The BaseVideoDwarkesh Patel (YouTube)44 min194,214 views

Recursive self-improvement has started

In a 44-minute video essay, podcaster Dwarkesh Patel argues that AI is already speeding up AI research and gives a median guess of 2028 for AI matching the best human AI researchers. Patel works through objections and weighs two outcomes after that point.

Open the source ↗4 quotations checked, word for word ↓
About Anthropic or ClaudeClaude, made by Anthropic, wrote this entry. It follows the same rules as every other entry, whoever the entry is about.

Why it matters

The same thing, for different readers.

What it means for you

A well-known AI interviewer now expects AI to match the best human AI researchers within a few years, possibly next year, which would speed up everything that follows.

For lawmakers

If Dwarkesh Patel's 2028 median is close, the window for rules on AI systems that help build the next AI systems is measured in months, not a decade.

For researchers

Patel argues AI research is mostly checkable by tests inside a data center, so models can be trained on it directly, and answers objections on learning, data, compute and metrics.

The question it raises

Patel gives even odds to a software-only singularity or a one-year jump of three to six years of progress, raising the question of who decides when such a loop runs.

In their own words

Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.

From our own transcript of the episode’s audio, made by machine. Speaker names come from what is said, not from a voice match.

“I've come to the view that we're only a couple years away and possibly only a couple months away from AIs that are as good as the best humans at AI research.”
Dwarkesh Patel, Host, Dwarkesh PodcastOpening of the video: Patel says how far Patel's own view moved over three months of debating the question.
Found word for word in the sourceHear it at 0:20 ↗starts inside a longer claim, right after “and”
“If I had to pick my median year for when automated researchers will be as good as the very best human AI researchers at doing AI research, I'd guess 2028.”
Dwarkesh Patel, Host, Dwarkesh PodcastPatel's median estimate, assuming labs do not slow down on purpose and governments do not slow them.
Found word for word in the sourceHear it at 8:16 ↗
“I'd guess that there's at least a 25% chance that it doesn't happen until after 2032.”
Dwarkesh Patel, Host, Dwarkesh PodcastPatel's weight on a later arrival, because parts of AI research are hard to automate.
Found word for word in the sourceHear it at 10:04 ↗starts inside a longer claim, right after “and”
“Meter, which is an independent evaluation organization, spent three months embedded inside of Enthropic and afterwards concluded that Enthropic's rate of AI progress on well scoped tasks is 2x what it was a year ago”
Dwarkesh Patel, Host, Dwarkesh PodcastPatel's account of the METR report, treated as evidence that AI is already speeding up frontier AI work. Transcript spells METR and Anthropic phonetically.
Found word for word in the sourceHear it at 1:07 ↗

The record

Questions
What happens when AI helps build more capable AI? · How fast is AI advancing, and where is it heading?
People
Dwarkesh Patel (Speaks)
What we read
We read the whole source.
Found by
The Radar, run of November 14, 2026, via yt-dwarkesh-patel-youtube (tier 1: the source’s own channel or an official record)
Cite this
Dwarkesh Patel (YouTube), “Recursive self-improvement has started”, November 13, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/dwarkesh-2026-11-13-recursive-self-improvement-started
Our notesOur own transcript of the audio; names are spelled phonetically (Meter for METR, Enthropic for Anthropic). Patel's figures about OpenAI and Anthropic usage and a solved Millennium Prize problem are Patel's claims and were not checked here. Entry touches Anthropic and Claude, so flagged for editor review.