(San Francisco) Mark Zuckerberg, the boss of Meta (Facebook, Instagram), declared that his group was now working on so-called “general” artificial intelligence (AI), that is to say computer systems equipped with capabilities human cognitive, just like OpenAI, the creator of ChatGPT.
“We came to the conclusion that, to build the products we want to build, we have to work towards general intelligence,” the billionaire said in an interview published Thursday by The Verge, an American publication specializing in technology.
“I think it’s important to get this out there because many of the best researchers want to work on more ambitious problems,” he added.
The idea of general AI was popularized by the start-up Californian OpenAI. It took the world by surprise at the end of 2022 by launching ChatGPT, the best-known generative AI software (production of various content on a simple query in everyday language).
Sam Altman, his boss, defines it as the technology that will be at work in programs “more intelligent than humans in general”.
The concept remains nebulous: Nick Clegg, head of international affairs at Meta, noted in Davos on Thursday that there is “no consensus on what exactly general AI means.”
“Ask computer scientists to define (it) and you will get a different definition from each of them,” he pointed out.
The emergence of generative AI over the past year has led to fierce competition between technology giants.
Microsoft (key investor in OpenAI), Google and Meta have deployed numerous tools and are seeking to attract the best engineers.
In society, this technology arouses a lot of enthusiasm but also an immense wave of concern on the part of observers and authorities, who fear in particular that it will lead to the loss of numerous jobs or that it will multiply the powers of ill-intentioned actors.
A fortiori, if humans manage to endow machines with an intelligence comparable to that of humans, but with much greater computing capabilities.
OpenAI says it wants to build general AI gradually, with the aim of benefiting all humanity. It relies on the large-scale use of its computer models to detect and rectify problems.
“I think one day we will create something that will be considered general AI, depending on whatever fuzzy definition you want,” Sam Altman said in Davos. “The world will have a two-week panic attack, and then everyone will go back to their normal lives.”