Machines like us - a brief history of artificial intelligence
Machines Like Us - A brief history of artificial intelligence
From horse manure and monsters to inscrutable language models. The dream of artificial intelligence is as old as myth itself. But why are we so eager for artificial minds to replace our own?
By Sam Sundberg
“AI is a bigger threat than climate change”, “AI could cause ‘civilisation destruction’”, “Humanity is failing the new technology’s challenge.”
As OpenAI launched ChatGPT in 2022, not only did people envision amazing new ways to use the technology for the good of humanity, but many AI scientists expressed grave concern that the technology would be used to flood the internet with disinformation or worse, that machine intelligence was about to surpass human intelligence, presenting questions we are not yet ready to answer.
Many have speculated that low-level jobs will soon be taken over by AI. But no longer are only simple, repetitive occupations at risk. Lawyers, physicians, artists, writers… as artificial intelligence approaches the human level we all should worry about – or look forward to – machines replacing us in the workplace.
I recently spoke to Max Tegmark about these developments. He is the author of “Life 3.0,” a professor at MIT and a renowned AI expert, and he is profoundly worried. Tegmark has been campaigning against nuclear weapons for years, but at present, he considers artificial intelligence an even greater existential risk. If we choose to replace ourselves, and let machines do all our work for us, the human species may simply lose the desire to carry on and to procreate. But why, Tegmark wonders, would we want to replace ourselves with machines?
Ada Lovelace is often described as the first computer programmer and is said to have created the first algorithm created to be processed by a machine.
Ada Lovelace is often described as the first computer programmer and is said to have created the first algorithm created to be processed by a machine.
In fact, this question echoes through the ages: Why have scientists and alchemists for so long strived to create not just useful machines, but machines like us?
The pursuit of artificial intelligence is not about merely making efficient tools, like calculators and word processors. It is about mimicking human intelligence, a quest to equal or even surpass it. In essence, turning the myth of creation on its head, making humans the creators of new life through intelligent design. This dream has ancient roots.
An awesome bronze giant
The Greeks told of the divine smith, Hephaestus, who forged automatons to serve its masters. Talos is the most famous of his creations, an awesome bronze giant who patrolled the island of Crete, protecting it against pirates. At Alexandria, the Egyptian scholar Heron built a spectacular array of automata for the theatre. Not intelligent, naturally, but appearing alive.
Around the thirteenth century and onward, many learned men, scholars and occultists were rumoured to possess mystical contraptions known as “brazen heads,” mechanical heads covered in bronze, which could answer any questions put to them. This may have been a legend borne out of the ignorance and jealousy of their scholarly wisdom. No evidence of any scientist or magician creating such a device exists. But soon automatons of a less supernatural kind became all the rage among the European aristocracy.
These cleverly constructed machines were no more than mechanical divertissements made of cogwheels and springs, inspiring awe and wonder. Magic tricks, to entertain guests, rather than actual dark arts. But alchemists and occultists were still hard at work, exploring the possibilities of creating some form of intelligent beings.
But alchemists and occultists were still hard at work, exploring the possibilities of creating some form of intelligent beings.
Indeed, in the sixteenth century, the Swiss alchemist Paracelsus claimed to have created a living, breathing homunculus by burying human sperm in horse manure for 40 days, magnetizing it, and then feeding it human blood for 40 more weeks. This little humanoid was said to work as an assistant to its master. Paracelsus promised, in words that could very well refer to the creation of artificial general intelligence far in the future:
“We shall be like gods. We shall duplicate God’s greatest miracle – the creation of man.”
Throughout history mankind has strived to create useful machines. This is a mechanical pinwheel calculator.
Throughout history mankind has strived to create useful machines. This is a mechanical pinwheel calculator.
In 1818, a sensational horror novel was released that tickled the imagination of thousands of readers. “Frankenstein,” by Mary Shelley, is the tale of a modern scientist following in the footsteps of alchemists like Paracelsus, consumed by the idea of creating an artificial man through scientific means. The Italian biologist Luigi Galvani had recently discovered that electricity is the means by which the brain sends signals to the muscles, so Shelley had Viktor Frankenstein animating his creature by electric current from a burst of lightning. The result, of course, is Frankenstein’s monster – a huge man, terrifying to its creator and woefully unhappy, who goes on a murderous rampage. The tale may serve as a warning of humankind’s troubles in controlling their greatest inventions.
Starting point of modern computing
Many historians would cite Charles Babbage’s designs for the Analytical Engine as the starting point of modern computing. In the 1830s, Babbage, a British inventor, engineer and mathematician, came up with two designs for machines capable of performing mathematical calculations. The first, called the Difference Engine, was funded by the British government and Babbage himself, but the project was never completed.
The second, called the Analytical Engine, was even more ambitious, and funding was hard to come by. Along with his companion Lady Ada Lovelace, he came up with different schemes to fund the project. At one point they proposed a tic-tac-toe-playing machine to entice investors, then they considered building a chess machine. Before they could build it, however, they came up with an even better idea. They would build the perfect system for betting on horse races, to fund the completion of the Engine. The scheme was meticulously planned by some of the brightest minds in England and ended in spectacular failure. Soon Lady Lovelace was deep in debt and rescued not by any ingenious machines but by her kind mother.
The Analytical Engine, like its predecessor, was never completed. But Babbage’s designs, along with Lady Lovelace’s ruminations on how the Engine would in theory be able to not only calculate numbers but to have those numbers to represent anything – for instance sounds in a musical composition – was an important step in the creation of the universal computer.
Technology shifts sometimes happens fast. Devices that not too long ago seemed very modern, can quickly go out of fashion.
Technology shifts sometimes happens fast. Devices that not too long ago seemed very modern, can quickly go out of fashion.
It would be another century before such a computer was finally realised. The world’s first programmable computer was built in the late 1930s by the German engineer Konrad Zuse. He called the mechanical, motor-driven machine the Z1. Although it was the first computer to be finished, many other engineers were tinkering with computers around the world. At this time, the field of psychology was also starting to understand the human mind as a biological network, and piece by piece figure out its workings. Perhaps the brain was best understood as a machine? And if so, might not a machine such as the computer, in the future, be able to perform the same work as the brain?
With these questions in mind, scientists were again starting to entertain ideas about thinking machines, mimicking human thought and behaviour. Their ideas were collected under names such as “cybernetics,” “automata theory” and “complex information processing”. It was not until 1956 that the American scientist John McCarthy came up with a new name for the field, that proved to be sticky: “artificial intelligence.” That summer he joined 19 other prominent academics at Dartmouth College in New Hampshire for a workshop brainstorming about the exciting new field.
Creating a computer mind
The participants of the conference were engineers, psychologists, neuroscientists, mathematicians, physicists and cognitive scientists; an interdisciplinary brain trust well suited to taking on the challenges of creating a computer mind. Their mission statement – brimming with the naïveté that comes from not yet having tried and failed – outlines their lofty ambitions:
“Every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it. An attempt will be made to find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans and improve themselves.”
The participants were confident that they would make great strides in this direction during the two-month workshop. It would be a stretch to say that they achieved their goal, but during those long discussions at the Dartmouth Math Department, they at least firmly established the new field of AI.
Mechanical divertissements, made of cogwheels and spring, to entertain guests were popular among European aristocracy.
Mechanical divertissements, made of cogwheels and spring, to entertain guests were popular among European aristocracy.
Human-level machine intelligence would turn out to be far harder to achieve than the early pioneers imagined. During the following decades, AI hype would be followed by AI winter in a cyclical pattern. Several prominent AI researchers, among them Marvin Minsky, had predicted human-like intelligence by the 1980s. When those predictions were proven wrong, some were deflated, but the Japanese government was eager to have Japan take the lead. In 1981, Japan initiated the Fifth Generation Computing Project, pouring 850 billion dollars into AI research, with the stated goal of creating machines that could carry on conversations, translate languages, interpret pictures and reason like human beings.
Progress was made during this time, primarily with limited systems tuned to play chess or give expert advice in narrow fields of inquiry. But anything even vaguely resembling the dynamic and wide-ranging intelligence of humans remained out of grasp. After the Fifth Generation Project came to an end, without fulfilling its promise, the field again found itself at a low point late in the 1990s.
Luckily, an offshoot of AI research was about to gain new traction. In parallel to the mainstream research going on at prestigious universities, well-funded government programs and hyped-up companies, some scientists had taken an interest in artificial neural networks. The network architecture was thought to resemble the human brain, offering new ways of representing machine thoughts compared to the strictly algorithmic manipulation of symbols of conventional AI systems. A neural network could be trained on appropriate data sets, much like a child learns, until its maze of internal connections becomes suitable for its designated tasks.
A fatal flaw
Artificial neural networks had a fatal flaw, however. As soon as you started to scale a network to do something interesting, its complexity increased exponentially, and the system ground to a halt. The computer hardware of the time, with architecture very different from human brains and far less processing power, simply could not keep up. So, this line of research remained theoretical, dormant for decades until, deep in the 2010s, the time had come for the AI field to enter a new era of machine learning.
Three developments of the new millennium came together to finally make neural networks practical:
Computer hardware kept getting faster, smaller and more energy efficient, as predicted by Moore’s Law.
Computer scientists developed more sophisticated architectures and algorithms for artificial neural networks.
An immense trove of digital text, images and sounds accumulated online, an all-you-can-eat buffet of information for neural networks to be trained on.
Looking back at what was then envisioned, artificial intelligence is finally living up to its name.
With the recent work of DeepMind, OpenAI, Google and Microsoft, we arrive at today’s state of the art. Artificial intelligence may have missed the deadline of Japan’s Fifth Generation Project, but looking back at what was then envisioned – or indeed, what the Dartmouth Workshop sought to achieve – artificial intelligence is finally living up to its name. ChatGPT and its rivals can easily hold conversations with humans; Google Translate and its ilk can translate text and speech in the blink of an eye; and many neural networks not only interpret images but also create beautiful pictures from natural-language prompts.
Several fundamental questions do remain, however. Can these machines truly reason? Can they think? Can they feel? Will they ever?
The French seventeenth-century philosopher René Descartes famously formulated a dualist theory where mind and body are metaphysically separate. He was inspired by the automatons on display in Paris at the time and concluded that mind and body must be different substances. The latter can be replicated by automatons, while the former is singular to man and intimately tied to what makes us us. We think, therefore we are.
Unexpected leaps
With AI science advancing – at times inching forward incrementally, sometimes striding with unexpected leaps – software engineers are getting closer to imitating the human mind as well. Chat GPT has repeatedly defeated the Turing test, designed by the British computer pioneer Alan Turing to settle the question: “Can machines think?”
Refined algorithms, humongous data sets and clever reinforcement learning techniques are pounding at the walls of dualism. Perhaps, as the Dartmouth Workshop proposed, the human mind is a mere machine after all. And if this is the case, why would we not be able to replace it with more efficient machines?
The history of artificial intelligence is a tale of scientific progress, of engineering failures and of triumphs. But it is also the story of our struggle to understand our own minds. Are we truly unique? Are our brains, like our bodies, simply machines governed by electrical impulses? When we dismiss the “thinking” of large language models as simply a series of predictions of what comes next, are we absolutely certain that this does not also apply to human minds?
It seems inevitable that we will soon be able to create genuine thinking machines – if we haven’t already.
At this point (as at every point in the history of AI) it seems inevitable that we will soon be able to create genuine thinking machines – if we haven’t already. There is still some disagreement about whether we can create feeling machines, however. Conscious machines. Machines that can do and experience everything that a human can and more.
Some aspects of this may be harder than we can foresee. On the other hand, it may be within our power sooner than we think, emerging incidentally as our models become increasingly complex, combining techniques from neural networks with symbolical AI.
Mary Shelley would be delighted to see modern scientists still hard at work trying to realise the ancient dream of godlike creation. The full original title of her famous horror novel is “Frankenstein; or, The Modern Prometheus.” The subtitle alludes to the crafty Titan from Greek mythology who stole fire from the Olympian gods and gave it to man. Fire, in this old tale, symbolises knowledge, invention and technology – powers Zeus had determined must be reserved for gods and withheld from humankind.
