Clicked Gallery

What is recursive self-improvement?

Highlighted from a real engineering doc. Explained by Clicked.

Used in a sentence

Engineering Notes · AI Systems

The authors argue that recursive self-improvement stays bounded by compute and data, not by algorithmic insight.

The reader highlighted one word in the docs. Clicked explained the technical term “recursive self-improvement” in simple terms:

Explained in three depths

Same facts, different vibe — Slang mode 😎

The Clicked way

●○○

Overview

Recursive self-improvement is a system doing the work of improving itself, where each new version is better at that same work. The clearest case is an AI writing part of the code used to train the next AI, which then helps build the one after that. Every version is more capable than the last and takes less time to produce, so progress speeds up instead of holding a steady pace.
●○○

Overview

Recursive self-improvement is when the thing gets better at the job of getting better. Normal progress is you grinding away at the same pace forever, and this is the grind quietly picking up speed on its own. Point an AI at building its own replacement, let that one build the next, and you have the whole idea. 😎

A quick take — often all you need.

●●○

Detail

An AI lab uses its current model to help write code and run experiments for the next one. That next model is both more capable and quicker to produce, so it takes on a bigger share of the work for the model after it, and the cycle keeps tightening. Why not just leave this to the engineers, which is how every other technology advances? Because then people set the pace, and each new model still takes about the same number of months no matter how good the last one was. No system does this whole job by itself yet. The pieces that do exist are narrow ones: programs that became unbeatable at chess and Go by playing millions of games against copies of themselves, and labs that use their own models to speed up their research. What people argue about is the ceiling. One side thinks the loop could run faster than safety teams can test each new model, and the other thinks it slows to the pace of building chips and running experiments, which no amount of intelligence shortens.
●●○

Detail

Quick version: you write a script that saves you two hours, then spend those hours writing a better script. Recursive self-improvement is that, except the thing doing the writing is also the thing getting better. In AI it means a team pointing today's model at the work of making the next one, and that next one is sharper and lands sooner, so it handles even more of the job after that. Why hand it to the machine at all? Because people are the speed limit, and every version still eats roughly the same months however sharp its predecessor was. Nothing runs the full loop solo yet, and the closest we have is board game engines that went unbeatable by drilling clones of themselves. Where it tops out is a genuine fight, with one camp saying it outruns the safety folks stuck checking every release, and the other saying it bogs down waiting on hardware and lab time. You will meet the phrase any time somebody starts predicting how fast this stuff can move. 😎

Want more? One click digs deeper.

●●●

Analogy

A 3D printer that prints toys will keep printing toys at the same speed forever. Point it at itself instead and something different happens, because it can print better parts for its own frame, and the upgraded printer then prints better parts than it could have managed before. Each version is built by a better machine than the one before it, so the improvements arrive faster and go further. It still needs plastic and power though, which is exactly where people disagree about how far this can run.
Deepest level reached.
●●●

Analogy

Some games hand you an upgrade that makes you level faster, and leveling faster is exactly what unlocks the next upgrade. An hour in, you are gaining way more per hour than you were at the start, and nothing about you got smarter or quicker. The build did that on its own, which is the entire trick and also why it feels slightly unfair. The only thing that stops it is the game running out of levels to hand you. 😎
Deepest level reached.

Unfamiliar concept? A real-world example makes it click — fresh analogies on tap.

AI explanations may contain errors · Not professional advice

Formal definition — The same term, explained the usual way

Recursive self-improvement describes a process in which a system applies its own capabilities to enhancing its architecture, training, or reasoning, such that each improved version is more effective at producing further improvements. The concept is defined by a feedback loop acting on the rate of capability gain rather than on capability alone, and its practical extent is constrained by compute, data, and the time required for empirical validation.

Want Clicked to explain terms like “recursive self-improvement” directly in your browser — including on PDFs?

Add to Chrome — Free

50 free Explanations · No credit card required