I t's the summer of 1956, and 20 young scientists meet for a workshop at Dartmouth College, in New Hampshire. They have an ambitious goal: to build a machine that simulates human intelligence. It is the birth of a field of research that, today, is being talked about more than almost any other topic: artificial intelligence (AI).
Decades pass, but progress is slow. Nearly 70 years later, the dream is still far away, even though programs like ChatGPT are impressive. AI can write poetry, defeat chess grandmasters and compose music, but its abilities are always limited to a specialized field. So far, it hasn't come close to human intelligence, but the goal is the same: to develop artificial general intelligence (AGI), a computer program that can do everything humans can do.
How would AI change our lives?
It would certainly change the world of work. In a recent study, the investment bank Goldman Sachs predicts that AI could automate 300 million jobs worldwide. Just a few years ago, many experts believed the technology would replace taxi drivers and factory workers. Now, it's become clear that the latest AI models, known as foundation models, mainly threaten white-collar jobs, such as those of lawyers and financial analysts, screenwriters and journalists.
Does it truly threaten those jobs? ChatGPT and similar programs are very good at doing tasks that aren't very exciting: composing emails, searching through documents, writing computer code. They are useful helpers, not all-rounders. The creativity of an author and the advice of a lawyer will still be needed in the future. What's more, according to the Goldman Sachs study, AI could lead to enormous productivity gains. Global gross domestic product could grow by seven per cent within ten years.
Nevertheless, politicians should already be thinking about how to adapt the welfare state to a world in which AI is even more powerful. In medicine, too, great hopes rest on AI. The technology could support doctors during operations and help in the development of new medicines. In 2022, the British company DeepMind succeeded in predicting the shape of almost all proteins in the human body — a breakthrough that could advance research into rare diseases, for example.
However, artificial general intelligence also brings dangers that worry researchers, including AI pioneer Yoshua Bengio. He doesn't believe AI will wipe out humanity. He's more concerned that people will let themselves be manipulated by it and believe the lies and fakes that AI sends out into the world.