In some versions of the myth, Prometheus gives us more than fire; moulding the human race from clay, he also gives us life. Millennia later, the fire he gave us is still burning bright, and we are now the ones doing the moulding. Not from clay, but from code.
Sam Sundberg
Freelance writer, Svenska Dagbladet
My favourite song the last decade: Light years – The National
A beginner’s guide to generative AI
A beginner’s guide to generative AI
We’ve all heard of AI, machine learning and ChatGPT. But how does it really work? Here’s a beginner’s guide to the technology behind – and what might come next.
By Sven Størmer Thaulow
A beginner’s guide to generative AI
We’ve all heard of AI, machine learning and ChatGPT. But how does it really work? Here’s a beginner’s guide to the technology behind – and what might come next.
By Sven Størmer Thaulow
In today’s rapidly evolving digital landscape, buzzwords like “AI” or “machine learning” are becoming increasingly common. Even if you’re not entirely sure what they mean, chances are you’ve encountered them in some form or another, perhaps through smartphone assistants like Siri or Alexa or in online customer service chats. However, a subset of AI, known as generative AI, is emerging as a transformative force in the digital world. Here’s a closer look at this technology and its implications for the future.
Unravelling the Mystery of Generative AI
At its core, generative AI is about creation. Much like an artist creates a painting or a writer crafts a story, generative AI can produce new content. But instead of paint or words, its tools are data and algorithms.
Imagine having a conversation with a friend about your favourite books. As you talk, your friend might suggest a new book for you to read based on what you’ve mentioned. Generative AI operates on a similar principle. Feed it with enough conversations about books, and it could suggest or even create a synopsis of a book that doesn’t exist but fits within the parameters of the conversations it’s analysed.
The Magic Behind the Screen
The magic of generative AI lies in its ability to produce content, be it text or images. But how exactly does it do this?
For text, generative AI models are trained on vast databases of written content. They analyse patterns, contexts, and structures within these texts. When given a prompt or starting point, they use this training to predict and generate what comes next. It’s like teaching a child to speak by immersing them in conversations until they start forming their own sentences.
On the image front, things get a bit more complex. Techniques like Generative adversarial networks (GANs) are often employed. Here’s a simplified explanation: imagine two AI systems – one is the artist (generator) and the other is the critic (discriminator). The artist creates a picture, and the critic evaluates it. If the critic can easily tell it’s a generated image and not a real one, the artist tries again. This back-and-forth continues until the artist produces something the critic can’t distinguish from a real image. Through this process, the AI becomes adept at creating realistic images.
Societal Impact and the Media Realm
The proliferation of generative AI doesn’t merely affect technological circles; its ripples will be felt across society. As AI-generated content becomes commonplace, our ability to discern between human-created and AI-created material might blur. This poses profound questions about authenticity, trust and the value of human creativity. For media companies like Schibsted, the implications are vast. On one hand, AI can generate news reports, write articles or even create visual content at a pace unmatched by humans, offering efficiency and cost savings. However, this also brings challenges. How do media houses ensure the credibility of AI-generated content? And as audiences become aware of AI’s role in content creation, how will this shape their trust and engagement with media outlets?
Charting the Evolution of Generative AI
Like all technologies, generative AI wasn’t born overnight. It’s been a product of years of research, improvements and refinements. As computational power increases and algorithms become more sophisticated, the capabilities of generative AI expand. Currently, we’re witnessing AI that can draft articles, compose music and generate artwork. Yet, this is just the beginning. The trajectory suggests
a future in which generative AI can create more complex, interactive, and nuanced content. Think of virtual realities indistinguishable from our own, or digital assistants that not only understand our preferences but can also predict our needs before we articulate them.
The Next Wave of Breakthroughs
Predicting the future is always a gamble, but based on the current momentum, several exciting developments appear on the horizon for generative AI.
- Personalised content: In a world saturated with content, personalisation is becoming paramount. Generative AI could craft experiences tailor-made for individuals. Imagine a movie that adjusts its storyline based on your preferences or a video game that evolves based on your playing style.
- Education revolution: Customised learning isn’t new, but with generative AI, it could reach unprecedented levels. Students might have access to study materials created on the fly, precisely addressing their weak points and reinforcing their strengths.
- Artistic collaboration: While some fear AI might replace human artists, a more optimistic view is a future where artists and AI collaborate. An AI could suggest melodies for a musician or sketch ideas for a painter, enriching the creative process.
In conclusion, generative AI lies at the intersection of art and science, holding the promise of a world where technology enhances creativity, personalisation, and efficiency. At Schibsted we feel we are on the cusp of this new era, and that it’s crucial to approach it with both excitement and caution. We must ensure that as we harness its potential, we also consider the ethical implications of AI shaping our reality.
Using an AI co-pilot: How did I make this article?
This article was a classic task for generative AI as it was a fairly generic piece more describing a well-known domain rather than being a very personal and opinionated article – so I used ChatGPT as a co-writer. I tried out a prompt describing the article I wanted. It became very “chatGPT-ish” – lots of numbered bullets with sentences. So I tried again with a prompt saying I wanted it in “New York Times” style. I got closer. I tried some more prompt variations and also limited it to the number of words. When I had the 80% text I wanted I started rewriting somewhat, cleaning up mistakes and adding some elements. And voila – a pretty decent article was born!
Sven Størmer Thaulow
EVP Chief Data and Technology Officer, Schibsted
Years in Schibsted: 4
My favourite song the last decade: Thinking of a Place – The War On Drugs
Human happiness must be our common goal
“Human happiness must be our common goal”
She thinks we’re discussing AI on the wrong level. And her vision is that everyone should understand how the technology works. Inga Strümke has become a tech celebrity in Norway, much thanks to her bestselling book, “Maskiner som tenker.”
By Ann Axelsson
“Human happiness must be our common goal”
She thinks we’re discussing AI on the wrong level. And her vision is that everyone should understand how the technology works. Inga Strümke has become a tech celebrity in Norway, much thanks to her bestselling book, “Maskiner som tenker.”
By Ann Axelsson
“If you talk about existential risks and appeal to people’s fears, you will get attention,” she says, referring to the dystopian warnings that AI will replace humans and take all jobs.
“These futuristic scenarios are not constructive, and they make it hard to debate the mechanisms behind the technology. What we really need to discuss is how we can develop today’s AI systems according to legislation, our goals, and our values.”
Inga Strümke is an associate professor in AI at the Norwegian University of Science and Technology, NTNU. She’s also a particle physicist, a frequent public speaker, and a best-selling author. “Maskiner som tenker” (“Machines that think”) has become a bestseller in all categories in Norway.
She also spent several years reaching out to Norwegian politicians, trying to get them to take AI issues seriously. That was a challenge. Until ChatGPT.
“Unfortunately, it takes bad news to get them to listen.”
As a scientist in the field, she of course welcomes progress, and she explains that the scientist’s mindset is to think about what is possible and then develop that.
“That mindset has given us X-ray, GPS, the theory of relativity. And atom bombs. As a scientist, you never know how your findings will be used. That’s where society needs to step in.”
Inga Strümke has spent several years reaching out to Norwegian politicians, trying to get them to take AI issues seriously. She was also a speaker at Shibsted’s strategy summit in October 2023.
Inga Strümke has spent several years reaching out to Norwegian politicians, trying to get them to take AI issues seriously. She was also a speaker at Shibsted’s strategy summit in October 2023.
And she believes that ChatGPT is a perfect example of how bad things can go when you share “fun” new technology openly, without having had discussions about its implications.
“I believe we have a serious problem when pupils are now thinking, ‘Why should I write a text if there is technology that can do it better?’ How will we now make young people understand that an idea starts with a thought in your head and that you need to grow and communicate that idea to get other people to understand it? And if you can’t do that, then you won’t be able to function in society.”
That might sound just as dystopian as the future scenarios. But her point is that we can and we must take the lead here in the Nordics and in Europe when it comes to discussing the role we want technology to play.
“If we can’t manage to figure out how to use what we develop here, then we will end up using solutions developed by tech giants that we are unable to influence.”
Impact across the society
And these discussions, she says, should involve experts from across the board – politicians, social scientists, economists, legal experts, ethics, apart from technologists – since the impact will be felt across all parts of society.
This is also why she thinks it’s so important that as many of us as possible understand what this is about and how the technology works. The things she explains in her book.
“My dream is that anyone can read it. That a woman past 80 would read it and think, ‘I can understand this if I want to.’ I have this passion to empower people on this subject. To make them see that they can look after their own interests.”
New class issues
What also has become clear to her, in discussions after writing the book, is that AI can spur a new kind of class issue. That the world might be divided between those who are able to use the technology for their own benefit and those who aren’t.
“Someone said that AI will not take the journalists’ jobs. But journalists who know how to work with AI will take the jobs from the journalists who don’t.”
Inga got into the science world as a physicist. She wanted to understand the universe. Then, when she took her bachelor’s at NTNU, she noticed there was a field of study on artificial intelligence, and curiosity led her that way.
“My driving force is to find out what is possible. The main reason that I’m still in this field is that I see the consequences, and they are extremely complex.”
Even though she thinks about this complexity day and night, she also finds time to use that curiosity and energy on other things – mainly outdoor activities. Her social media is filled with pictures of her mountain biking, climbing and hiking. And those things are important.
“No matter what happens with technology and politics, there is one important thing that we can’t forget: to have a nice life. Human happiness must be our common goal – if not, nothing else matters. That’s very important to me to remember, every day.”
Ann Axelsson
Senior Product Manager, Strategic Communication, Schibsted
Years in Schibsted: 25
My favourite song the last decade: Paper Doll – John Mayer
AI for good or for bad?
AI for good or for bad?
How will AI impact our lives and society? Joacim Lund compares it to the breakthrough that came with the internet – and sees many similarities. AI will solve many problems in our daily lives. We just need to make sure it doesn’t create new ones.
By Joacim Lund
AI for good or for bad?
How will AI impact our lives and society? Joacim Lund compares it to the breakthrough that came with the internet – and sees many similarities. AI will solve many problems in our daily lives. We just need to make sure it doesn’t create new ones.
By Joacim Lund
Artificial intelligence is a flop. Ok, not really. But we are talking about it the wrong way.
In Norway, an opinion piece from 1996 pops up on a regular basis. The headline of the newspaper clipping is crystal clear: The Internet is a flop.
Every time it appears, people have a good laugh. But the person who wrote it (incidentally, a reader I keep getting meaningful emails from) wasn’t irrational in his argument. He believed that people who work on computers will get more than enough of it during office hours (this definitely applies to some of us), that humans are social beings and, moreover, that it would not be profitable for business to offer their services online.
“When we rent our videos, we will visit the rental store and make our selection in visual surroundings,” he opined.
27 years later, much of the debate around artificial intelligence is not entirely dissimilar. People are either for or against artificial intelligence. They think it’s either good or bad. They think it will eradicate us or not. They want to stop it or let it be.
As new technologies arrive – the sceptics speak up. New habits sometimes take time to accept.
As new technologies arrive – the sceptics speak up. New habits sometimes take time to accept.
At the same time, AI developers from around the globe are creating autonomous weapons, racist algorithms and extreme surveillance systems that challenge human rights. Other AI developers are creating systems that revolutionise medical research, streamline the public sector, or help keep the temperature of the planet below boiling point. And everything in between.
The possibilities seem to be endless. So, shouldn’t we rather be talking about how AI can be used responsibly?
It’s changed everything
Today, everyone knows that the internet was not a flop. The authorities communicate with us using the internet. Ukraine and Russia are bombing each other using it. The propaganda war is almost exclusively waged online.
But perhaps even more importantly: the internet solves problems and has made life easier for most people. I charge my car online, pay bills and rent my videos (but sometimes I do long to go back to discussing movies with the movie nerd who worked on Bislet Video instead of getting flimsy recommendations by an algorithm).
I listen to my music online, remotely activate the heaters at the cabin, where I can also stream my news broadcasts. People find life partners online and discover like-minded people who also get excited by photography, old Italian scooters, 16th-century clocks, or Talk Talk bootlegs. We have access to everything, everywhere, all the time.
