Yann LeCun, a leading AI expert and chief AI scientist at Meta, recently dismissed concerns about artificial intelligence posing a threat to humanity, labeling such fears as 'complete B.S.'.

In a recent interview, LeCun emphasized that current AI models are less intelligent than household pets, arguing that they merely predict upcoming words in text without genuine understanding.

He explained that large language models (LLMs) like ChatGPT can manipulate language but do not possess true intelligence, stating they lack essential qualities such as persistent memory and reasoning.

LeCun further elaborated that these models are not capable of understanding the physical world, which limits their functionality.

While acknowledging the potential for artificial general intelligence (AGI), he argued that today's LLMs will not lead to AGI and that achieving it is a long-term goal that could take decades.

LeCun criticized predictions from figures like OpenAI's Sam Altman, who suggested AGI could be developed within 'a few thousand days,' asserting that such timelines are unrealistic.

He contrasted his views with those of industry leaders like Sam Altman and Elon Musk, who believe AGI could emerge in the near future.

LeCun emphasized that discussions about AGI exceeding human intelligence are premature, advocating for fundamental advancements in AI before such concerns arise.

In the regulatory landscape, California Governor Gavin Newsom recently vetoed an AI safety bill, citing concerns that it would provide a false sense of security by only targeting larger companies.

He is particularly interested in research that mimics how baby animals learn, highlighting ongoing projects at Meta that analyze real-world video.

Yoshua Bengio, another AI pioneer, shares LeCun's skepticism about current AI capabilities but advocates for government involvement to ensure public safety in AI development.

Bengio has criticized the AI startup community for overhyping current capabilities and making unrealistic predictions about the future of AI.