# Dwarkesh Podcast

Dwarkesh Podcast features long conversations with researchers, builders and thinkers about technology, history and the future.

Type: show. Context selected by Escape Meridian; inclusion is not endorsement.

[Collection](/entities/dwarkesh-podcast)
[Original website](https://www.dwarkesh.com/)

## Sources

- [Dwarkesh Podcast](https://www.dwarkesh.com/): Follow extended conversations with researchers, builders and historians into specific claims and disagreements.

## Perspectives

- [Adoption can move fast and still take work](/perspectives?position=amodei-fast-diffusion) (Dario Amodei): Amodei expects AI to spread through the economy unusually quickly, while procurement, permissions and organizational change still take time. He separates models writing code from complete job automation and reduced demand for workers.
- [More agents can buy speed. The savings still need measuring.](/perspectives?position=brown-coordination-cost) (Noam Brown): Brown argues that parallel agents can reduce waiting time while increasing total compute. He expects better models to organize themselves more effectively and credits the underlying model with most of the mathematical result, while saying very large teams remain poorly compared with smaller ones.
- [Someone still has to decide what a good result means](/perspectives?position=schulman-choosing-objectives) (John Schulman): Schulman separates carrying out experiments from deciding what to seek. Useful research also involves judgment, realistic tasks and objectives that reflect what people actually want.
- [A clear target lets a model try what people have not demonstrated](/perspectives?position=millidge-learning-beyond-demonstrations) (Beren Millidge): Millidge argues that models can attempt harder goals than people can demonstrate when success is verifiable. Whether that learning transfers far enough to remove human guidance remains uncertain.
- [Yesterday's useful knowledge can become today's wrong assumption](/perspectives?position=oneill-changing-work-context) (Charlie O’Neill): O’Neill distinguishes technical advances that remain useful from workplaces whose relationships, routines and circumstances keep changing. Accumulating knowledge and keeping it relevant are different demands.

## Terminal notes

- [A faster answer can still cost more work](/notes/noam-brown-2026-09-17.md): Noam Brown describes agent teams as a way to finish sooner by spending more compute. He expects stronger models to coordinate better, while acknowledging limited measurement of very large teams.
- [When answers get cheaper, choosing the question matters more](/notes/research-panel-choosing-next-question.md): John Schulman, Beren Millidge and Charlie O’Neill largely agree that choosing objectives and the next experiment remains difficult for models. They disagree about how long that limitation lasts and what could overcome it.

## Parallels

- [More capable. More in control?](/parallels?parallel=delegated-judgment): When does handing over the work give people more room to decide?
- [A discovery is a beginning](/parallels?parallel=discovery-to-benefit): What has to happen for a technical advance to become a shared benefit?
- [The answer is right. Was the question?](/parallels?parallel=right-answer-wrong-question): When producing an answer gets cheaper, where does the useful work move?

## Context references

- [https://www.dwarkesh.com/](https://www.dwarkesh.com/)

Latest published distillations: [Publications API](/api/pipeline/publications?entity=dwarkesh-podcast)