Technology

Dwarkesh Podcast - Page 3

Page 3 of 3

50 episodes 100 views

Copy link

Some thoughts on the Sutton interview

Why pre-trained language models may be a stepping stone rather than a dead end, and what continual learning would take.

11:39 9 min read 5 views

Richard Sutton – Father of RL thinks LLMs are a dead end

A pioneer of reinforcement learning argues that systems which only predict text cannot learn from the world the way animals do.

01:06:21 53 min read 1 views

Fully autonomous robots are much closer than you think – Sergey Levine

Robots learning from real-world work could kick off a self-improving loop within years, if data and hardware scale.

01:28:28 80 min read 6 views

How Hitler almost starved Britain – Sarah Paine

Britain's wartime victory over Nazi Germany hinged on winning the Atlantic: convoys, codebreaking and new ships

01:35:17 68 min read 6 views

Evolution designed us to die fast; we can change that — Jacob Kimmel

A handful of transcription factors might reset an aged cell's epigenome, but finding the right mix is the hard part.

01:44:39 99 min read 1 views

China is killing the US on energy. Does that mean they’ll win AGI? — Casey Handmer

A 5-gigawatt AI data center could need roughly 50,000 acres of solar and batteries, and the math is surprisingly concrete.

01:08:21 67 min read 1 views

Artificial meat is harder than artificial intelligence — Lewis Bollard

Cheap factory-farmed chicken left welfare out of the equation, but small sums for corporate campaigns may spare billions.

01:08:04 59 min read 2 views

Sarah Paine — How Imperial Japan defeated Tsarist Russia & Qing China

A closing railway window and a surprise attack on Port Arthur: how Japan's 1904 gamble foreshadowed Pearl Harbor.

01:55:36 85 min read 1 views

Stephen Kotkin — How Stalin became the most powerful dictator in history

Why failed constitutional reform and peasant land hunger helped pave the way for Stalin's dictatorship.

02:12:40 89 min read 1 views

Why I don’t think AGI is right around the corner

Why today's LLMs can't learn on the job like employees do, and what that means for AI handling tax filing.

14:00 12 min read 1 views