Autocracy in America The Computer Scientist
26 snips
Sep 5, 2025 
Garry Kasparov chats with Gary Marcus, a cognitive scientist and AI researcher known for his critical insights on AI hype. They explore the balance between AI's potential benefits and risks, emphasizing that AI is merely a tool in human hands. Topics include the misperception of AI as understanding complex contexts and the ethical implications of its integration into society. They stress the importance of accountability in AI usage and the need for public awareness to prevent technological authoritarianism.
AI Snips
Chapters
Books
Transcript
Episode notes
Kasparov's Human–Machine Story
- Garry Kasparov recounts beating 32 chess computers in 1985 and then facing Deep Blue in 1997, showing rapid machine progress.
- He frames that arc as emblematic of recurring public hype and fear around so-called intelligent machines.
Scale Creates Illusion, Not Understanding
- Gary Marcus explains modern large language models mainly exploit massive data to mimic text patterns rather than genuine understanding.
- He argues their apparent intelligence is an illusion produced by scale, not deep conceptual reasoning.
Verbal Rules Don't Equal Internal Models
- Marcus shows that LLMs can verbalize rules yet fail to follow them in practice, revealing lack of internal models.
- He uses chess illegal-move examples to illustrate that LLMs lack dynamic, abstract representations of the world.
