https://rsi4science.org/news/ Selected primary-source research on recursive self-improvement and its scientific applications. Source checks are AI-assisted and do not establish independent replication.en Mon, 12 Oct 2026 20:16:17 GMT https://arxiv.org/abs/2610.12176v1rsi4science:news:7f79a91ca454996507ecMASS trains a shared model on selected agent trajectories, then uses the updated model to execute, optimize and evaluate the next round. It connects learning from task experience with later self-improvement. Evidence boundary: Evidence covers two cycles with one backbone. Token efficiency excludes total training and search cost, and transfer varies across tasks. The study leaves longer-run scientific improvement open.Thu, 08 Oct 2026 15:44:02 GMTSelf-improvementResearch agentsModel learninghttps://arxiv.org/abs/2610.02975v1rsi4science:news:d25f019836e0f7fb48fdCISE uses calibrated reward intervals to guide materials search. It addresses a practical dataset question: how can feedback prevent promising-looking but infeasible candidates from steering later trials? Evidence boundary: The coverage results depend on independence and covariate-shift assumptions. This is evidence about feedback reliability in materials search; recursive revision of the improvement procedure was outside the study.Fri, 02 Oct 2026 08:07:27 GMTScientific discoveryMaterialsFeedback reliabilityhttps://arxiv.org/abs/2609.26457v1rsi4science:news:f330de6f12df20ba842dAIDE² evolves research-agent code. Its separate ignition test puts an evolved agent back into the role of improving agents—the return path that matters for recursive self-improvement. Evidence boundary: The main evolution run uses a fixed outer improver. The ignition test does not establish a stronger or more efficient improver; scientific transfer remains limited.Tue, 22 Sep 2026 00:00:00 GMTImprover handoff experimenthttps://arxiv.org/abs/2607.05297v1rsi4science:news:87762250e8461166d7b2MetaSkill-Evolve edits the meta-skills that guide its improvement pipeline. The revised meta-skills are reused in later rounds, including subsequent edits to those meta-skills, while the base model stays frozen. Evidence boundary: These are not scientific-discovery tasks. The baseline descriptions differ on branching and sharing, so the reported gain cannot be cleanly attributed to recursion alone.Mon, 06 Jul 2026 00:00:00 GMTRecursive mechanism outside sciencehttps://arxiv.org/abs/2609.17523v1rsi4science:news:8ed7a0f4faed007f9ce5ScienceBuddy couples harness edits and model learning. It is useful for examining the boundary between a scientific system that keeps improving and a demonstrated recursive improvement advantage. Evidence boundary: The auxiliary reflector remains fixed. Better scientific task performance does not establish that the improvement mechanism became stronger or isolate a recursive benefit.Tue, 15 Sep 2026 00:00:00 GMTScientific co-adaptation boundary case