That’s why everyone laughs at the “flop” prediction. He was absolutely certain and he was wrong. But that’s easy to see in hindsight. And it’s hard to predict.
In 1996, people were concerned about Charles and Diana’s divorce and Bill Clinton’s re-election. Who could have imagined that Diana would die in a car accident in Paris a year later? Or that two years later, Bill Clinton would be explaining why his sperm was on a dress?
Or that the internet was going to change everything?
Tomorrow is only a day away
I have no idea how artificial intelligence will have impacted society, people and the lives we all live in 2050. But I see several similarities between the internet in 1996 and artificial intelligence today:
Artificial intelligence solves problems and will make life easier for most people. Artificial intelligence is changing assumptions.
Also for people who don’t have good intentions.
“Whoever takes the lead in this field takes the lead of the world,” Putin said during a live address to 16,000 schools at the start of the school year in the fall of 2017. By “this field” he meant artificial intelligence. Xi Jinping had recently launched an ambitious plan to make China a world leader in artificial intelligence by 2030.
It almost makes you want to just ban the whole thing. Impose international sanctions and throw the baby out with the bathwater. The problem is that artificial intelligence opens up so many positive possibilities as well.
Things like evaluating X-rays, AI can solve much faster and better than humans.
Things like evaluating X-rays, AI can solve much faster and better than humans.
The toe is broken
I was once playing with my son. I chased him around the apartment, arms straight ahead, like a zombie. As I made my way around the kitchen table, my little toe hooked onto a chair leg. There was no doubt that the toe was broken. It pointed straight out to the side. Still, I spent most of the day in the emergency room.
The reason for this was a bottleneck in the system.
When people come in with minor fractures or just a suspicion that they’ve broken something, for that matter, an X-ray is taken. A doctor (the bottleneck) must study each individual image to see if there is a fracture or not. If there is no fracture, the doctor sends the patient home. If there is a fracture, the patient is placed somewhere on a priority list.
However, minor fractures are not life-threatening. If there is a lot to do in the emergency room, the X-rays will be given low priority until there is more time to look at them. When the doctor finally has time, he or she will study the picture and determine that there is no fracture – in about 70% of cases. The patient, who by then may have waited seven hours, is then told to go home, take two ibuprofen and three glasses of red wine (which my dentist once recommended), and turn on Netflix.
It’s things like this that artificial intelligence can solve much faster and better. And it’s already doing it, actually.
Level up
The other day I was visiting a hospital in Bærum, just outside Oslo. An enthusiastic, young, bearded radiologist pointed to an X-ray image on a screen in front of us. The picture showed a foot, and it looked quite similar to the picture taken once upon a time of my foot (except that the little toe didn’t point straight out to the side).
But one thing was different. The image had been assessed by an artificial intelligence.
Above the ankle bone, a yellow square had been drawn, lined with the text “FRACT.” That means there’s a fracture there. The software goes through all the X-rays as they come in. Seven out of ten patients are told immediately that they can go home. The rest automatically end up in a priority queue.
Doctors do not have to spend valuable time finding out that everything is okay, and patients do not have to wait. This is an extreme efficiency improvement in a health service that will experience greater and greater strain in the decades to come.
Should this have been banned? Some think so.
Sense and sensibility
A few months earlier, two Norwegian politicians warned that artificial intelligence leads to everything from polarisation to eating disorders, and perhaps even the extinction of humanity. The government should immediately “impose a public sector moratorium on adopting new commercial tools based on artificial intelligence,” they argued.
This is an absurd approach to artificial intelligence. The pressure on the healthcare system only increases as people age. To have any hope of maintaining good health services for the population, we must make use of the tools at our disposal. The AI tool at Bærum Hospital happens to be delivered already fully trained from abroad. All patient data is deleted, so as to avoid all privacy issues. Of course, there shouldn’t be a ban on such things. But the two politicians still had a good point:
“The development of AI has for a long time taken place without adequate regulation from the authorities.”
Now it’s happening
There has been a Wild West approach from the tech companies. Naturally. Development is rapid, and work on laws and regulations is slow. But the EU has been working diligently on the issue.
The EU’s first draft regulation of artificial intelligence, the so-called AI Act, was presented two years ago. It is likely to be formally approved within 2023. The EU is adopting a risk-based approach. For services that pose a low risk, it’s full speed ahead. Unacceptable risk means it’s prohibited. And for everything else in between, there are two more levels: acceptable risk and high risk.
The purpose of the AI Act is to ensure that artificial intelligence systems are safe and respect fundamental rights and values. This means, for example, that facial recognition in public spaces is banned. It’s not allowed to single out citizens for the authorities to keep an eye on in case they do something illegal. Stuff like that.
AI should be open and honest, not closed and manipulative. The resistance the AI Act has faced from tech companies suggests that regulation is needed. For example, Sam Altman, the man behind OpenAI and ChatGPT, has threatened to pull out of Europe if the regulations become too extensive.
Perhaps now it’s time to revisit the crystal ball.
A willingness to solve problems
In September 2023, Norway’s Prime Minister, Jonas Gahr Støre, held a press conference where he proudly announced that his government would allocate one billion Norwegian kroner to artificial intelligence research, to be used over the course of five years. On the same day, the government leaked that it would spend five billion on a new tunnel to shave a few minutes off the drive between two villages in the mountains of western Norway somewhere. But OK, a billion is money too.
A large and important part of the research will focus on how artificial intelligence can be used for innovation in industry and in the public sector. Like in hospitals, when people come in with sore fingers and toes. Or in building applications, so people don’t have to wait several months for the overworked caseworker to get far enough down the pile. Or to provide public services with a faster, larger and better basis for decision-making. Or to improve data security, in fact, and not worsen it.
And in so many other ways that I can’t possibly imagine.
That’s what politics is all about. To follow social developments and govern society in a way that makes it as good as possible for as many people as possible. Norway is just an obvious example because that’s where I live. The same goes for every other country and continent, and globally, for that matter.
As in other areas of society, international resolutions and treaties and sanctions must be adopted to ensure that artificial intelligence is used in a way that solves humanity’s problems, rather than create new ones.
That work is underway.
OK, here’s what the crystal ball says
If I’m going to allow myself to try to look 27 years into the future, to 2050, I’d guess that people are more concerned about themselves and their nearest and dearest, and not so much about what people were thinking back in 2023. But those who bother to read old newspapers might chuckle a bit at the banal discussions we had about artificial intelligence ’way back when.’ And the fact that many were either for or against. Maybe it’ll be the demand for a ban and the call to halt development that everyone will laugh at (try asking Putin to stop the development of artificial intelligence, by the way).
I’m guessing that my future grandchildren will experience an education system much more attuned to each student’s learning disabilities, learning styles and skills. That their health will be taken care of much better than by the GP they see every two years. That potential health problems will be discovered before they become major and serious. I’m guessing the car will be a safer driver than the human. That public transport will be much better adapted to people’s needs. That precise weather forecasts will control the heating in houses. That everyone will be better protected from abnormal activity, whether it’s in their bank accounts or in their apartments. Maybe I won’t have to think about shopping for food or cleaning the house anymore.
I’m guessing it will seem strange that society spent so much time and resources on having people perform repetitive and simple tasks. And that major and important decisions were made on a razor-thin knowledge base.
I am absolutely certain that artificial intelligence will be subject to international regulations. And that artificial intelligence will lead to global, regional, local and personal changes that are difficult to imagine today.
Because by then humanity will know better.
If, of course, it still exists.
Joacim Lund
Technology commentator, Aftenposten
Years in Schibsted: 18
My favourite song the last decade: Bråtebrann – Kverletak
On speaking terms with machines
On speaking terms with machines
We have interacted with our computers in mostly the same way for almost 60 years. But now we’re entering the age of conversational interfaces. Schibsted’s Futures Lab has experimented to understand more of their capabilities and constraints. The experience was surreal.
By Christopher Pearsell-Ross
On speaking terms with machines
We have interacted with our computers in mostly the same way for almost 60 years. But now we’re entering the age of conversational interfaces. Schibsted’s Futures Lab has experimented to understand more of their capabilities and constraints. The experience was surreal.
By Christopher Pearsell-Ross
With the invention of the mouse in the 1960s, command-line interfaces gave way to a visual paradigm defined by graphical user interfaces (GUIs). Icons, menus and windows made computing more accessible to more people, and more applicable to a broader range of tasks.
In the mobile age, we have left our cursors behind in favour of the touchscreen. Now more than ever, we are reliant on visual metaphors to interact with our machines. We discover, create and explore our digital worlds with clicks and scrolls, taps and swipes, but this reliance on two-dimensional GUIs does not fully reflect our shared vision of how future technology should look.
These visions, exemplified by scenes in science fiction film and television, help define our shared expectations for what our technology should be capable of. In the future we are often shown, machines will speak and understand us. They will know us, anticipate our needs, and for better or worse, have the agency to act on our behalf. Large language models and tools like ChatGPT appear to be changing the paradigm again, bringing these sci-fi visions closer to reality.
Developed in 1964
These conversational interfaces are by no means new. Eliza, the first convincing chatbot, was developed at MIT in 1964 using simple pattern matching and rule-based responses. Siri was launched in 2011 as part of iOS using machine learning to recognise speech and to make sense of our intentions, letting many of us speak to our computers with our own voices for the first time.
But these interfaces have been limited to the responses and actions their programmers pre-defined. AI might have changed the input side of the equation, but these tools are still a lot closer to Eliza than we might care to admit. Advancements in AI technology over the last few years are radically altering this equation.
The introduction of generative AI, built on advanced neural networks called transformers, is reshaping the way our computers understand, process, and even create text. These AI models are what revolutionary new products like ChatGPT are built on, but they are also driving incredible improvements beyond text generation, including new capabilities in speech recognition, voice synthesis, sentiment analysis, image and video generation, and even the creation of 3D assets and animations.
In the year since ChatGPT was released, several key tech trends are shaping the future of conversational interfaces. Context windows are growing, essentially giving these tools longer memories and leading to more nuanced and relevant conversations. These tools are also getting connected to external data sources and digital services, enabling them to provide reliable and referenced answers, perform calculations and data analysis, and even take actions on behalf of the user. Lastly, as a recent release from ChatGPT shows, these tools are becoming multi-modal, meaning they are capable of processing not only text but also audio and images as both inputs and outputs, further expanding their versatility.
Until now, conversational interfaces have been limited to pre-defined responses and actions. AI is radically altering this equation.
Until now, conversational interfaces have been limited to pre-defined responses and actions. AI is radically altering this equation.
Aside from technology, social trends are also shaping this conversational paradigm. Firstly, populations in the developed world are ageing as birth rates decline, life expectancies increase and immigration and healthcare systems struggle to keep up. At the same time, feelings of loneliness and isolation are growing. In 2022, the number of single-person households in Sweden grew to over two million, and in 2023, the US Surgeon General warned of the public health effects of a growing epidemic of loneliness. Finally, in many parts of the world, education gaps are also growing. Inequities like racial, gender and economic disparities mean more people around the world are left out and left behind when it comes to the opportunities that education affords.
Taken together, we are seeing signs that point toward a future in which we increasingly rely on our technology for tasks and companionship that have traditionally been performed by people. There are opportunities and risks here. Conversational tools might enable new forms of healthcare and companionship services, give knowledge workers new superpowers or provide personalised tutors to children who need them. And they might also replace human connection, displace workers, or widen inequities.
Conversational user interfaces can bridge the best of what computers and humans have to offer.
While looking at hypothetical scenarios and possible outcomes is an important part of how we inform our strategy, our mission at Futures Lab doesn’t end there. To learn more about what we can and should do today, we need to get our hands dirty with practical experimentation.
Speculative prototyping is like a kind of time travel – it allows us to rehearse possible futures, and to experience what it might feel like to be there ourselves. In this case, we built a phone powered by ChatGPT to learn about how we might talk with AI-enabled devices in the future.
Inspired by science fiction examples like Samantha from the film “Her,” we set out to build an audio-only interface. Our goal was to explore the technical maturity, usability and applicability of CUIs in today’s landscape.
