Innholdsfortegnelse for EG02 i New Scientist - The Essential Guides

THE second in our series of Essential Guides examines perhaps the most revolutionary technological development in a generation, and one that is happening with bewildering speed. From small beginnings just decades ago, artificial intelligence in 2020 is already embedded in our daily lives. Feeding on vast amounts of often-personal data, it helps to determine the information we see, the news we read, the products we buy and much, much more. Technologies from driverless cars to personalised medicine form just part of its future game-changing potential. Yet AI stands accused of all sorts of misdemeanours, from undermining democracy through the spread of information to widening inequality to embedding racism and discrimination. That’s not to mention fears that machine minds might one day supersede our own. Curated from New Scientist’s extensive archive…

With traditional computer programs, the machine gets line-by-line instructions. With machine learning, however, the computer must work out how best to solve the problem. The result is a machine that essentially programs itself. Imagine instructing a robot to make soup. The conventional approach would be to write out a precise recipe for SoupBot to follow. First peel the onion, then cut the onion. But a SoupBot based on machine learning would instead work out what to do on its own, perhaps by watching thousands of videos of people making soup and trying to come up with its own soup-like recipe, or by attempting to make soup again and again and learning from feedback on the results of each attempt. In the case of SoupBot, the conventional approach would be most…

Although originally based on a loose analogy with the network of connections between neurons in the human cortex, a neural network is not a physical network in any sense. Rather, it is a complex mathematical object: a suite of algorithms constructed in such a way that, rather like the human brain, it can recognise patterns in data and so “learn” from it. The basic units – the nodes or “neurons” of the network – are algorithm routines specialised in recognising some aspect of the data the network is designed to process. In an image-recognition network, for example, lower-level neurons detect simple features – straight lines, say – and feed that information to higher-level neurons, so that increasingly complex properties can be detected. In that way, the network can begin to…

CREATED over 2500 years ago, Go is a challenge for the nimblest of minds. Its 19 by 19 board allows for 10171 possible layouts, dwarfing the roughly 1050 possible configurations on a standard 8 by 8 chess board. “Go is probably the most complex game ever devised by man,” says Demis Hassabis, founder of DeepMind, the Google-owned company that created AlphaGo. Alpha Go’s five-game match with human Go champion Lee Sedol was held in the swanky Four Seasons hotel in the heart of downtown Seoul in March 2016. Lee was a national hero in his native South Korea, known for his brashness and unconventional and creative play. Hundreds of reporters from around the world were in attendance, and the match was televised live across Korea, China and Japan. Lee flashed…

IF YOU want to beat ‘em, you must first join ‘em. That was the principle the first AI gamers followed: they were trained, at least initially, through observing thousands of instances of how humans played the games. Not so AlphaGo Zero. The zero stood for “zero data”: AlphaGo Zero was programmed simply with the rules of Go, and then started playing at random against itself, learning what worked. Three days and 4.9 million such games later, it took on AlphaGo. In a 100-game grudge match, it won 100-0, becoming the world’s best game-playing AI. “Humankind has accumulated Go knowledge from millions of games played over thousands of years,” the DeepMind programmers wrote in their paper announcing the breakthrough. “In the space of a few days… AlphaGo Zero was able to…

AI artwork renders questions of ownership and copyright in a fuzzy hue. Who owns the output: the person who built the algorithm, the person who fed it data, so it could recognise and mimic patterns, or the person who selected the specific output? The consensus among artists and lawyers seems to be that an AI is something akin to a word-processing program: if nobody types into one, there is no essay. By extension, anyone who feeds data into an AI owns the resulting work, and the AI’s creators have no claim to the outputs. “It can seem like the machine is doing everything, but without an artist to actually collect the data, the algorithm has no agency,” says AI artist Janelle Shane. It doesn’t stop there. The data sets used…

How did you end up applying your work with machine learning to oncology? There is a technical answer and a pragmatic one. Virtually every aspect of life today is regulated by machine learning, whether you know it or not. The only area that isn’t is healthcare, which involves a lot of prediction tasks. When your doctor tries to find you a treatment, they look at different clues together and make a prediction. With personalisation, which we’re all trying to achieve in medicine, the goal is matching you and your unique characteristics to the correct drug. I got into oncology because I was a breast cancer patient at Massachusetts General Hospital (MGH). Why do we need machine learning for personalised medicine? Even very experienced doctors have only seen a limited number…

