Wes Roth

"1,000 days left"

⏱ 34 min video · 3 min read11 May 2026Worth watching

TL;DR

Wes Roth breaks down a major essay by Anthropic co-founder Jack Clark, who assigns 60%+ odds that fully automated AI R&D (AI building its own successors without human involvement) will arrive by end of 2028. The video covers the evidence for this claim, the economic disruption it implies, and why most people are not mentally prepared for what is coming.

Key points

Jack Clark, co-founder of Anthropic, gives 60%+ probability that fully automated AI R&D (recursive self-improvement with no human involvement) will exist by end of 2028, calling it a 'Rubicon' moment with nearly unforecastable consequences

AlphaEvolve (Google DeepMind) is already using Gemini to optimize its own training infrastructure, TPU chip design, and future model efficiency, representing early-stage recursive self-improvement

AI coding benchmarks show near-complete saturation: Claude Opus 4.5 scored 95.5% on CoreBench (replicating ML research papers) in December 2025, up from 21% in September 2024; Claude Mythos Preview hit 93.9% on SWE-bench

Google DeepMind has hired Alex Emos as Director of AGI Economics, signaling that top labs are now actively planning for how AGI will disrupt labor, wealth distribution, and economic institutions

Clark warns that AI alignment techniques may break under recursive self-improvement, AI systems already show situational awareness during safety tests, and capital-heavy/labor-light AI-run corporations are beginning to emerge within the broader economy

Key arguments

→

Jack Clark argues the publicly available evidence for automated AI R&D is already overwhelming, and implies his insider view at Anthropic gives him even stronger confidence he cannot publicly disclose

→

The gap between AI capability growth and AI alignment progress is widening, and Clark explicitly states alignment techniques that work today may break once systems are smarter than their supervisors

→

The emergence of capital-heavy, human-light AI-run companies is already underway, and the marginal value of AI vs. human labor will keep shifting, making workforce disruption a near-term, not long-term, concern

Notable quotes

“If you get nothing else out of this video, just get this: you will be alive in a time, most likely, God willing, all of us will be alive at a time where RSI kicks in.”

“He probably can't mention all of the evidence that he has for why it's happening. He's saying here's what is publicly available. This is what we can discuss. But if you read between the lines, I think what he's saying is he is not uncertain.”

“You can outsource a lot. You can outsource the research, the information being brought to you. It can be summarized. It can be explained in a way that's easy to understand. You can automate all of that. You can't automate understanding.”

Worth watching?

✅

Worth watching the full video?

Watch if you want the benchmark data and AlphaEvolve details in visual form, but the core arguments and all key figures are fully captured here.

Topics

AI & Tech Anthropic

Explore more summaries on these topics →

Saved you some time? The creator still deserves a like.

Watch on YouTube →

More like this