We scoured Finn.no for a suitable device to house our new tool and settled on a 1970s-era Ericofon 700. To provide a context for our experiment, we decided to explore workplace productivity and set out to design a weekly de-briefing tool to help us reflect on our work and keep our stakeholders updated.
We were able to use the original speaker but replaced the dialling mechanism and speaker with a Raspberry Pi minicomputer, new microphone and a proximity sensor so we could tell when the phone was lifted. Using OpenAI’s Whisper service for voice recognition, we sent a transcript of what users said to ChatGPT using a custom system prompt. This prompt helps GPT know how to respond, what role to play and which tone of voice to use. Finally, the system’s response is played back to the user using Google Cloud text-to-speech functionality.
The result was compelling and eerily similar to some examples from science fiction. While you still need to take turns speaking and listening, the conversation flows fairly naturally. Our AI agent can ask highly relevant follow-up questions, keep the conversation on-task and help users reflect on their work in new ways. Once the system determines it has enough information (usually after a few minutes of back-and-forth conversation) it writes a summary for the user, which it can either re-write or submit to a database at the user’s instruction. From there the summaries can be used in any number of ways, from providing a searchable archive of our progress to creating tailored newsletters and Slack updates.
The audio-only experience allows us to assess what actually speaking with our machines in open-ended, flowing conversations might be like, without relying on the graphical and visual indicators we normally use.
Using these new interfaces has been as informative as it has been surreal. The scenes from “Her” and “Star Trek” that we took as inspiration are very quickly becoming reality. Testing prototypes like this can help us understand the capabilities and limitations of the technology, how to design usable products, and where and when CUIs are an appropriate choice.
Impressed with the quality
The people who have tested our phone interface were impressed by the overall quality of the conversations and the relevance of the follow-up questions. Being able to go off-script and have an actual voice conversation with a computer has been revelatory, though not without its frustrations.
Audio-only experiences are clearly outliers, but prototyping in this extreme format and comparing the experience to conventional chatbots has highlighted some important usability considerations. The things we may take for granted when using well-designed GUIs – namely, seeing the system status, understandable actions with clear icons and buttons, and information hierarchies that prevent cognitive overload – become more complicated when we only have our ears to rely on.
When it comes to usability and user experience, user preferences are strongly divided between the audio and text-based interfaces. Some users felt the intimacy, distraction-free focus, and ability to speak plainly without pretension or self-editing created a novel experience, one in which they were prompted to reflect and share a sense of openness and safety. Other users expressed a strong preference for text-based communication. They cited the efficiency of typing, the ability to refer to questions and previous answers, having time to formulate and edit their responses, as well as the ability to read and paste in other materials as important factors for them.
An important consideration in both text and audio-based CUIs is expectation management. These tools have come a long way and are able to converse at such a high level that many users will expect them to have capabilities and understandings far beyond their current capabilities. We can blame this partly on the quality of synthesised voices available today – the more human the system sounds, the more human we expect it to behave.
ChatGPT and other conversational tools like it are already demonstrating two key superpowers. First, they are great conversationalists and interviewers – they are able to understand our meaning clearly, provide tailored answers, and ask highly relevant questions. They are also able to translate human language into data consumable by machines, and to take complex data and translate it back into comprehensible human language.
We see these tools being most useful in contexts in which both of these abilities can be leveraged. Obvious applications include games and interactive media, personalised content production in news media, customer service, sales and product discovery. They are already proving highly useful as task-specific assistants in programming, research and data analysis, and we expect them to be applied as pervasive personal assistants and tutors in the very near future. Less obvious, and perhaps more far-fetched and ethically challenging, applications include AI therapists, healthcare advisors and personal companions for the vulnerable.
Combination of superpowers
Conversational user interfaces can bridge the best of what computers and humans have to offer. They can leverage the high-speed data analysis and computational superpowers of computers while making sense of the messy, creative and intuitive understanding we have as humans. In the best-case scenario, this combination of superpowers will help make the world more accessible to people with visual and cognitive differences, help make education more accessible and tailored to individual needs, increase our productivity at work and free up more of our time for the things that truly matter. On the other hand, these tools also have significant potential to disrupt labour with new forms of automation and to create emotionally impactful, biased content that drives further social isolation, misinformation, and inequity. The reality is that both scenarios are likely to coexist.
This is a rapidly changing landscape and things we thought of as science fiction are already becoming reality. We can’t predict the future, but through foresight and experimentation, we can better prepare ourselves for the changes, challenges and opportunities that are coming. That’s the approach we try to take at Schibsted’s Futures Lab. We are seeing a new paradigm of interaction on the verge of being born. CUIs have incredible potential to empower people in their daily lives… if we can get it right.
This text was human-generated by the Futures Lab team. ChatGPT was used as a sparring partner and writing critic throughout the process. Special thanks to our summer intern Lucía Montesinos for driving much of this work.
Christopher Pearsell-Ross
UX designer, Schibsted Futures Lab
Years in Schibsted: 2.5
My favourite song the last decade: Your Best American Girl – Mitski
Applying AI in Schibsted
Applying AI in Schibsted
At Schibsted, we experiment and work with AI in all our business areas. These are some examples of applications that have improved the way we work and our products.
Applying AI in Schibsted
At Schibsted, we experiment and work with AI in all our business areas. These are some examples of applications that have improved the way we work and our products.
Video-subtitles
Over the last few years, the demand for subtitled videos has increased due to accessibility needs and because more and more users watch videos without sound. But the manual timing and writing of subtitles take a lot of time. With the help of the OpenAI-model Whisper, video editors can now upload videos to a service built by the Aftonbladet TV Operations team and it will automatically generate subtitles. These subtitles can then be used in Adobe Premiere.
“The subtitles still need to be edited in Premiere a bit to fix minor errors, but the time gained is enormous, which creates more time for other tasks in the newsroom,” says Vasilios Hatciliamis, Head of TV Operations at Aftonbladet.
Language model
In a small cabin during summer vacation 2023, a Schibsted LLM was born. Simen Eide and Anders Haarr from AI foundations in Schibsted started training a model with Schibsted content to create SEO optimised headlines for Schibsted newspapers. It turned out to be five times as good at the job than ChatGPT and other open-source models. So far it’s been tried out on VG but the internal interest to use it is big. The ambition is also to implement it on other products in Schibsted, like marketplace brands, but exactly where this will go is not decided.
“It’s really cool that we are able to work on a project like this, without a final goal,” says Simen. ”I think that’s thanks to the culture of innovation that we have in Schibsted.”
AI helps engineers in Schibsted to write code.
AI helps engineers in Schibsted to write code.
AI helps with writing code
Hundreds of software engineers in Schibsted now use artificial intelligence to help them write code. “My productivity has grown at least 15%,” estimates one of them, Pedro Goncalves.
Good software engineering requires lots of creativity as well as superb skills in solving problems. But the daily work is also full of tedious and repetitive tasks. There are tests and failures as new code is produced – and it all takes time and energy.
That’s why, after a pilot project, Schibsted decided to let all its software engineers use the AI tool GitHub Copilot in their daily work. GitHub Copilot is like a ChatGPT for programmers, it just suggests software code instead of normal text.
After only a few weeks, 34,000 lines of code had been accepted by the software engineers. (By comparison: some estimate that an average programmer writes about 25,000 lines of code in a year.) And nearly 400 programmers have already started using it.
Indexing offers at Prisjakt
At Prisjakt, a price comparison service within the Schibsted family, machine learning (ML) plays an integral role in their ever-so-important product matching system.
The system utilises ML algorithms, such as neural networks, natural language processing (NLP) and computer vision methods, that support the categorisation of items from shops and matches the items to Prisjakt’s database of products, called “product matching.” This is a redundant system that evaluates several inputs, such as text and images from the product descriptions, along with the price.
There are almost 300 million items to process in the system and millions are matched every day, effectively reducing the need for manual labelling for just a small group of people.
Manual labelling combined with the creation of new products helps to continuously improve the performance of the ML models, which results in more product matches. For the users on Prisjakt’s site, this results in a seamless experience as they search for and compare products, ultimately enhancing the shopping experience and driving more traffic to the shops.
Overall, Prisjakt’s well integrated ML solutions for automated product matching serve as a competitive advantage, as well as being the backend for the price comparison service.
A feature built with GPT-4, creates short summaries of news articles.
A feature built with GPT-4, creates short summaries of news articles.
AI-generated summaries
On VG, Aftonbladet and Aftenposten, readers can get short summaries of news articles. This is a feature built with GPT-4, created by a cross-brand team in Schibsted. When an article is ready, the journalist simply toggles on the functionality in the content management system, and a summary is generated. The journalists can then review it and hit ‘publish’ when it’s ready.
To make sure that mistakes do not slip through the cracks – there is an extra safety mechanism. The team behind it has asked GPT-4 to double-check that the text and the summary are aligned before it’s published, using text classification. And the readers like it. The overall click-through rate on VG’s summaries is 19%, and for young readers, it’s 27%. You might think that there is a risk that these readers don’t read the whole article – but it turns out that they often do. They simply use the summary as an introduction.
Transforming sound to text
When software engineer Johannes Andersen had lunch with a VG colleague who was moaning about having to transcribe an interview, he was sympathetic but didn’t think more about it until he saw a forum post about the OpenAI model Whisper. Then he tried the model out on a hack day and it turned out to work very well.
As it became clear how much time the tool could save, Johannes Andersen and his team invested three weeks to work on an application. They’ve built an interface and an app where journalists can upload their sound files and turn them into texts. At the time of writing, a few months after launch – the web application has saved 13,433 hours for journalists who can now use that time to do other things.
You can use it for several languages and the word error rate is 9.5% for Norwegian and only 4.2% for English. With the app, the journalists can work locally on their computers, so they don’t have to share any data.
A way to listen to all articles
In the last Schibsted Future Report, we told the story about Aftenposten’s synthetic voice. In cooperation with the company BeyondWords, a team from Schibsted’s subscription newspapers trained an artificial voice, using sentences from published articles that were recorded by podcast host Anne Lindholm. Since then, the project has grown. Now you can listen to most articles published on Aftenposten.no. And soon there will also be playlists with recommendations for articles to listen to.
At Aftenposten Junior, where the project started, you can choose between nine different languages, including Arabic, Ukrainian and Somali. And more newspapers in Schibsted are on their way to implementing the technology. But this project is not only about convenience. It’s also a question of giving everyone access to the same information.
“For instance, we learned from teachers that 92% of them have students who struggle to read in their classrooms,” says product manager Lena Beate Hamborg Pedersen.
Unleashing the potential of AI in news
Unleashing the potential of AI in news
In the fast-paced digital world, the news media industry stands on the brink of a revolutionary shift. AI will shape the future of journalism and content creation. Ian Vännman from Schibsted Futures Lab predicts several phenomena that will drive the transformation, as he looks into the technology behind it.
By Ian Vännman
Unleashing the potential of AI in news
In the fast-paced digital world, the news media industry stands on the brink of a revolutionary shift. AI will shape the future of journalism and content creation. Ian Vännman from Schibsted Futures Lab predicts several phenomena that will drive the transformation, as he looks into the technology behind it.
By Ian Vännman
AI is the catalyst for a transformational wave that’s redefining our reality, akin to the monumental changes brought about by the birth of the microprocessor, the emergence of personal computers, the spread of the Internet, and the ubiquity of mobile phones.
To comprehend this future better, the Schibsted Futures Lab team delves into and explores recent technological advancements. We function as scouts, scanning beyond the Schibsted horizon and using our insights to influence our colleagues to apply emerging technologies in our businesses. We also identify and examine smaller breakthroughs, as they provide clues about plausible futures.
Breakthroughs that spark innovation
History has taught us that seemingly minor technical breakthroughs can spark innovations that, over time, dramatically reshape our world. Consider, for example, Intel’s creation of the microprocessor in 1971. This paved the way for Apple to launch the personal computer in 1977. The convergence of these technologies with Stanford’s invention of TCP/IP, the networking protocol that forms the backbone of the internet, truly took off when the World Wide Web became globally popular with Netscape’s introduction of its web browser in 1994. These innovations, combined with the GSM digital mobile networking standard developed in Europe in 1987, led to the birth of the smartphone.
