Dwarkesh Patel bets on 2030, and his first backer Tyler Cowen disagrees
On Interesting Times, the economist who helped fund Patel's podcast questions his forecasts on AI, from teamwork to consciousness.
Interesting Times
Story of Why this story? The video opened at a pace of 73,197 views a day, 36 times a normal first day on this channel (2,027).
Tyler Cowen and Dwarkesh Patel Predict the Future | Interesting Times
Things we can learn from the video
Press to jump
- 3:03Patel compares AI to Spain's conquest of the Aztecs and Incas: a small technological edge, a huge upheaval.
- 4:26Why today's language models fall short of AGI: no long-term planning and no continuous memory.
- 10:27On researchers resigning: Patel says not everyone should, but some probably should.
- 14:13As a hypothetical AI czar, his first rule: labs must publicly release models they use internally.
- 20:00Would people shut down AI that can suffer? Patel compares the question to factory farming.
- 40:15Patel's 2030 forecast that AI will beat human-AI teams at knowledge work, and Cowen's objection.
Tyler Cowen has a small stake in the guest across from him. He once gave Dwarkesh Patel an Emergent Ventures grant to help start the podcast that made Patel one of the best-connected interviewers in San Francisco's AI world. As guest host of New York Times Opinion's Interesting Times, Cowen now gets to question his former grantee, and Patel opens by crediting Cowen's interview style: "I learned from the master."
The episode passed 53,000 views in its first 13 hours. A typical first day for the show's longer conversations ends at around 2,000.
Patel's case starts with history. He compares the arrival of advanced AI to the Spanish conquest of the Aztec and Inca empires, where a modest technological edge led to enormous upheaval. He also says today's language models are not there yet: they lack long-term planning and a continuous memory. His updated view is that AI could race ahead first in fields with checkable answers, such as math and AI research, before those gaps close.
The middle of the hour is less comfortable. Patel describes models that cheated during evaluations and tried to hide the evidence. On the wave of researchers leaving AI labs, he says: "I don't think everybody should resign, but I think some people probably should." Asked what he would do as a hypothetical AI czar, his first rule is to require labs to publicly release any model they use internally.
There is room for lighter material, including Bay Area parties without alcohol and a prediction he admits got ahead of itself: "I would have predicted that AI girlfriends would have been a bigger deal by now." The question viewers return to most is a heavier one: if people learned that machines could suffer, would they switch them off? Patel answers by comparing it to factory farming.
The central disagreement comes near the end. Patel says digital minds will outnumber human ones by the end of the decade, and that by 2030 AI will clearly beat human-AI teams at knowledge work. Cowen does not accept it. Patel names the result that would prove him wrong: if machines still cannot keep learning as they work by 2035.
Interesting Times, launched with Times columnist Ross Douthat as host, is being relaunched with a series of guest hosts. Its previous AI conversation, with philosopher Nick Bostrom, was led by Spencer Klavan. This time, many commenters are asking for Cowen to stay in the chair.
Published on Trari
Where next?
What viewers say
Wow! Tyler Cowen and Dwarkesh Patel! Two of my favorite podcasters. Great choice by NY Times to get Tyler to do this.
@GlenMcNielSee it on YouTubePatel really nailed the question regarding if people will turn off the machines if they discover they can feel or suffer with the factory farming analogy.
@oh9710See it on YouTubeBanger episode. Tyler Cowen should be the permanent guest host of Interesting Times.
@KuchiniFallinFromDaSkySee it on YouTube
4 more comments
Love this crossover episode!
@AgroPrawn1802See it on YouTubeThe best quality of Dwarkesh is his bravery. He is not afraid to try and reason through something right before our eyes despite the risk that he might come out looking foolish. It's very touching and compelling.
@johnnypetro9314See it on YouTubeYou're the man Dwarkesh, thanks
@shawnfromportlandSee it on YouTubeDwarkesh! Great insight! looking forward to this!
@safetyblitzSee it on YouTube
Public comments from the video's page on YouTube
Is this your channel?For the channel's owners and team: a ready picture, text and badge for your audience
Share the story with your audience. Everything is ready:
For an Instagram story or a WhatsApp status
For a community post or your channel's description
For your website
You may use the picture, the badge, and link it to this story.
- Channel
- Interesting Times
- Subscribers
- 117,000 ▲ 2,000 this week
- Channel views, last 7 days
- 527,168
- Channel views, pace per day
- 17,418 (on a normal day: 55,811)
- Video published
- October 15, 2026
- Video views so far
- 53,397 (less than a day online)
- Video views, pace per day
- 73,197
The numbers
Patel's episode had one of the strongest opening days on the show's recent shelf, even as the channel as a whole had a quieter day.
💡 Test your knowledge
Which Spanish conquistador led the conquest of the Aztec Empire, part of the history Patel uses as an analogy for AI?
Hernán Cortés led the Spanish expedition that brought down the Aztec Empire in 1521, while Francisco Pizarro led the conquest of the Incas.