50 - Eli Lifland on AI 2027

Remember AI 2027? Not AI 2040, the newest coolest thing AI Futures Project has done, but AI 2027, their OG product? At long last, we have an AXRP episode about it. Enjoy!

Transcript: https://axrp.net/episode/2026/08/03/episode-50-eli-lifland-ai-2027.html

Topics we discuss, and timestamps:

0:00:12 What is AI 2027?

0:10:17 What happens in AI 2027?

0:18:08 Why two endings?

0:21:15 Who did what?

0:23:59 Why superhuman AI in 2027?

0:32:53 Forecasting time horizon growth

0:48:34 When do time horizons go infinite?

1:02:13 Forecasting effective compute growth

1:07:09 From superhuman coders to superintelligence

1:20:58 How many AI companies?

1:26:29 What AGI will want

1:39:48 What misaligned AI does

1:52:20 Will AIs be able to align their successors

1:57:21 Why so long until AI takeover?

2:03:06 Would misaligned AGI kill us?

2:04:53 Will there just be one AGI?

2:13:25 The reception of AI 2027

2:18:57 What do you now think about takeoff?

2:26:31 What's next for AI Futures Project

2:31:36 How to work on AI forecasting

2:38:01 Following Eli's and AI Futures Project's work

Links to AI 2027 and related research:

AI 2027: https://ai-2027.com/

AI Futures Project blog: https://blog.ai-futures.org

AI Futures Research Notes: https://aifuturesnotes.substack.com/

AI Futures Model: https://www.aifuturesmodel.com/

X/Twitter links:

Eli Lifland: https://x.com/eli_lifland

Daniel Kokotajlo: https://x.com/dkokotajlo

AI Futures Project: https://x.com/AI_futures_

Research we discuss:

Task-Completion Time Horizons of Frontier AI Models: https://metr.org/time-horizons/

What Happens When Superhuman AIs Compete for Control?: https://blog.ai-futures.org/p/what-happens-when-superhuman-ais

How AI Takeover Might Happen In 2 Years: https://www.alignmentforum.org/posts/KFJ2LFogYqzfGB3uX/how-ai-takeover-might-happen-in-2-years

Guive Assadi on AI Property Rights (AXRP): https://axrp.net/episode/2026/02/15/episode-48-guive-assadi-ai-property-rights.html

Titotal: A deep critique of AI 2027's bad timeline models: https://titotal.substack.com/p/a-deep-critique-of-ai-2027s-bad-timeline

Response to titotal's critique of our AI 2027 timelines model: https://www.lesswrong.com/posts/G7MmNkYADKkmCiumj/response-to-titotal-s-critique-of-our-ai-2027-timelines

What you can do about AI 2027: https://blog.aifutures.org/p/what-you-can-do-about-ai-2027

Episode art by Hamish Doodles: hamishdoodles.com