Thus, minor breakthroughs converge with other advancements and innovations to generate new innovations that, over time, revolutionise the world.
History has taught us that seemingly minor technical breakthroughs can spark innovations that, over time, dramatically reshape our world.
History has taught us that seemingly minor technical breakthroughs can spark innovations that, over time, dramatically reshape our world.
Recently, the Futures Lab team has been delving into groundbreaking technologies such as neural radiance fields (NeRFs) and diffusion models. NeRFs is an impressive AI-based technology that allows us to construct 3D environments using only 2D images. In essence, it enables us to use standard cameras to generate 3D objects and environments, as showcased in Luma’s apps. Diffusion models are being used to create artistic and lifelike images with only text as input, as seen in applications such as Midjourney, Dall-E, and Stable Diffusion.
While these technologies are impressive in their own right and seem almost magical from a user perspective, they pale in comparison to the innovations spurred on by the transformer architecture. This technology, developed by Google in 2017, now underpins all the leading chat-based AI services, such as ChatGPT, Anthropic’s Claude, Google’s Bard and Meta’s open-sourced Llama.
The real magic
The transformer architecture is leveraged to create large language models, often referred to as LLMs. These LLMs are trained on enormous volumes of text data, enabling them to form artificial neural networks that capture and store patterns from the data. The real magic lies within these LLMs. To draw an analogy, if ChatGPT were a car, the LLM would be its engine.
Building on the transformer architecture, OpenAI introduced another breakthrough: a new type of LLM known as Generative Pre-trained Transformers, or GPT, as in ChatGPT. Fast forward to 2023, OpenAI and its contemporaries have enhanced GPT with the ability to build tools. In simpler terms, GPT can now generate and execute code to accomplish tasks.
The advancements we’ve made in the past 50 years will likely pale in comparison to what we’ll achieve in the next 50 years, or even the next 15 years.
The advancements we’ve made in the past 50 years will likely pale in comparison to what we’ll achieve in the next 50 years, or even the next 15 years.
Several academic studies have already explored the impact of using ChatGPT across various professions, including law, customer support, programming, and creative writing. The consensus is clear – AI significantly enhances the productivity of lower-performing individuals, enabling them to accomplish more with better quality. High performers see less improvement, and in some cases, even a drop in productivity. Interestingly, early indicators suggest this productivity boost is consistent across many, if not all, white-collar disciplines.
This can be attributed to two primary factors. First, chatbots have become remarkably adept at simplifying complex tasks. Second, Gen-AI enhances creativity. While there’s ongoing debate in the scientific community about whether large language models can truly be creative, from a productivity standpoint, this is a moot point. After experiencing ChatGPT’s “creativity,” it’s clear that it’s quite adept at it.
Something bigger
But is the so-called AI revolution merely about increasing productivity by using ChatGPT and its counterparts in office work? Or is there something bigger at play here?
Comparing the CPU, the central processing unit of a computer, with the human brain, we find that they complement each other remarkably well. The CPU excels at rapidly executing instructions provided in code with structured data – tasks that humans find challenging.
Conversely, we humans excel at learning, a capability entirely absent in a CPU. We possess agency, intuition, creativity, and are multi-modal, meaning we process input and output through most of our senses.
The LLM sits somewhere between these extremes. It’s as fast as a CPU, but also capable of learning in the sense that it can be trained and fine-tuned. It possesses contextual understanding, a characteristic more akin to our brains than a CPU.
Low costs
The key takeaway is that we now have access to human-like intelligence at nearly zero cost. It’s more than just about chatbots. Large language models enable us to infuse human-like analysis, creativity, decision-making and more into workflows and processes at virtually no cost.
With this perspective, the advancements we’ve made in the past 50 years will likely pale in comparison to what we’ll achieve in the next 50 years, or even the next 15 years, for better or worse.
How can all of this play out more concretely, in one of Schibsted’s core business areas – news media?
The answer to this is that its practical implications will be vast and far-reaching. The expected transformations will challenge the very core of our traditional business models. To grasp the full breadth of AI’s potential impact, let’s first consider the fundamental business structure of the industry.
Most online businesses can be simplified into three core activities:
- Creation of goods
- Customer acquisition
- Distribution of goods
- From a financial perspective, these activities respectively translate into:
- Cost of goods sold
- Sales and marketing expenses
- Other operating expenses
Historically, the advent of the internet drastically reduced distribution costs in the news media, triggering substantial shifts in how content reached consumers and removing most barriers to entry into the market. Now, as we usher in the era of AI, we stand on the precipice of another profound change: a potential collapse in content creation costs. The ramifications of such a shift could be as transformative, if not more so, than the internet’s earlier influence on the business models and the broader industry landscape. In the short term, I predict several phenomena that are set to drive our transformation:
Democratisation of programming
Anyone can develop software using tools like ChatGPT and Replit. All it requires is a bit of curiosity and courage. This democratisation signifies not just more efficient programming, but an increase in the number of programmers, which will further accelerate digitalisation and innovation. As Sam Altman, CEO of OpenAI, puts it:
“I think the world is going to find out that if you can have ten times as much code at the same price, you can just use even more.”
Automation of content creation
Content with predictable production processes and performance, often format-driven, such as news summaries, listicles and interviews, will likely be generated either entirely by AI or more commonly in collaboration with journalists.
Unbundling of research and narrative
Traditionally, journalism involved researching facts and weaving them into a cohesive narrative. With AI, we can separate these processes. For instance, we can publish research material alongside articles, enabling readers to explore the underlying research through an AI-driven chat interface. Newsrooms may even have teams dedicated solely to establishing and verifying facts and other information building blocks, which are then used to automatically create content using AI.
Writing of previously unwritten stories
Many individuals possess important stories that remain untold due to a lack of competence in content production. With AI, these barriers between lower and higher performers are reduced, allowing many more voices to be heard.
Personalised consumption
Every individual has unique consumption preferences. With AI’s ability to transform text into various formats, we can cater to these individual needs more effectively, especially when mastering the arts of unbundling research and narrative, as well as the automation of content creation.
With the collapse in costs and barriers in distribution and content creation, customer acquisition becomes the primary competition area for both new and incumbent brands.
To succeed in this new paradigm, I’ve identified at least four distinct, but not mutually exclusive, strategies that news media brands can deploy.
1. Creating an addictive product
Develop a service so engaging that it captures users’ attention far more than traditional news outlets. The prime example is TikTok, which holds users’ attention for an average of 90 minutes daily. Achieving this is extremely, challenging, likely impossible, but the payoff is tremendous if accomplished.
2. Fostering a movement
Tap into deeper emotions such as fear and hope to capture audiences’ energy and passion, generating extraordinarily high engagement and loyalty. Fox News, for better or worse, has done this. There is no doubt that in these times of high uncertainty, audiences are yearning for hopeful narratives.
3. Nurturing a trusted brand
This is the go-to strategy for established brands. Establishing and maintaining credibility in an era of information overload should be rewarding. However, in a future hostile media landscape, no matter how strong the brand is, brands will require greater degrees of discipline, transparency, and accountability than before.
4. Building a community
In a world of increasingly personalised experiences, individuals will seek shared interactions and rewarding experiences. This insight isn’t new for news media, but most previous attempts to build communities encountered scaling issues as the community grew, leading to its downfall. This paradox may be resolved if we can leverage AI to address the challenges that arise as the community expands.
Technology of the present
AI is not a technology of the future anymore; it’s very much a technology of the present. Every media organisation must actively engage with AI tools and platforms. Training your teams on platforms like ChatGPT or similar AI tools can lead to innovative storytelling techniques, streamlined content production, and a deeper understanding of audience behaviour.
On a personal level, embracing this new paradigm means integrating AI into your daily routine. You need to incorporate it into your life to such an extent that you automatically turn to it whenever you face challenges that require collaboration, or that can be solved faster and more effectively than you or your colleagues can do on your own. Only when it becomes an integral part of your life will you be able to fully understand it and its potential.
Rethink the pipeline
The barriers to software development are being lowered every day. Embrace this democratisation by encouraging your teams to experiment. Host internal hackathons or workshops. Foster a culture of prototyping; this not only breeds innovation but also promotes a fail-fast mentality in which learnings are quickly integrated.
With AI’s capabilities, media organisations have the opportunity to rethink their content production pipeline. Centralising certain production elements can help maintain consistency while leveraging AI can ensure content is tailored to audience preferences. Moreover, AI can assist in identifying content trends and predicting audience interests.
The transformative power of AI in the journalism industry is undeniable. We stand at a crossroads, facing a horizon with enormous uncertainty, limitless opportunities and inevitable challenges. The technological power that AI presents has profound implications on how we produce, distribute and consume news. As AI shapes a new paradigm for humanity, it becomes imperative for the journalism industry to not just adapt but lead the way. By wholeheartedly embracing AI, media brands can redefine their narrative in this new era. This journey won’t be without pitfalls, but the rewards – both for the industry and society at large – are immense. The future of journalism, powered by AI, awaits.
Ian Vännman
Strategy Advisor, Schibsted.
Years in Schibsted: 23
My favourite song the last decade: I Don’t live Here Anymore – The War on Drugs
Meet Our People 2024: Technology
Meet our people
Hanna Lindqvist is leading a project to create a new common platform for Schibsted’s marketplaces. Eivind Hjertholm Fiskerud is coordinating AI initiatives within news media and Deng Wuor Joak is detecting vulnerabilitis. Meet three Schibsted people working with tech.
Meet our people
Hanna Lindqvist is leading a project to create a new common platform for Schibsted’s marketplaces. Eivind Hjertholm Fiskerud is coordinating AI initiatives within news media and Deng Wuor Joak is detecting vulnerabilitis. Meet three Schibsted people working with tech.
Giving users the best experience
After leading some large transformation projects at Svenska Spel and PayPal, Hanna Lindqvist was sure of one thing – never again. Nevertheless, she’s once again in charge of just such a project in Schibsted, creating a common platform for all Nordic marketplaces.
But Hanna seems calm and confident. The change is motivated by a clear goal: to give users of Schibsted marketplaces the best possible experience.
“When I came to Blocket two years ago, we worked as if conditions were the same for selling different things on the same platform, but different in different countries. Now we know that the user needs are similar when selling a car in all countries – but different from selling a t-shirt.”
That’s why Finn in Norway, Blocket in Sweden, Tori and Oikotie in Finland, and DBA and Bilbasen in Denmark are now cooperating to develop the best service in four categories. They will do so from the same new platform, a platform developed from one of Schibsted’s four exciting marketplace platforms, but good stuff from the others will be integrated.
“We talk about closing gaps,” Hanna explains. “This means testing and finding out what works and fixing what doesn’t.”
Making a shift like this is very business critical – and complicated. You can’t fail, and as you make the transition, business must run as usual.
The strongest upside for Hanna is working with all the people in different countries, and that confidence of hers relies on one simple fact.
“The transformation is made by our own people, people who know the platforms and the business.”
Hanna Lindqvist
SVP Technology, Schibsted and SVP Aurora Foundations
Years in Schibsted: 2
My favourite song the last decade: Wake Me Up – Avicii
He hacks to keep companies secure
He started his Schibsted journey as an engineer at Mitt Anbud four years ago. After coming second in a Schibsted hacking competition, a part of Schibsted’s security program, Deng Wuor Joak transitioned into the field of cyber security. He has since built a security program for both Mitt Anbud and Distribution Innovation to make sure that their systems are safe.
Distribution Innovation has developed new technology for logistics and more than 2.2 million deliveries each night are enabled by their services in Sweden, Norway and Finland.
“If we get hacked, most of the newspapers and parcels that are supposed to land on the doorsteps in a large part of the Nordics wouldn’t get there,” Deng says.
When he started at DI, the company lacked a security program, so he built one from scratch. The company’s security has gone through a large transition, since the development of this security strategy.
And Deng knows what he’s fighting for. After the hacking competition, his boss at Mitt Anbud asked him to ethically hack their site. As a result of the hacking, Deng found a severe vulnerability in their systems that he presented to other engineers. This led to a job offer from Distribution Innovation.
