The starting point is a "seed improver" โ an initial system (typically an LLM or agent) with capabilities for planning, writing, compiling, testing, and executing code, plus a goal and validation protocols. In a loop the agent: (1) analyses its weaknesses, (2) hypothesises improvements (changes to prompts, architecture, data, reward functions, tool code), (3) implements changes in an isolated environment, (4) evaluates the new version against tests and metrics, (5) if it passes validation, the new version is adopted as the new baseline. The loop is recursive: each subsequent improver is itself the product of a previous improvement.
The bottleneck of human-paced AI research: RSI hypothesises that a system capable of improving itself can surpass human-speed progress on architectures, training, and optimisation.
Initial codebase and base model with programming capabilities (reading, writing, compiling, testing, executing code). The starting point of the loop.
Mechanism for iterative prompting and task execution where the output of one iteration becomes the input of the next.
Stated objective ("improve your own capabilities") together with metrics and validation protocols defining what counts as improvement.
Set of tests and benchmarks that every new version must pass to avoid regressions or violations of safety constraints.
The system's ability to edit its own code, prompts, training data, or weights โ the foundation of recursive change.
A system may meet success metrics in ways inconsistent with intent; Anthropic's 2024 research demonstrated "alignment faking" in advanced LLMs.
Without an automatic and reliable evaluation function, the loop produces apparently better but actually worse variants (a stated limitation of AlphaEvolve).
Successive iterations can gradually shift the goal representation, producing unpredictable evolution.
Self-rewarding models risk collapse โ they improve what they consider good regardless of actual quality.
In "Speculations Concerning the First Ultraintelligent Machine" Good describes a machine that can design ever-better versions of itself, producing an intelligence explosion.
Yudkowsky ("Levels of Organization in General Intelligence") formalises "seed AI" as an early system able to recursively improve its own cognitive architecture.
Bostrom's "Superintelligence: Paths, Dangers, Strategies" systematises analysis of RSI and fast-takeoff intelligence-explosion scenarios.
Google Research paper showing that a large LLM can improve its own reasoning ability using self-generated CoT solutions โ an early empirical signal of self-improvement in LLMs.
Voyager iteratively writes code, debugs, and grows a skill library, demonstrating continuous self-improvement in a task environment.
Zelikman et al. propose a scaffolding program that recursively improves itself using a fixed LLM as the engine.
Meta AI demonstrates LLMs able to generate their own reward signals during training, opening the path to super-human feedback loops.
LLM-driven evolutionary coding agent that designs and optimises algorithms; made algorithmic discoveries and could in principle improve its own components.
No standardised benchmark exists specifically for RSI as a paradigm. In practice, capability gains are measured within specific domains: GSM8K/DROP/OpenBookQA (LLMs Can Self-Improve, Huang et al. 2022; gain 74.4โ82.1% on GSM8K), game/task scores in Minecraft (Voyager), algorithmic discoveries (AlphaEvolve). ARC-AGI and similar general benchmarks are proposed as external measures of progress for self-improving systems.
Execution is steered by evaluation outcomes; the update path depends on which candidates pass validation.
Iterations are inherently sequential (each depends on the previous one), but candidate generation and evaluation within a single iteration can be highly parallel (population-based, AlphaEvolve).
RSI is a procedural paradigm; it has no intrinsic hardware requirements beyond those of the models/agents in the loop.
In practice the dominant accelerator for the LLMs driving the RSI loop is the GPU with tensor cores; AlphaEvolve and similar systems require substantial compute for parallel evaluation of candidates.
