[ad_1]
Yann LeCun, the chief AI scientist at Meta, mentioned that computer systems are removed from attaining human-level intelligence and that fears of AI taking on the world are primarily based on a misunderstanding of the know-how. LeCun, one of many main specialists in synthetic intelligence, has dismissed the concept that AI might pose a risk to humanity as “preposterously ridiculous.”
He spoke at a press occasion in Paris on Tuesday, the place he defined the present state and limitations of AI methods that may generate pure language however lack widespread sense and reasoning talents. LeCun mentioned that computer systems would surpass human intelligence sometime, however that was far sooner or later, and we will keep away from constructing it if it’s unsafe.
He dismissed the concept that AI would dominate the world as a mirrored image of human tendencies towards machines. He additionally argued that proscribing AI analysis could be a grave error.
“It’s as if you happen to requested in 1930 to somebody how are you going to make a turbo-jet protected? Turbo-jets weren’t invented but in 1930, similar as human degree AI has not been invented but. Turbo jets had been finally made extremely dependable and protected,” LeCun mentioned, highlighting that the identical would occur with AI.
Throughout the dialogue, LeCun said that machines surpassing human intelligence sooner or later shouldn’t be perceived as a risk. He famous the advantages of this development, envisioning AI assistants which might be smarter than people, appearing as supportive instruments in day by day life.
LeCun emphasised the significance of making certain that these AI methods stay beneath human management and subservient to them. Moreover, he dismissed the notion popularized by science fiction that robots would harbor intentions of world domination, highlighting the absence of a connection between intelligence and a need to take over.
In different information, Meta introduced granting researchers entry to sure parts of a novel AI mannequin referred to as I-JEPA. In line with the corporate, this mannequin can analyze and successfully full unfinished photographs, surpassing the accuracy of present fashions. Not like standard generative AI fashions that solely depend on neighboring pixels, I-JEPA leverages background information concerning the world to fill within the lacking components of photographs.
Learn extra:
[ad_2]
Source link
