Neuroscience Expert Gary Marcus Speaks at US Senate on AI

Marcus has been recognized in academia for some time, but his prominence has recently increased due to his newsletter (“The Road to AI We Can Trust”), a podcast (“Humans vs. Machines”), and his concerns about the uncontrolled rise of AI. He has also been featured in media outlets such as Bloomberg Television, The New York Times Sunday Magazine, and Wired.

Risks of Current AI

When asked about the risks of current AI, Marcus stated that he does not consider current AI to be an existential threat, but it is still dangerous. According to him, current AI poses a danger to democracy, although it does not threaten humanity as a whole. He also highlighted potential dangers to democracy, such as deliberate or unintentional misinformation and manipulation of individuals’ political beliefs.

Read also: ChatGPT: A Powerful Tool with Debates

Debate with Yann LeCun

Marcus also mentioned his debates with Yann LeCun, the AI Chief Scientist at Meta. Their disagreement mainly concerns the use of large-scale language models and the potential risks they pose. Marcus believes that LeCun is wrong to think there is no danger in using these models, as they can be used to manipulate individuals and spread misinformation on a large scale.

Regulation and Openness Questions

During the hearing, few questions were asked about open or closed systems in AI. Marcus believes it is important to allow some level of open source, but there also need to be limits on what can be achieved and how it can be implemented. He also expressed concerns about Meta’s strategy of making its language model available to the public for modification, calling this decision “reckless”.

An AI Regulatory Agency

Marcus advocates for the creation of an AI regulatory agency, similar to the FDA for drugs, which would oversee the large-scale deployment of AI technologies. He believes this agency should be global, neutral, and non-profit. Marcus is willing to play a role in this agency and share his neutral voice to help strike a balance between the benefits and risks of AI.

Reflections on the Senate Hearing

When asked about his experience at the US Senate, Marcus said he was deeply moved by the event and felt that everyone was there to give their best for the United States and humanity. He would not be surprised to be invited to testify again, but he does not yet know if that will happen.