LEVEL 0 No autonomous features, may have automatic gear shift. Most cars currently on the road LEVEL 1 Some autonomous features, such as automatic braking, cruise control. Many newer car models LEVEL 2 Automated steering, braking and acceleration, but requires human oversight. Eg Tesla Model S, Mercedes-Benz E Class, Volvo S90 LEVEL 3 Car can monitor its environment and drive autonomously, but may request human intervention at any time. Eg Audi A8, Nissan ProPILOT 2.0 (2020), Kia DRIVE WISE (2020) LEVEL 4 Car can drive independently but may request human intervention in unusual conditions, such as extreme weather. Eg Volvo, Tesla, Ford (2021), BMW iNext (2021) LEVEL 5 Car can drive independently in all conditions.…

As with everything in the world of modern AI, the questions start with data. Should we really be giving so much of it away? WHEN Joanna Bryson, an AI researcher at the University of Bath, UK, is in a house where she knows there is an Amazon Echo, a Google Home or any similar voice-activated digital assistant, she says she holds a more guarded conversation. She’s all-too conscious of the possibility of her words being observed, recorded, dissected. Most of us haven’t thought that far, seduced by the usefulness of such devices. Data-driven AI both feeds on and fuels a huge data-gathering infrastructure. Since the advent of the world wide web, engineers and entrepreneurs have invented myriad ways to elicit and collect information about us and our habits: asking users…

Stare into these images of static and tell me what you see. That is, in effect, what cognitive scientists Chaz Firestone and Zhenglong Zhou at Johns Hopkins University in Maryland asked 200 people to do in a recent study. The static images had been generated specifically to fool machine minds, even though there is no meaning to them as far as a human is concerned. Even so, when the people were prompted with images of a robin and asked to say which static image the AI would mistakenly classify as a robin, 81 per cent of them got it right, a higher proportion than would be expected by chance. Answer: The AI thought the right-hand image was a robin…

Why do we need to keep tabs on algorithms? We have a new kind of agent in the world. It’s not an animal or a human, it’s a machine. Today, only computer scientists explore the behaviours of algorithms. My team is studying machine behaviour as we would a human, animal or corporation – in the wild. Are you saying algorithms behave like animals? Animals are much more complicated, yet we have a long history of studying their behaviour. It’s the same for human psychology: we are far from understanding the brain, but we know a lot about human behaviour. Even if we could see behind corporate walls and access algorithmic source codes, we wouldn’t understand their impact on society. The crucial thing is that algorithms act in the world, so…

WITH BEWILDERING SPEED, AI HAS ADVANCED TO UNDERPIN MANY ASPECTS OF OUR EVERYDAY LIVES – AND THAT’S JUST THE BEGINNING. THIS SECOND ESSENTIAL GUIDE FROM THE MAKERS OF NEW SCIENTIST MAGAZINE TELLS YOU ALL YOU NEED TO KNOW ABOUT A TECHNOLOGY NO ONE CAN AFFORD TO IGNORE, INCLUDING: ❶ What AI is – and isn't ❷ How neural networks and deep learning works ❸ AI applications from medical bots to driverless cars ❹ The social and economic challenges of AI ❺ What the future holds for intelligent machines £9.99…

Like any transformative new technology, AI brings both risks and opportunities. To truly understand what they are, we must first cut through the hype – and, as Peter Norvig sets out, get to grips with some very basic questions. PROFILE PETER NORVIG Peter Norvig is director of research at Google, and previously director of its core search algorithms group. He is co-author of the classic textbook Artificial Intelligence: A modern approach ARTIFICIAL intelligence is about engineering machines that act intelligently. That raises a vexing question: what is “intelligent”? In many ways, “unintelligent” machines are already far smarter than we are. But we don’t call a computer program smart for multiplying massive numbers or keeping track of thousands of bank balances; we just say it is correct. We tend to reserve…