“Hacking got me into this field and I still find it challenging and exciting to find potential vulnerabilities.”
Deng Wuor Joak
Head of Cyber Security, Distribution Innovation, Schibsted
Years in Schibsted: 4
My favourite song the last decade: Mamma Sa – Jonas Benyoub
Using AI to strengthen news
Eivind Hjertholm Fiskerud has a long and storied history in Schibsted, having worked in several positions within Schibsted and Bergens Tidende. Now, he serves as Project Lead for Nextgen Newsrooms in Schibsted News Media, taking a central role in our work with AI.
“I’ve long been very interested in how we can use data to strengthen our news media,” Eivind says. When it comes to using AI in News Media, Eivind says it comes down to strengthening the journalists in their profession. Enhancing investigative journalism, using AI for creative storytelling, and making every day easier for our journalists by using AI for time-consuming tasks, letting them focus on the journalistic core.
News Media’s work in AI begins with a clear stance on how we use AI, grounded in our longstanding tradition of journalistic ethics.
“It’s important for Schibsted to be in the forefront in this area, to take responsibility and make sure AI is used in a responsible and trustworthy manner,” Eivind says.
Eivind Hjertholm Fiskerud
Project Lead AI, Nextgen Newsrooms
Years in Schibsted: 16
My favourite song the last decade: To Mminutter – Lars Vaular & Röyksopp
Making immersive sound accessible for all listeners
Making immersive sound accessible for all listeners
Trondheim-based Nomono’s goal is to make recording just as easy as having a conversation. Now their sound capsule is ready for the market.
By Sylvia Brudeli
Making immersive sound accessible for all listeners
Trondheim-based Nomono’s goal is to make recording just as easy as having a conversation. Now their sound capsule is ready for the market.
By Sylvia Brudeli
The first snow of the season would soon blanket the city of Trondheim. With an intent gaze, Audun Solvang explored the NxtMedia 2016 conference. He was there to listen to the latest buzz in the world of audio and video. New innovations in virtual reality and 360 videos were popping up. And smartphones were on course to replace professional cameras. But it occurred to him that audio was not on par. Both in quality and ease of use. While video production was light years ahead.
With a PhD in 3D audio and ambisonics, Audun saw the untapped potential.
“I knew that If I could put the power of spatial audio in the hands of every content creator, it would transform the industry. In a big way,” he says.
Spatial audio basically means a feeling of the sound coming from all around you. Audun describes the experience as immersive, and then and there on that snowy day in Trondheim, he knew he wanted to make that feeling accessible to all listeners.
With extensive experience in the field, he went back to where he worked, down in the labs at Sintef Digital, an institute that conducts research and innovation in digital technologies and technology-oriented social sciences.
Peter Ringset, from the Nomono team.
Peter Ringset, from the Nomono team.
In 2018, he made a breakthrough with spatial positioning of microphones. Getting really close to the experience he had imagined.
By this time, podcasting was booming and support for spatial audio playback was coming to headphones and speakers. But the big bottleneck was still recording and post-production of spatial audio.
To get things going, he realised he needed help. Audun reached out to his former colleagues and experts in the area – Sigurd Saue and Jonas Rinde – about his idea. At the time, Jonas was CEO at Huddly and Sigurd was an associated professor in music technology at NTNU, the Norwegian University of Science and Technology.
“Audun had no problem convincing me that the project was technically sound and had great potential. The prospect of creating an audio tech company right here in Trondheim was truly appealing,” says Sigurd Saue.
Full perspective
Together, they set up a team of people with different backgrounds and experience levels across signal processing, user experience, consumer electronics and media production, so that they had a full perspective on what needed to be done and which challenges to solve. In 2019, Nomono was born.
From the very start, the goal has been to make recording and sharing stories just as easy as having a conversation. And to make it feel just as natural. The team envisions a new way of producing audio that is intuitive, creative, and connected from start to finish.
“Many challenges exist between recording and publishing. Getting ready and setting up equipment can feel like a time drain. Cables, microphones and obtrusive stands get in the way of natural conversations,” Audun explains.
Helge Berg Bache, Head of supply chain, opens up a Space Recorder.
Helge Berg Bache, Head of supply chain, opens up a Space Recorder.
Being bound to a physical studio can also limit imagination, and laborious editing and audio cleanup can distract the creator from being creative with the content. Faced with heaps of hurdles and complexities, even the most experienced creators sacrifice quality when crafting a story.
These are the fundamental problems that Audun and his team set out to solve.
“The whole idea is that anyone can just bring our kit to wherever you have a conversation, and simply push one button to start recording it.”
Reduces noise
Early 2023, the first Nomono portable kit for recording conversations and interviews in the field was ready for market. It consists of a recorder, four wireless mics, a charging case and access to a cloud service. But a challenging compliance process put sales on hold.
One of the secrets behind Nomono’s high-quality sound is an AI-based denoiser that identifies and reduces background and handles noise. Another AI-based method reduces crosstalk between the voice tracks. And then there’s that issue of spatial sound. Nomono solves that by spreading out voices spatially using localisation data from the mics.
With the Nomono workflow, your stories are also easy to transfer from recording to publishing. When you have finished your recording, all content is automatically uploaded to the cloud service, where you then can choose to enhance or spatialise your audio recording. It solves your needs for basic editing, including adding intros, outros and jingles, making the podcast fully ready for publishing either as a spatial audio podcast or as a normal stereo podcast.
At the beginning of September 2023, the sound capsule was fully compliant and ready for on-demand sale. Since then, Nomono has started to execute on their refined go-to-market strategy, inviting the 40,000 potential customers from their waiting list to finally be able to buy the kit after a long wait.
As a source of extra encouragement, they also won the 2023 SXSW Innovation Award in the audio category at the SXSW tech festival in Austin, Texas.
“With this award, a big milestone in the journey towards our vision of enabling spatial audio for storytellers was reached. A really nice proof that we are doing the right thing,” says Audun.
Beyond the technology, the team is also convinced that storytelling is a powerful thing, and that it connects people.
“Great stories entertain our emotions and expand our knowledge. And it has done so since the era of cave paintings all the way to the rise of podcasting today,” says the Chief Design Officer of Nomono, Viktor Rydal.
Fact
2019 Nomono was founded.
34 people are employed at the company.
Awarded the 2023 SXSW Innovation in the audio category.
Schibsted invested in the company in September 2022.
Sylvia Brudeli
Chief Product Officer, Nomono
Years in Schibsted: 1
My favourite song the last decade: Smilet i ditt eget speil – Chris Holsten
Tech trends 2024
Tech trends 2024
5G, quantum computing and clean tech. Here are some tech trends we will hear more about in the upcoming year.
Tech trends 2024
5G, quantum computing and clean tech. Here are some tech trends we will hear more about in the upcoming year.
Companies are saying goodbye to “the world’s factory”
We’re seeing a lot of companies moving their production out of China, “the world’s factory,” for several reasons. Many decided to start diversifying their supply chains following the chaos of the pandemic, and several countries, especially in South Asia, made it clear that they wanted in. The Indian government started reducing taxes and regulations on businesses and investing in infrastructure to make it a more attractive option. We’ve already seen Apple move some of its production of the iPhone to India, now producing 7% of its iPhones there. Other big players like Intel, Microsoft and Nike are talking about moving production elsewhere. This is likely only the start of this trend.
Will we be able to develop completely new strategies and devices with quantum computing?
Will we be able to develop completely new strategies and devices with quantum computing?
Is quantum computing the next revolution?
We’re expecting to see continued development of quantum computing. In 2024, IBM is set to launch a European quantum data centre complying with GDPR. The advancements in quantum computing could revolutionize banking, medicine, cryptography and other industries handling large amounts of data. As we wrote back in 2022, a major task that lies ahead will be to explore the applications of quantum computing, to think differently and to develop completely new strategies and devices for this technology.
3,3% is the expected growth rate for energy consumption in 2024, according to the IEA.
3,3% is the expected growth rate for energy consumption in 2024, according to the IEA.
Global power grid instability
3,3% is the expected growth rate for energy consumption in 2024, according to the IEA. Following the economic downturn of the last year and the subsequent energy crisis, an increased demand could lead to power grid instability globally. Businesses need to prepare to be more energy-efficient and self-reliant in the coming years.
Big tech investing in healthcare
Healthtech will be revolutionized by AI, especially in diagnosis and drug creation where a massive amount of data points need to be analysed. We also expect remote healthcare to grow, along with wearable technology being used for preventative care. Apple is working on an AI-powered health coaching service and new technology for tracking emotions, and Spotify founder Daniel Ek launched the preventative care startup Neko this year.
5G networks have the potential to lower costs and improve efficiency in smart ciities.
5G networks have the potential to lower costs and improve efficiency in smart ciities.
5G and smart cities
As 5G networks expand, we can expect the development of new technologies and services as well as the evolution of current ones. Smart cities with updated infrastructure have the potential to lower costs, improve efficiency, and increase security and interconnectivity.
We can expect more premium versions of social media accounts.
We can expect more premium versions of social media accounts.
Social media can come at a cost
Social media platforms are looking for more revenue streams as restrictions on targeted advertising are coming from tech giants like Apple and EU regulators. One avenue they’re exploring is paywalling: X (formerly Twitter) is charging for verification and Snapchat is offering additional features for a monthly fee – which over two million people have paid for. Meta and TikTok are also exploring these options, and we’re likely to see more premium versions of social media accounts soon.
Global investments in clean energy technologies is expected to rise faster than investment in fossil fuels.
Global investments in clean energy technologies is expected to rise faster than investment in fossil fuels.
Clean energy ahead of fossil
As global climate and energy concerns keep growing, we expect to see developments in cleantech with further transitioning to renewable energy sources. The IEA expects global investments in clean energy technologies to rise faster than investments in fossil fuels, and climate fintech is one of the strongest innovation areas for European startups.
71% of people would already rather utilize voice tech than a keyboard when searching online, PWC research suggests. We’ll continue to see voice-powered technology develop through AI and natural language processing, and voice assistants will be further enhanced.
Less connection, more consumption
Social media has become less social and more media. There have been several signs that social media platforms aren’t living up to their proclaimed mission. All the most popular spaces have, or are, moving from a place to connect with friends to another place to consume media. They’ve entered a space of increased competition for the consumer’s attention, fighting not only other social platforms but also all other media.
EU regulations – a balancing act
EU regulations – a balancing act
A tsunami of digital regulation is on its way. Schibsted is calling for regulations that promote circular economy and innovation while protecting consumers.
By Petra Wikström
Today, we live in a very different world than it was just a few years ago. Everything has changed; the geopolitical landscape, the energy market, and the cost of living. But what has also changed is the view on regulating the Internet. Some years ago, most politicians in the Nordics believed that the Internet should be free from rules and that a liberal regime was the only guardian of innovation. They said that tech companies should not be liable for the content on their platforms and big tech should be able to grow by acquiring start-ups.
Instruments with huge impact
The pendulum has now swung the other way, and in 2023 we see overwhelming support from Nordic decision-makers for the EU landmark regulations of the digital landscape, namely the Digital Services Act (DSA) and the Digital Markets Act (DMA). Both regulations were adopted by the EU institutions in 2022 and will become a reality at the start of 2024.
These instruments will have a huge impact on the way platforms must deal with liability for illegal content, but also how big tech companies should deal with their business users. The guiding principles have been to create fair and transparent rules that level the competition in the market and protect users and consumers from unfair commercial practices.
But while the EU wants to boost the European digital economy, it also wants to create a safe internet based on European values. This is a tricky balancing act between those who want liberal rules that allow for innovation and increased global competition and those who want heavy regulation that protects consumers of digital services.
EU is balancing between liberal rules and heavy regulation to protect consumers.
EU is balancing between liberal rules and heavy regulation to protect consumers.
Recent global events have undoubtedly strengthened the perception among EU policymakers that the EU needs to be its own strong force and promote a distinct “third way” in regulating the digital economy, somewhere between the USA’s “laissez-faire” and Chinese authoritarianism.
This will and has already led to more regulatory oversight and enforcement powers at the EU level, making Brussels an ever-more important focus for tech regulation.
