His name may mean nothing to you, but Jeffrey Hinton is at the center of the current revolution, the AI revolution. Now 75 years old, this cognitive psychologist and computer scientist is dubbed the ‘Godfather of Artificial Intelligence’ and this British Canadian just announced his resignation from Google in a statement to The New York Times.
If he recognizes that his age is one of the justifications for this withdrawal, he also makes it clear that he is afraid of malicious uses of AI tools, such as Chat GPT that he helped create through his work. He told the New York Daily that until last year he thought Google was a “good manager” of technology, but that changed when Microsoft began integrating a chatbot into its Bing search engine. “I have come to the conclusion that the kind of intelligence we develop is very different from the intelligence we have.” Explain this world to BBC. We are biological systems and they are digital systems. The big difference is that with digital systems you have many copies of the same set of weights, same world pattern. »
Soon smarter than man
Copies capable of learning on their own and feeding on billions of data: “All of these clones can learn separately, but share their knowledge instantly. It’s like having 10,000 people and every time someone learns something, everyone knows it automatically. This is how these chatbots can know so much more than any other. another person “.
For this world, human beings will quickly fall behind the strength and development of artificial intelligence! “At the moment they are not smarter than us, as far as I know. But I think they will soon be.” Expect. In the short term, he worries, people will no longer be able to tell right from wrong with images, videos and text generated by AI tools. Unfortunately, this is actually the case.
Fear that artificial intelligence is at the service of dictators
What still allows humans to have superintelligence today is their ability to think, as artificial intelligence is still only in its infancy. “So far what we’re seeing is things like Chat GPT-4 outperforming and outperforming someone in terms of general knowledge. In terms of reasoning, it’s not very good, but it’s really capable of performing simple reasoning. Given the pace of progress, we expect things to get better. Quickly. So we have to worry about that.”
In the medium term, it even evokes ” nightmare scenario Where AI will be at the service of tyrants: One can imagine, for example, that a bad protagonist loves [le président russe Vladimir] Putin decides to give robots the ability to create their own subgoals like “I need more power”.
