Artificial intelligence itself shouldn’t be targeted by regulators, but how it is used, the senior Facebook official who develops the technology told AFP.
“I am generally in favor of regulating a particular application rather than a technology” in general, said Yann LeCun in an interview.
Facebook’s chief AI scientist and one of the company’s vice presidents, LeCun is one of the computer scientists behind modern artificial intelligence.
He’s skeptical of the recently proposed EU regulations, but less scathing than former Google chief Eric Schmidt, who called them a disaster and said the technology was too young to be regulated.
While he believes the proposed EU regulations were drafted “with good intentions,” LeCun fears they will hold the EU back.
“AI systems need to be secure, that they don’t endanger people, that they respect privacy,” he said.
“But we have to be careful not to slow down the research and the creativity of researchers, who are kind of the engine of innovation in the economy,” said LeCun.
“Europe runs the risk of falling behind” if it adopts overly restrictive AI regulations, he warned.
– Facial recognition –
Speaking of facial recognition, for example, he said regulators need to distinguish between apps that serve “good purposes” and others.
Last week, EU data protection agencies called for an outright ban on the use of AI to identify people in public places, underlining the “extremely high” risks to privacy.
“Facial recognition or biometric recognition is absolutely essential for some countries that do not have simple means to verify the identity” of people, said LeCun.
“It could help people get access to bank accounts, to social services,” or as the Bill Gates Foundation has done, which uses AI to compare images of feet to make sure babies don’t not receive two vaccines, he said.
“On the other hand, we need very strict regulations to protect privacy, to prevent people passing through public spaces from having their faces identified,” he added.
– Self-supervised learning –
LeCun was one of the guests at an event marking the sixth anniversary of Facebook’s AI research lab in Paris.
Today, around 90 people work at Fair Paris and the laboratory has published work recognized by the international scientific community.
Facebook has used its work to improve its text and image recognition tools to better moderate content on its social networks.
LeCun is now taking a keen interest in researching what is known as self-supervised learning.
Machine learning has so far used data sets prepared by humans, but in self-supervised learning, computers try to learn a lot on their own.
“In the near future, or beyond, depending on the progress we make, we may have intelligent personal assistants who understand almost everything we say and are not frustrating to interact with them,” he said. declared.
“We could also develop somewhat intelligent and autonomous robots: machines that we give a task to but don’t explain how to do it,” said LeCun.
lby / rl / spm