Increased call for liability
During the past five years, we have seen a “tsunami” of proposed legisation affecting online marketplaces especially. Schibsted is currently monitoring or actively lobbying more than ten different EU proposals that have an impact on our businesses. The main reason for this regulation frenzy is the lack of trust by decision-makers in new online platforms operating in the EU. Many platforms are established outside Europe, which has led to an increased call for greater liability over the products these online marketplaces offer, as well as more control over the businesses selling the products and more collaboration with national authorities.
In addition to increased consumer protection, the EU has intensified its efforts to promote a transition towards a circular economy through various proposals. These are expected to lead to more durable and repairable products in the EU and more trustworthy environmental information being communicated by businesses.
The EU is also updating its VAT rules for the digital age and has increased the reporting requirements to tax authorities regarding sellers who sell many products via online platforms. All these rules will require changes to many digital services, including Schibsted’s second-hand marketplaces. In particular, we will have to make changes to how we collect information about our business sellers and we must start conducting random checks for dangerous or illegal products among the products listed on our platforms. Although we support the regulation, we think that it is important that the regulation does not require us to control the sale of second-hand products between consumers, as that would risk alienating consumers from selling their used products and throwing them away instead.
Many proposals are promoting a more circular economy. These are expected to lead products that will last longer and be easier to repair.
Many proposals are promoting a more circular economy. These are expected to lead products that will last longer and be easier to repair.
Perhaps the biggest and most difficult initiative of the EU Commission is its aim to be the first in the world to regulate artificial intelligence (AI).
The EU’s proposal for an AI Act was presented in April 2021, and it is still being negotiated by the EU institutions. The proposed legal framework intends to regulate the specific use of AI-systems associated with risks to citizens’ health and safety, building on a “risk-based approach.”
Some AI systems presenting “unacceptable” risks are proposed to be prohibited, such as subliminal techniques, remote biometric data (such as facial recognition cameras in public spaces) and AI systems for social scoring.
According to the proposal, a wide range of “high-risk” AI systems can be used and put on the EU internal market, if they comply with a broad list of requirements and obligations, including an extensive risk assessment procedure. High-risk AI use cases include AI systems used within education, access to essential private and public services, recruitment and HR systems, law enforcement and administration of justice and democratic processes. And some AI systems that only present a “limited risk” are proposed to be subject to transparency obligations.
Allow for innovation
Schibsted actively uses AI in its businesses. We, therefore, support some regulation of AI that focuses on the apparent risks to health and safety but still allows for digital companies in Europe to innovate and develop AI tools. This possibility should not only be left to the global big tech companies. We also need to ensure that we have good large language models for our Nordic languages and that we have control over how our own editorial content is used to train AI models.
Over the next five years, we expect that the EU will continue to regulate online marketplaces. The EU Commission initiated a fitness check of EU consumer law in 2022, to assess its efficacy in protecting consumers in the digital domain. According to a public consultation by the EU Commission, consumer representatives listed manipulative interface design (88.7%) and unfair use of personal data to personalise commercial offers (74.3%) among the five most problematic and frequently occurring practices.
We believe that these sentiments among consumer representatives, such as the national consumer associations, may lead to proposals for new consumer regulation during the next EU Commission mandate. This includes, for example, practices or so-called dark patterns that are perceived to manipulate and “steer” users’ choices and inhibit freedom of choice online.
One of the most discussed and contested issues in digital regulation is the use of personal data for targeted advertising. Many EU decision-makers are frustrated with the way big tech companies in particular use personal data for targeted advertising. And there is a clear political push to do more to protect privacy online.
We also think that the EU will continue to regulate AI, depending on when the AI Act will enter into force and which issues may be left unregulated due to time not yet being ripe for strict regulation of AI.
Schibsted is contributing to these various proposals by calling for a regulatory landscape that promotes a circular economy, and allows for the innovation of state-of-the-art digital products and services, all while protecting consumers from harmful practices in the online market. As the next EU mandate approaches, we are developing our positions on increased consumer protection, enhanced privacy and new AI regulation, which we plan to present to the EU institutions in 2024, in an effort to influence the next EU Commission work programme.
Petra Wikström
Senior Director of Public Policy, Schibsted
Years in Schibsted: 5
My favourite song the last decade: Blinding Lights – The Weeknd
Building a Norwegian language model
Building a Norwegian language model
ChatGPT’s ability to generate human-like language is impressive. But now it’s important to also build specific models for the small languages like Norwegian.
By John Einar Sandvand
Building a Norwegian language model
ChatGPT’s ability to generate human-like language is impressive. But now it’s important to also build specific models for the small languages like Norwegian.
By John Einar Sandvand
“We need a Norwegian language model, built mainly on Norwegian text,” says Schibsted’s Chief Data & Technology Officer, Sven Størmer Thaulow.
Sven is currently the chair of the Norwegian Research Center for AI Innovation (NorwAI) at NTNU (the Norwegian University of Science and Technology) in Trondheim. Schibsted is one of several industrial partners of NorwAI – and contributes both with both competence and data.
One of the big projects at NorwAI is to build a generative language model for the Norwegian language. The work has been ongoing for more than two years, and the first version was launched last summer. Schibsted has contributed thousands of articles for the model to be trained on. This is a non-commercial research project at present, and Schibsted will be among the first to test how well it performs compared to the big American models.
Why is a Norwegian language model needed when ChatGPT works quite well in all the Scandinavian languages?
Sven shares three main reasons:
Better in Norwegian: A model trained primarily on content in the Norwegian language will likely also be better in Norwegian. To compare, we estimate that only 0.1% of the content ChatGPT was trained on was in Norwegian.
Control over our own infrastructure: Large language models are becoming part of our digital infrastructure. But we see that artificial intelligence is already turning into a global industrial political race. It is not obvious that the technology will be democratised. Therefore, we must develop our own large language models that serve our societies well and can be the basis for innovation and the development of new services for our population.
Consistent with Norwegian culture: We need language models that reflect the value sets of our Nordic societies rather than being dominated by American perspectives. Language models can easily be biased, for instance, because of the content they are trained on or how they have been adjusted. By training our own models there is a greater chance that the output will better reflect our culture and values.
Building a large language model is an enormous effort. It requires vast amounts of text, specialised competence, and enormous computing power. And early this year, Sven invited all media companies in Norway to contribute content to the work of building the model.
“We need content that is representative of the full Norwegian society, from news articles, simple chats, government documents, court verdicts – to even the most beautiful novel,” Sven says.
He adds that a successful Norwegian language model can be a shared resource that will create much value – for Norwegian society at large, for companies like Schibsted and others, as well as for individuals.
John Einar Sandvand
Senior Communications Manager, Schibsted
Years in Schibsted: 30
My favourite song the last decade: Save Your Tears – The Weeknd
Redefining our digital lives
Whatever shape the metaverse eventually takes, and whatever name we end up calling it, profound changes are coming. We’d best be ready.
Redefining our digital lives
The metaverse is a term that defines the optimism and hubris of the tech industry today. Will this utopic vision for the future of the internet usher us into brave new virtual worlds, or will it fall by the wayside as Meta’s last-ditch effort to save their business?
Few topics in the tech world are attracting such unbridled optimism and derisive scepticism as the metaverse, with, perhaps, the exception of NFTs. And this year, Meta and countless other tech companies have doubled down on reorienting their businesses towards building “the next chapter of the internet.” Around the world, executives and the consultants they hire took note, with one of McKinsey’s forecasts topping out at a staggering USD 5 trillion valuation for the metaverse by 2030.
Countless definitions
But what are we actually talking about when we refer to the metaverse? Countless definitions have been written this year, and they generally follow a similar recipe to this one offered by Gartner: the metaverse is a “persistent, immersive and all-encompassing joining of our physical and digital lives, bringing together work, entertainment, marketplaces and social spaces in one seamless experience.”
In practical terms, this definition relies heavily on rapidly improving extended reality (XR) technologies like virtual reality (VR), augmented reality (AR) and ambient computing (voice assistants and home automation).
While this definition may still be somewhat vague, the idea behind it isn’t new. The term itself was coined by Neal Stephenson in his 1992 novel Snow Crash. Stephenson describes a dystopian future in which the only escape from failed states and corporate warfare is to strap on a headset and dive into a three-dimensional digital world complete with personalised avatars, virtual real estate, transportation networks and nightclubs.
Epic Games’ Fortnite is a traditional computer game that has successfully integrated purchases of virtual goods.
A profound cultural impact
Stephenson’s concept builds on the cyberpunk traditions established in works like William Gibson’s novel Neuromancer, which popularised the term Cyberspace and imagined a virtual reality environment he called “the Matrix”. The cyberpunk genre of science-fiction has had profound cultural impact, not least of which is a long line of dystopian films, a surprising number of which star Keanu Reeves.
The people who read books like Snow Crash and Neuromancer as teenagers grew up to found global tech companies. But they focused on an escapist, tech-utopic vision of the metaverse, leaving aside the critique of unbridled capitalism and technocratic rule that defines the cyberpunk genre. While these contemporary visions of the metaverse should be viewed with healthy scepticism, they offer a profound reimagining of our digital lives and crystalise a vision for where current technological developments might lead.
Get to know Hanna Ek’s avatar – Hanna. The Schibsted Tech experiment Team created her in just a bit over 30 minutes. The original Hanna is part of the team.
The obvious starting point here is Meta. With Facebook losing daily users and facing increasing obstacles for their ad business due to privacy initiatives in Apple and Google’s mobile platforms, it’s no wonder that Meta is searching for new revenue streams and new ways of connecting to users. Mark Zuckerberg has famously tied the future of his business to a future augmented reality platform that he hopes will replace the mobile phone, breaking Meta’s reliance on the mobile giants for access to users.
A major milestone
The release of the Oculus Quest Pro, a pass-through VR/AR headset that allows for both full immersion and for virtual elements to be added to real environments, is a major milestone towards this vision. Meta hopes that one day we will all use improved versions of headsets like the Quest Pro to access Horizon, a virtual reality social network made up of virtual environments for socialising, entertainment and work.
The vision here is compelling. In one direction, VR technology lets us physically enter virtual worlds and create powerful experiences. Imagine history classes around the world taking place in virtual reconstructions of historic sites, or the higher levels of remote collaboration and social connection enabled by a true feeling of shared presence. Going the other way, AR technology allows us to pull virtual elements into our physical world, opening entirely new possibilities for fashion, art and education and augmenting everything from buying new furniture to sports analysis.
The potential of Meta’s new technology and platforms is as impressive as the lurching struggles the company has gone through as they try to realise this vision. Buggy experiences and unclear use cases have meant that most people who try Horizon don’t return to the platform after the first month. Furthermore, more than half of previous generation Quest VR headsets aren’t in use within six months of purchase. Most damning, perhaps, are the recently leaked memos from within Meta that suggest the teams building Horizon aren’t even using it themselves, instead they are preoccupied with making critical improvements to basic elements of the user experience.
Will the metaverse be a system of closed and proprietary platforms?
These struggles shouldn’t be confused with a condemnation of the visions of a metaverse that Meta and others put forward. Apple is reportedly getting ready to release their own VR/AR headset. While they remain tight-lipped, Apple is no doubt expected to produce high-end hardware complete with an accompanying closed platform and marketplace for apps and experiences.
In anticipation of an increased threat from Apple, Mark Zuckerberg has called this a “competition of philosophies and ideas”, where they believe that by doing everything themselves and tightly integrating that, they build a better consumer experience. Meta’s businesses, including Facebook and Instagram, rely on accessing the closed platforms controlled by Apple’s iOS and Google’s Android. Unsurprisingly, then, Meta has championed interoperability, co-founding the Metaverse Open Standards Group (Google is a member, while Apple is not) to develop mechanisms that let users seamlessly travel between different virtual worlds, complete with whatever virtual purchases they may have made.
Will the metaverse be a system of closed and proprietary platforms?
This push for open standards, as well as the sheer number of partner organisations participating, only highlights how pervasive and motivating the idea of the metaverse is in the tech world today. Each new investment and product is being evaluated in terms of how it relates to the metaverse.
