Officially retired from Google, Dr. Geoffrey Hinton now talks about his fears about artificial intelligence, which he helped bring to light during his career.
“I console myself with the classic excuse: if I hadn’t done it, someone else would have done it.” In 2004, Geoffrey Hinton focused his efforts around an idea that no one had succeeded in realizing. He developed the technological concept of the neural network, now the basis of all artificial intelligence.
This discovery earned him the Turing Prize in 2019 (similar to the Nobel Prize in Computer Science) along with two colleagues, including Yann Le Cun – now head of AI at Facebook. But this Monday, Geoffrey Hinton joined the camp of skeptics. In an interview at New York Timesthe 75-year-old Briton now shows regret about his work.
“It’s hard to see how bad actors can be prevented from using it for malicious purposes,” points out Geoffrey Hinton, often dubbed “the godfather of AI”.
Photos and texts created from scratch
These statements were made after Dr. Hinton left Google. He worked for more than ten years within the company. However, he believes he can speak freely about the risks posed by artificial intelligence.
A few weeks ago, tech figures (including Elon Musk and Steve Wozniak) signed a petition calling for a moratorium on AI, wanting to halt advances and projects for six months. A few days later, another grandstand was posted by the Association for the Development of Artificial Intelligence. Here again, it was a question of the risks carried by the technology.
If Geoffrey Hinton did not sign either of the two letters, he remains worried. One of his main concerns are photos, videos and texts created from scratch. According to him, this could lead to “no longer being able to know what is true”.
The economic impact of artificial intelligence is also at the heart of his fears. Moreover, a report by the American bank Goldman Sachs estimates that the technology could replace 300 million jobs. At the same time, Chris Pissarides – Nobel Prize in Economics in 2010 – believes that models such as ChatGPT could help implement the four-day week.
But Geoffrey Hinton prefers to remain cautious. “It removes the tedious tasks. It may well remove more than that,” he concludes.
