Dr. Eli David, Top AI Researcher and Entrepreneur, on Free Speech, Woke AI, and Why We’re so Much Smarter Than Elephants

Mar 30, 2026

Dr. Eli David, AI researcher and serial entrepreneur known for neural-network breakthroughs and outspoken views. He discusses viral X fame, why current LLMs are inefficient, alternatives like sparse architectures, the energy-driven US–China AI race, and the tension between free speech and platform moderation. Short, punchy, and provocative takes throughout.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Why LLMs Flatten Humor And Edge

Eli says statistical LLM outputs flatten human emotion and dark humor because models learn average public tone from filtered data.
He A/B tested GPT-4 drafts versus his own tweets and found AI lacked the cynical punch that works on his audience.

INSIGHT

RLHF Imposes Trainers' Values On LLMs

Eli argues RLHF (reinforcement learning from human feedback) intentionally flattens models by teaching them to avoid 'not good' outputs, importing trainers' worldviews.
He cites GPT-3's nickname 'woke GPT' where RLHF prevented producing pro-Trump content while allowing pro-Biden output.

ADVICE

Handle Content Disputes By Building Alternatives

Draw the free-speech line at legal versus illegal content and avoid platform-level moral policing.
If you dislike a platform's content policies, Eli's practical remedy is to build or join a different network instead of universal censorship.

Get the Snipd Podcast app to discover more snips from this episode

Get the app