Epic Games, the creator of Fortnite, is proving the role that game engines and graphics-focused companies have in building the metaverse. Epic’s Unreal Engine is being touted as one of the defining creative tools used in creating virtual worlds and virtual humans.
Crypto spurring virtual worlds
There is, of course, a role here for crypto as well. Virtual worlds built on blockchains have emerged in recent years, spurring a run on virtual real estate and playing host to massive exhibitions of NFT art.
It’s a star that shines so bright that long-standing investments by established tech players also are being recast as bets on the metaverse. After the seeming failure of Google Glass, the search engine giant has planted a new flag in this arena, demoing an incredibly slim pair of glasses that offers live translation in a heads-up display.
Microsoft already has a long history of building AR solutions for industry. Their Hololens products enable workers in heavy industry to view instructions contextually as annotations in their environments. Along with their Azure cloud platform and advancements in AI, Microsoft is also developing data-driven digital twins. These virtual copies of real-world equipment enable new levels of simulation, monitoring and preventative maintenance. Digital twins enable x-ray vision to workers wearing the Hololens, giving them a data-enhanced view into the inner workings of complex machinery.
Taking a different approach with the potential to complicate Meta’s vision, Amazon is building its strategy around what it calls “ambient intelligence”. They are building on the massive popularity of their Alexa-enabled devices. Amazon hopes to create a predictive, invisible layer of computing in our physical world, and in so doing, ship what you want to buy before you know you want it.
Looking across all these examples, we see common elements that point towards a more specific definition of the metaverse. Matthew Ball, a venture capitalist and the author of landmark articles and a book on the subject, identifies a range of technologies that would require advancements to fulfil a system like the Metaverse, including consumer devices, computation and networking, digital platforms, governance, standards and payment and ownership frameworks (did someone say crypto?), to name a few.
So, the metaverse cannot be seen as a single technology in and of itself. Rather, the term should be viewed as a catch-all concept that brings together a range of technologies in different states of development. Breakdowns like these help us make sense of the bold claims and grand visions coming from the tech industry, but do these visions of a future internet hold water? Can we ever really know what is coming?
Three core uncertainties
As history shows, we can’t fully predict what the metaverse will truly be or enable. But three core uncertainties will define the future shape of the metaverse: technology, power structures and experience.
First, we can’t accurately predict what emergent properties will arise from the combination of novel and developing technologies. Will these technologies converge, allowing the metaverse to become a new dominant platform for accessing digital content and replacing the web as we know it? Or will it remain separate yet connected to today’s web, simply offering new spaces and modes of interaction for us to explore?
Second, what power structures will define the metaverse? Will the metaverse represent an open ecosystem, freely accessible across devices and platforms, or will it instead be a system of closed and proprietary platforms, with limited interoperability and a high level of control resting in the hands of private companies? And on top of that, will national and regional authorities learn from early attempts at regulating the internet and take a stronger approach to governing the metaverse?
Finally, how will our personal experience of the metaverse be? Will we spend more of our time diving into the metaverse, leaving behind our physical reality in favour of the digital, or will the metaverse come out to meet us in the physical world, fit into our daily lives and maybe even become mundane?
Many possible futures
Developments are underway right now that will radically redefine our digital lives. The visions put forth by tech companies like Meta represent a clear image and a path to one possible future, but this future is just one among many. We can’t say with any certainty what form the metaverse will take, but something is coming. Despite the uncertainties defining the future of the metaverse, we do know two things.
One, our digital lives are becoming more spatial, more human-shaped. As the lines between the digital and physical blur, we will use more of our bodies and more of our senses to interact with computers, just as computers will become more embedded and invisible in our daily lives.
And two, our digital lives are also becoming more social. Younger generations don’t differentiate between their physical and digital lives, just as it’s becoming harder to differentiate between what is a game, what is a concert and what is a social network.
All signs suggest that the metaverse will be far more spatial and far more social than the web we know today. Whatever shape the metaverse eventually takes, and whatever name we end up calling it, profound changes are coming. We’d best be ready.
Many different technologies and trends are part of what we call the metaverse. Schibsted Futures Lab guides us to a few that are mayor drivers for our expanded digital lives and experiences.
Christopher Pearsell-Ross
UX Designer, Schibsted Futures Lab
Years in Schibsted: 1.5
Technologies pushing our digital lives
Technologies pushing our digital lives
Many different technologies and trends are part of what we call the metaverse. Schibsted Futures Lab guides us to a few that are major drivers for our expanded digital lives and experiences.
Our extended realities are growing
We have seen continuous growth in the adoption of extended reality (XR). It is predicted that the worldwide AR and VR industry would reach USD 209 billion by 2022. Aside from the gaming, healthcare and education sectors, there is also an emerging demand for digitally immersive customer experiences. Interacting with the goods digitally before visiting the store, from shopping for clothes to trying on make-up, or even searching for your next dream home, has made XR crucial in the purchasing process.
A hurdle to the widespread adoption of these technologies is the lack of multi-platform standards. To break down this barrier, WebXR has been introduced as the future web standard to facilitate XR experiences through a URL. It is cheaper to implement without any additional hardware or software. Not only does web XR make 3D content generation widely accessible to users, it could also be a game changer for enterprises to attract younger generations and scale up new businesses rapidly.
Yifan Hu,UX Designer
Game engines help speed up development
As experiences become digitally immersive, we need a framework to provide tools and libraries for building digital worlds and physics for their interactions. This led to the idea of game engines. Unity and Unreal Engine are the clear market leaders. Since their first debut in the late 90s and early 2000s, they have evolved and expanded into other industries. Unreal Engine’s metahuman creator has proved its ability of rapid avatar generation and customisation. What used to take a 3D artist months now only takes minutes.
Many other industries have also begun using game engines to create prototypes and digital twins, from car companies like Porsche to architecture firms to manufacturing plants. When more and more designs get populated in UE5, how long will it take to see your digital avatar driving a Porsche 911 down the street?
Yifan Hu,UX Designer
Synthetic media has opened Pandora’s box
If you have been online at all in the last year, you’ve likely heard something about synthetic media. AI-enabled tools for generating text and images have exploded in recent months. In September of this year, an AI-generated artwork won the Colorado State Fair art prize, likely the first time a rural art competition has been at the centre of global controversy in the fine art world. AI tools like Midjourney, which was used to generate the winning artwork, are related to the technology behind deepfakes.
Neural network AI models are trained on vast amounts of existing works, whether text, images, or, as recently released by Google, short video clips. These models can then be used to generate new works, potentially revolutionising creative industries, and in the process sparking heated debates about intellectual property and the role of human creativity. Synthetic media likely has a large role to play in content creation, not only for the metaverse, but in countless existing industries as well, not least of which is news media. Pandora’s box is open. The question now is whether we can find ways to make responsible and ethical use of what’s inside?
Christopher Pearsell-Ross, UX Designer
Blend the real with the virtual
The scope and areas of application for digital twins have significantly expanded. A digital twin works as a digital replica of a real-world object or system. It enables you to run one or several simulations with access to real-time data. This opens the possibility of studying, tracking and monitoring multiple processes with various application areas, including urban planning, healthcare services, the automotive industry, manufacturing operations, big structures and power-generation equipment. Digital twins allow for a seamless integration between digital and physical spaces, blending the real with the virtual.
Eline Wong, Junior Creative Technologist
Gaming has become social
Games are already the new social hubs for next-gen consumers. According to Newzoo’s latest report, more than 90% of Generation Alpha, Z and Millennials have engaged with video games in holistic form, including live streaming, podcasting and attending in-person live events. The revenue growth for the gaming industry has exceeded USD 200 billion.
Social gaming still performs at a high level of monetisation via advertising, in-game offers and virtual goods purchased by players. Minecraft, Roblox and Fortnite top the charts for Gen Alphas. These games share three key concepts: players can build worlds, host events, or go on adventures with their customised avatars. Since 2021, demand for building and creating tools has increased by 7% and is most relevant for young players in the eight to 15-year-old range. While many adults are still pondering what the metaverse is, these young master builders have already begun hammering down their proto-metaverses.
Yifan Hu, UX Designer
Avatars will need to become more realistic
Meta’s new release of the XR headset Quest Pro underlines the importance of avatars, which have surpassed their rudimentary perception as animated Memojis. Rather, live, hyper-realistic avatars will be crucial to how we express ourselves in the future. Meta’s so-called codec avatars and instant avatars will allow the user to generate a photorealistic digital self through a face scan with a mobile device.
These avatars should excel in the non-verbal cues, such as facial expressions and eye contact, which we rely on to enhance communication. That being said, securing your avatar, whether through encryption or an authenticated account, will become a critical challenge. Nvidia recently announced an open-source platform for building cloud-native, AI-powered avatars. From customer support agents to teaching assistants, digital humanoid interfaces will soon no longer be a fantasy.
Yifan Hu, UX Designer
Meet our People: She builds a universe of football
She builds a universe of football
At Stavanger Aftenblad, the 13 to 14-year-olds have become a loyal target group – thanks to Elin Stueland and Mååål (“Scooore”, in English). For more than a year, this service has been reporting on all the local football matches in every series. The secret? A combination of robot journalism and human reporting.
“Mååål is a universe where you want to stay – it’s close to people’s lives,” says Elin as she explains the success.
As a digital editor she was on the lookout for new projects when she and her team realised that football was an obvious choice. There are more than 100 different football clubs in the area, playing more than 7,000 matches each season. The key has been to cover all matches played by 13-year-olds and older.
“Imagine being 13 years old and seeing your own picture in Stavanger Aftenblad. Now they know us as a brand.”
To be able to do this, Stavanger Aftenblad developed a robot in co-operation with the Norwegian news agency NTB. Now local trainers give input, and the robot automatically transforms it into results and content. Two reporters are also working full time to create real stories, podcasts and live studio reporting. The service also includes live streaming from more than 80 matches.
Another important thing has been to build Mååål as a concept with its own web page, social media content, and interactions. And during the first six months, the number of subscribers who renewed their subscriptions of Stavanger Aftenblad almost doubled.
“We have shown that we take the sport and the audience seriously.”
Elin Stueland, Deputy News Editor, Stavanger Aftenblad
Years in Schibsted: 16
Einar Otto Stangvik
A hacker devoted to storytelling
Einar Otto Stangvik is a security officer at VG – and a hacker, dedicated to developing editorial tools to support news investigations and presentations. He started out as a software developer and security consultant, but he wanted to get more out of his skills. As he investigated a politician who was hacking women to access their photos, he got in touch with VG, who helped him get the story out. He ended up getting hired and revealing child pornography on the dark net – a story that went global.
Currently, he is into 3D-maps to create a new dimension of journalistic storytelling. It started off with the insight that simple maps don’t really tell you much.
“I wanted to show how the places we write about really look and what that could mean to a story.”
In a recent story about a murder case, in which two young Norwegian girls were kidnapped and killed in 2001, maps have revealed information that indicates that one of two convicted men might be innocent.
“I am really a technology sceptical person. But I believe that our newspapers need to find out how we can use technology in the best possible way to give the readers the best possible understanding of a story.”
Einar Otto Stangvik, Head of Information Security, VG
Years in Schibsted: 9
Björn Schiffler
Contextual ads don’t need to track the users
Björn Schiffler comes from academia where he specialised in cognitive neuroscience. But when he switched to data science, Schibsted became an attractive place to work – as with all its content, the opportunities to work with natural language processing (NLP) are plentiful.
Björn and his team are developing machine learning-models that support the development of contextual ads.
“It’s all based on the content itself, so we don’t need to track users or use cookies,” he explains.
Contextual ads are ads matched with relevant content, like news articles on Schibsted’s media platforms.
“For instance, if you want to sell bikes, articles about the Tour de France would be such a relevant context.”
But Björn also has his eyes on the future. He’s looking forward to developing even more advanced models that can understand the overall essence of an article and contribute with even better input to contextual advertising. Looking even further ahead, generative models could create summaries of news stories adapted to different target groups.
“This is how our field, in a concrete way, can contribute to making a positive impact.”
Björn Schiffler, Staff Data Scientist
Years in Schibsted: 1.5