After five false starts, AI is now the real deal
Artificial intelligence is more powerful than ever.
We are living through a technological revolution. Over the past 18 months, generative artificial intelligence (AI) has emerged as the technology of the moment, with a new story published by the world’s media every 24 seconds in 2024, according to my analysis of Nexis, a media database.
Even if ChatGPT isn’t going to replace your job just yet, watch out – artificial intelligence is set to affect 40 per cent of jobs worldwide, according to the International Monetary Fund.
Excitement over AI is nothing new. In my new book,
How AI Ate the World, I write about five occasions that AI has previously been heralded as a world-changing technology.
1 ALAN TURING, 1950
The brilliant mathematician had long thought that the human brain was, in some ways, a biological version of a digital computing machine. Turing (inset) developed on paper an AI rule-following computer program that could play chess, which he called Turochamp. As technology developed, he thought it was important to develop a test that could be used to distinguish between computers and humans. His eponymous 1950 test would become the benchmark against which AI was measured for decades.
He made two bold claims: that computers would eventually have about 100 megabytes of memory, and that this would help machines pass the test.
The first part was correct: an Apple PowerBook G4 released around the millennium had 128 megabytes of RAM as standard. The second part wasn’t.
2 REAL-TIME TRANSLATION, 1954
In January 1954, the world’s media gathered at IBM’s building at 590 Madison Avenue in New York to witness the future.
Using an IBM 701 computer, researchers at the computer company and Georgetown
University demonstrated an
AI tool that could translate 250 words from Russian into English near-instantaneously.
The New York Times quoted Georgetown researcher Léon Dostert as saying: “Five, perhaps three, years hence, interlingual meaning conversion by electronic process in important functional areas of several languages may well be an accomplished fact.”
As anyone who’s received a wonky translation from Google Translate knows, that isn’t quite true – even today.
3 THE PERCEPTRON, 1958
In the 1950s, a psychologist named Frank Rosenblatt received money from United States Office of Naval Research to develop his AI tech, which he called the “perceptron”. The navy wanted it to help distinguish between harmless fish and harmful Soviet subs.
The New York Times claimed that “when completed in about a year, [it was] expected to be the first non-living mechanism able to ‘perceive, recognise and identify its surroundings without human training or control’”.
The perceptron never worked. But the principles behind it would become popular decades later.
4 THE ALVEY PROGRAMME, 1984
AI went in and out of vogue with Western researchers in the following decades.
It took geopolitics to bring it back in from the cold. Japan had pressed ahead undeterred with its own AI research, spending hundreds of millions of pounds on development.
In 1983, the US launched a 10-year plan to build AI systems, and investment in AI tripled between 1984 and 1988. In 1984, the UK began the Alvey Programme, its response to the AI revival, which would cost £350m. It wasn’t money well-spent – AI took another 40 years to find its purpose.
5 DEEP BLUE BEATS GARRY KASPAROV, 1997
More than half a century after Alan Turing first dreamt up the idea of an AI chess expert, IBM’s Deep Blue system managed to defeat grandmaster Garry Kasparov in a rematch held in New York City.
Kasparov had beaten the supercomputer a year earlier in Philadelphia. But this time round he was spooked when, in the 44th move of the first game, Deep Blue made a deeply illogical action, as a result of a coding error.
But Kasparov believed that Deep Blue had used brute force computing power to see a future opening the chess genius couldn’t.
It rattled him. For the rest of the series, Kasparov thought Deep Blue could see into his soul. He ended up losing. Machine had triumphed over man, yet it took another two decades for AI to become big.
WHY THIS TIME IS DIFFERENT
Not only have computers become much more powerful, but the way we code AI systems and the way they work has, too. A revelation in AI research in 2017 helped kickstart the technology underpinning generative AI tools.
Today, we’re more realistic because the tools are in our hands to see both their benefits – and limitations. We’re more prepared. And we need to be more cautious.