Gary Marcus joins Steve to discuss calls to slow down AI development, Anthropic’s IPO and mounting debt, Trump’s self-policing approach, the Hugging Face incident, rogue AI agents, anthropomorphizing AI, and whether LLMs, at their core, are incapable of ever truly aligning with human values. They also discuss whether AI is better than we are at writing literature or doing math (and if so, whether we should let it be), mathematician Tristan Buckmaster’s accusation that OpenAI stole from his unpublished work, how to regulate AI, and whether politicians who barely understand how it works are up to the task.
Gary Marcus joins Steve to discuss calls to slow down AI development, Anthropic’s IPO and mounting debt, Trump’s self-policing approach, the Hugging Face incident, rogue AI agents, anthropomorphizing AI, and whether LLMs, at their core, are incapable of ever truly aligning with human values.
They also discuss whether AI is better than we are at writing literature or doing math (and if so, whether we should let it be), mathematician Tristan Buckmaster’s accusation that OpenAI stole from his unpublished work, how to regulate AI, and whether politicians who barely understand how AI works are up to the task.