Finding an AI that solves all problems is “nonsense”, says researcher

Jordan, professor of computer science at the University of California, Berkeley, replied thus the statements of industry figures such as Sam Altman, head of OpenAi, or Dario Amodei, of Anthropic, who consider that the so -called artificial general intelligence (AGI) is Just a few years away.

The AGI is the coveted “Holy Grail” of the investigation in AI that would supposedly overcome humans in all fields.

“This Super Agi who can answer any questions, knows everything … this is nonsense,” said Jordan, an expert in automatic learning and statistics, at a conference in AI in the prestigious Polytechnic Engineering School of France.

Jordan criticized the fact that companies are spending “billions” to achieve what he said was, for now, an illusory objective.

Today’s models can “really predict, very good” thanks to access to vast amounts of data generated by humans, including those of the Internet, Jordan said.

But the AI ​​“cannot know everything, because what I am thinking is currently the context of what I am going to do next, today. He can’t know that, ”he added.

“And if you don’t know, you can’t give me good advice,” he settled.

The great language models (LLM) that drive chatbots of AI as chatgpt are like “a lot of small experts” capable of answering highly specific questions but “are not intelligent entities.”

Jordan characterized current developments in AI as “a new emerging engineering field, such as civil engineering,” but working with “human decisions, data and networks in networks.”

That makes it appropriate to solve specific complex problems such as improving transport or medical care, he suggested.

“The development of this wonderful engineering field is really braking for all this aspiration of the AI ​​of building a super robot that is responsible and helps you with all your homework and all these other things,” said Jordan.

“I don’t think there has ever an era in the history of humanity in which a new technology field has emerged with so much hype and hysterism,” he added.

The debate about whether the AGI is realistic, and if it can be a threat to humanity if its creators lose control, it is one of the main controversies between scientists and developers.