BACK in 2008, over 90 per cent of all emails sent and received were spam. Over the following decade, that proportion dropped by half, standing at just over 45 per cent in 2018. It just wasn’t so worthwhile sending unsolicited emails advertising dubious products and services any more – thanks to the steady improvement in AI-driven spam filters preventing them from ever reaching users’ inboxes. Spam filters, like many other AI technologies, got their first fillip from user-generated data. Individual email users provided the gold standard by labelling messages in their inboxes as “spam” or “not spam”, laboriously dragging junk emails into dedicated folders. Spam filters used this information to break down each message into features that identify an email as spam: individual words or phrases, the time of day…

“I’M NOT afraid to say that I’m afraid,” said Garry Kasparov, the world chess champion, after the fifth game of his six-game 1997 series against Deep Blue 2. The match between man and machine was all square at 2 ½ points apiece as Kasparov spoke to the chess aficionados and world press assembled at the Equitable Center on New York’s Seventh Avenue. The statement raised to an extraordinary pitch the tension surrounding the final game the next day. Kasparov was fiery, flamboyant and a merciless chess competitor, and people loved him for it. When he had first taken on Deep Blue, his current opponent’s first iteration, in a six-game match the year before, he had an historically unprecedented rating. After a wobble losing the first game, to few people’s surprise…

LIKE other human champions facing a machine opponent, Grzegorz “MaNa” Komincz rated his chances. “A realistic goal would be 4-1 in my favour,” he told an interviewer before the match. One of the world’s best players of the video game StarCraft II, Komincz was at the height of a successful esports career. Artificial intelligence company DeepMind invited him to face its latest AI, a StarCraft II-playing bot called AlphaStar, in December 2018. Komincz was expected to be a tough opponent. He wasn’t. After being thrashed 5-0, he was less cocky. “I wasn’t expecting the AI to be that good,” he said. “I felt like I was learning something.” Many others at the receiving end of an AI thrashing say the same – and used the experience to get even better.…

ALPHAFOLD is an AI from Google-owned DeepMind, the same stable as game-playing AIs such as AlphaGo. It has a far loftier aim, however: to apply the same deep-learning smarts to working out the intricate structures of proteins. A better understanding of how proteins work would help us control everything from disease to food production. But a protein’s function is determined by its unique, complex tangled structure, formed by the intricate folding together of its constituent amino acids. To untangle things, researchers currently rely on laborious, expensive structure-determination methods that only work for just a few proteins. Starting from genetic information about the constituent amino acids, AlphaFold learns to make predictions about what parts of the different amino acids are in contact with one another. And it works. In a head…

PADDING through a deserted office in Waltham, Massachusetts, a quadrupedal creature stops in front of a pair of heavy doors. Resting on its haunches for a moment, as though contemplating the obstacle, it turns and seems to summon a friend. The two could almost be Dalmatians. Until, that is, one of them turns what appears to be its head into an articulated arm that grabs the handle, twists it and pulls open the door. It makes for a weirdly enthralling scene. So enthralling, in fact, that millions of people have watched the YouTube video of these SpotMini robots performing tricks such as opening inward-swinging doors, getting back up after slipping on a banana skin, doing the Running Man dance and negotiating narrow staircases. For their humanoid companion Atlas, the highlights…

IN DECEMBER 2017, Ed Bridges was mingling with the crowds of Christmas shoppers on the streets of Cardiff, UK, when the police snapped a picture of him. He hadn’t been convicted of a crime, nor was he suspected of committing one. He was simply one of a vast number of people who have been quietly added to face-recognition databases without their consent, and most often, without their knowledge. Bridges took the police force, South Wales Police, to court to get them to delete the images. Eventually, in September 2019, a court ruled that the police’s actions were lawful. The next month, however, the country’s official in charge of data privacy, information commissioner Elizabeth Denham, flagged serious concerns about the lack of laws regulating the use of face-recognition technology. Face recognition…

