measured-rd-uplift-is-below-the-self-sustaining-threshold CLM-0087observationsingle paperReviewed by AI and Russ Miller
On the best current calibration, AI is not yet accelerating its own development in a self-sustaining way: the modelled threshold is that a one-unit gain in model capability must buy at least 15% higher AI R&D productivity, and the back-of-envelope figure from reported engineer uplift since coding agents launched is about 9% — below it, but rising, so the gap is a current reading rather than a ceiling.
Capability: Whether the measurement made the finding · Evaluation
Observed on
2026, calibrated on the Epoch Capabilities Index and reported engineer uplift.
Sources
- Derives the self-sustaining acceleration condition from the elasticities around the core feedback loop, then calibrates it: the condition is met at roughly 15% R&D productivity gain per unit of capability, and the reported figure is around 9%. States plainly that the return is not constant and has been increasing, so the finding is that we are not there yet, not that we cannot get there.
Either way, bring evidence: a link, or what you saw.
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- For deciding when a robot skill has finished and the next subtask should begin, general-purpose VLMs prompted zero-shot are near useless on long-horizon mobile manipulation (under 4% of transitions detected on BEHAVIOR-1K, including a frontier proprietary model), while a sub-1B student fine-tuned on compact self-explanations condensed from teacher reasoning over annotated demonstrations detects about 90% and lifts closed-loop task success from 0.30 to 0.53 against a 0.63 oracle.Turning plans into physical action · reviewed by AI
- On 170 software and research tasks timed with human domain experts, the length of task that frontier agents complete at 50% reliability, measured in human time, has doubled roughly every seven months since 2019 and was around 110 minutes for o3 in mid-2025 — but the horizon at 80% reliability follows the same trend at roughly five times shorter, and success drops as tasks carry more of the 16 "messiness" factors that separate benchmark tasks from real work, so the headline horizon overstates what an agent can be relied on to finish.Completing tasks that take many steps or many hours · reviewed by AI
- When a calculator-using model is trained with reinforcement learning on verifiable final-answer rewards for an arithmetic search task (Countdown), the gain lands almost entirely at low k — pass@1 rises sharply while pass@16 moves little or falls — because the update can only reinforce correct trajectories the starting policy already samples, and prompt groups with no correct sample supply no gradient.Using the tools it is given · reviewed by AI
- In agent pipelines that generate whole domain-specific ML codebases (medical imaging) from scratch, execution/assembly validation and domain knowledge bases fix different failures: removing runtime and assembly testing mainly costs autonomy (more human interventions to debug), while removing the curated domain knowledge base mainly costs output quality (syntactically fine but domain-inappropriate pipelines) with little change in intervention count.Generating and editing working code · reviewed by AI
- A large share of an agent benchmark score is attributable to the harness around the model rather than the model itself: holding the model fixed and changing only guides, sensors, retries and state management is reported to move the same model from 30.9% to 74.6% on GAIA and 30th to 5th on Terminal Bench — so a score reported for "a model" on an agentic benchmark is a score for a model-plus-harness, and comparisons across labs with different harnesses are not comparisons of models.Whether the measurement made the finding · reviewed by AI
Notes
Filed as an observation, not a mechanism. The 15% threshold comes from a model whose structure could be wrong, and the 9% is a back-of-envelope from self-reported uplift — both will move. The durable part is the shape of the argument, which is held separately in input-telemetry-cannot-establish-self-improvement. Relevant to this project rather than merely interesting: docs/ambitions.md proposes that this system improve itself from its own catalog. This is the serious attempt to say what would have to be true for that to be happening, and its answer is a ratio nobody currently publishes.