AI’s Leading Critic: Gary Marcus on the Risks, Myths, and Failures of AI

In this episode of AI-Curious, we speak with Gary Marcus, author of the new book "Taming Silicon Valley" and one of the most respected—and controversial—voices in artificial intelligence. Known as “AI’s leading critic,” Marcus shares his candid perspectives on the limitations of large language models (LLMs), the risks of concentrated power in Big Tech, and why the current AI landscape is in need of foundational change.

Marcus explains:

•Why LLMs like ChatGPT lack true reasoning capabilities and are prone to hallucinations.

•His argument that LLMs are hitting a developmental plateau, despite billions of dollars being invested in scaling.

•Concerns over Big Tech monopolies and the growing surveillance potential of companies like OpenAI.

•How AI risks, from misinformation to bias and environmental impact, mirror the early industrial age’s “privatized profits and socialized costs.”

We discuss his latest book, Taming Silicon Valley, which critiques the unchecked influence of tech giants and calls for grassroots action to demand better AI regulation and fairness. From online skirmishes with Sam Altman and Elon Musk to his thoughts on artificial general intelligence (AGI), Marcus offers a refreshingly candid take on the urgent challenges facing the AI industry.

Timestamps:

•0:00 Introduction: Who is Gary Marcus?

•3:15 Gary’s early work on neural networks and reasoning.

•7:30 The plateau of large language models: why scaling isn’t enough.

•15:20 Surveillance, Big Tech, and the dangers of AI oligarchies.

•21:45 Hallucinations and why generative AI isn’t all it’s cracked up to be.

•28:00 Grassroots action: how the public can push for better AI.

•34:10 Gary’s hopes for a proper debate with AI’s key players.

•39:20 Final thoughts: The future of AI and AGI.

Taming Silicon Valley: How We Can Ensure That AI Works for Us

The Gary Marcus substack
https://garymarcus.substack.com

Gary Marcus
http://garymarcus.com/index.html