ALGORITHMS are not intrinsically mysterious things. They are simply sets of instructions that tell a computer how to perform a task. The trouble starts with the way the details of many in use today are proprietary – whether it’s Google or Facebook’s all-pervasive examples, or one determining your next mortgage or credit card application. The companies behind them want, of course, to protect their intellectual property and so continue to make money. Governments around the world are increasingly on the tech companies’ case. The European Union fired the first salvo with its General Data Protection Regulations, or GDPR, which came into force in 2018. Besides giving the public beefed-up rights of consent for how companies use data, it also introduces a right to be informed when automated decisions are made…

Could superintelligences of our creation could in some way turn against us or supersede us on Earth? That idea is a staple both of science fiction and discussions surrounding the future of AI, and it makes a natural finishing point for our discussions. But while machines may well in the long run outsmart us, there are any number of reasons to believe they’ll never usurp us, argues Toby Walsh PROFILE TOBY WALSH Toby Walsh is professor of artificial intelligence at the University of New South Wales in Sydney, Australia, and author of books including Machines that Think: The Future of Artificial Intelligence and 2062: The World that AI Made HOWEVER you look at it, the future appears bleak. The world is under immense stress environmentally, economically and politically. It’s hard…

IT WAS in 1936, at the age of just 24, that Alan Turing laid the theoretical groundwork for the computing revolution. Up until that point, the word “computer” meant a person who did calculations either manually or with the help of a mechanical adding machine. These human computers were an essential part of the industrial revolution and performed often repetitive calculations, such as those necessary for the creation of books containing vast tables of logarithms. Turing provided a recipe for how a machine could take over these sorts of tasks, by being provided with a set of internal rules and programmed with step-by-step procedures, or algorithms, to complete them. He also showed what such a computer could and could not be expected to do. While the idea of this “Turing…

The key is pattern recognition, a fundamental part of intelligent behaviour, for machines as well as humans. That leads to the paradigm at the heart of most modern AI: deep learning, a machine-learning technique that, implemented on architectures called neural networks, attempts to mimic how the human brain works. THE need to generalise and find patterns in data poses a whole host of new challenges for those developing machine-learning systems. For a start, how do you choose the right features that your system should light upon to make decisions? This is one of the most critical problems in machine learning. Take the example of a book-recommendation algorithm. What does it mean to say that two items are “similar”, or both relevant to a given user? We could describe a book…

STIFF neck, headache, tingling in your fingers. You list your symptoms, answer a few questions about how long they’ve lasted and whether they seem to be getting worse. Then, without ever leaving home or queueing at the clinic, you get the diagnosis: a strained neck. Or, at least, eight out of 10 people with those symptoms have one. This isn’t some future scenario. The app Ada, which offered up this diagnosis, boasts it has completed 15 million medical assessments worldwide since it was launched in 2016. It took six years for 100 data scientists to train the artificial intelligence behind Ada, using real medical records. With each case it sees, it gets a bit smarter. Health insurers and healthcare providers see a bright future for AI-based medical apps. “A machine…

ON THE night of 3 September 2010, 33-year-old Brian Wood was driving along a highway in Washington state in the northwestern US. Asleep in the passenger seat was his wife Erin, seven months pregnant with their first child. The couple were on their way from Vancouver, Canada, to spend time at her parents’ vacation home by the picturesque Puget Sound. Out of nowhere, a Chevy Blazer came hurtling towards them. By the time Wood saw it, it was too late. He braked hard and swerved right to take the brunt of the impact. He died instantly, but his wife and their unborn daughter survived. We hope it never happens to us, but any driver might find themselves making such a split-second, life-and-death decision. They are part rational, part reflex, and…

One harmless early craze was to paste the actor Nicolas Cage into films in which he never appeared. But then there were the faces of female celebrities pasted over those of people in pornographic videos, and the researchers who used AI to generate convincing footage of world leaders saying words that never actually left their mouths. Many people are worried about the potential harm that such manipulated pieces of video footage – “deepfakes” – could cause. “I’m usually a tech optimist but this is one where the red flags went up,” says Kate Devlin at Goldsmiths, University of London. “I think it could be really insidious.” New laws are an option, but they could backfire if they end up outlawing beneficial applications: deeply immersive films or video games, novel forms…