High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / belief

Published · transcript-backed

Leopold Aschenbrenner: belief

4 Jun 2024 Dwarkesh Podcast Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history

“We talked about the unhobbling. I think of a GPN token as one token of my internal monologue.”

— Leopold Aschenbrenner

Source trail

Everything needed to verify it.

Speaker
Leopold Aschenbrenner
Attribution
Verified speaker
Claim type
belief
Recorded
4 Jun 2024
Publisher
Dwarkesh Podcast

Transcript context

…Here’s an analogy. Today, we generate more patents in a year than during the actual physics revolution in the early 20th century. Are we making more physics progress in a year today than we did in half a century back then? Generating all these tokens doesn't necessarily equate to generating as much codified knowledge in the initial creation of the Internet. Internet tokens are usually final output. We talked about the unhobbling. I think of a GPN token as one token of my internal monologue. That’s how I do this math on human equivalents. It's like 100 tokens a minute and then humans working for X hours. What is the equivalent there? This goes back to something from earlier. Why haven’t we seen huge revenues from AI yet? People often ask this question. If you took GPT-4 back ten years, people would think it would automate half the jobs. There’s a modus ponens, modus tollens here. Part of the explanation is that we’re on the verge and we just need to do these unhobblings. Part of that is probably true. But there is another lesson to learn there. Just looking at a set of abilities at face value, there are likely more hobblings behind the scenes. The same will be true of AGIs running as AI researchers.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence