Langford: RSI Is a Spectrum, Not a Switch; 10x Annual Research Speedup Plausible

On November 15, machine learning researcher John Langford commented on Dwarkesh Patel's new article and held a multi-round discussion with MIT researcher Dimitris Papailiopoulos on "what research looks like in the era of recursive self-improvement (RSI)." The core claim: a 10x yearly improvement in AI research efficiency would not be surprising, and RSI is more likely a gradual process than a switch-like flip.

Confirmed

Commenting on Dwarkesh's central argument, Langford noted that current AI capabilities are spiky/jagged—below human level in most domains, but consistently ahead in areas with verifiable proxy metrics (e.g., mathematics, AI research itself).
In the discussion with Papailiopoulos, Langford said a 10x speedup in AI research is a reasonable and unsurprising judgment today, and another order of magnitude a year from now would not shock him either.
From this perspective, Langford argued that "full RSI" or amplification by several orders of magnitude may not be what distinguishes future states.
On what research looks like in the RSI era, Langford said the hardest part is coming up with interesting questions.
Papailiopoulos pressed further: will RSI also automate "high-taste questioning" itself? Langford responded that this could happen, but he doesn't expect a switch-like flip—rather a gradual process, where as a larger share of "taste" is automated, the research-speed multiplier gains another zero within a year on top of the previous one.

Why it matters

The discussion grounds the abstract topic of "AI accelerating AI research" in concrete magnitudes: Langford's estimate (one order of magnitude per year) offers an actionable benchmark for assessing RSI progress.
Whether "high-taste questioning" can be automated is the crux of the debate: if RSI also unlocks the ability to pose interesting questions, the uniquely human role in research narrows further, with direct implications for researchers' roles and how science is organized.

2026-11-15 ~ 2026-11-16 · 6 related posts

Full story(3 episodes)→

Episode 1: Dwarkesh Radically Shortens ASI Timeline to Median 2028(2026-11-14, 13 posts)
Episode 2: Debate erupts over Dwarkesh's recursive self-improvement claims(2026-11-15, 3 posts)
Episode 3: Langford: RSI Is a Spectrum, Not a Switch; 10x Annual Research Speedup Plausible(2026-11-15, 6 posts)

Primary sources

John Langford: 10x AI research speedup now unexceptional, another OoM next year unsurprising — JohnCLangford · 2026-11-15 09:02 (UTC+08:00)
John Langford on post-RSI research: asking interesting questions comes last — JohnCLangford · 2026-11-15 21:08 (UTC+08:00)
Langford: RSI isn't a switch, it adds a zero to research speed over a year — JohnCLangford · 2026-11-15 22:59 (UTC+08:00)

[source] John Langford: 10x AI research speedup now unexceptional, another OoM next year unsurprising — JohnCLangford · 2026-11-15
Langford: a 10x AI research speedup is 'reasonable and unexceptional' now — DimitrisPapail · 2026-11-15
[source] John Langford on post-RSI research: asking interesting questions comes last — JohnCLangford · 2026-11-15
Papail: could RSI automate high-taste question asking too? — DimitrisPapail · 2026-11-15
[source] Langford: RSI isn't a switch, it adds a zero to research speed over a year — JohnCLangford · 2026-11-15
Langford: RSI is a spectrum, not a bit — and we're already on it — JohnCLangford · 2026-11-16