Harvard/MIT paper FINSKILLOPS makes financial AI self-improve via regression-tested skills
rohanpaul_ai · x · 2026-09-22
A paper from Harvard, MIT and other labs introduces FINSKILLOPS, a way for SEC-filing analysis agents to continuously learn from their mistakes without regressing.
- Core idea: turn a static RAG system into one that safely updates its behavior, requiring every new behavior to pass regression tests first.
- Mechanism: a small skill registry converts repeated errors (wrong reporting periods, mixing up companies) into reusable instructions.
- A new instruction deploys only if it fixes the target failure and still passes checks on cases the system already answers correctly — turning repeated mistakes into tested skills.
More from Research
- SteerDuplex: full-duplex speech model gains 44.5pp in steerability, new SteerBench released — ScaleAI · 2026-09-22
- Sarah Hooker suspects fake AI paper submissions cluster at a few universities — sarahookr · 2026-09-22
- Xiaomi MiMo's CodeMIDAS scales agentic coding RL environments from code itself — _akhaliq · 2026-09-22
- Rollout scheduling: the underappreciated infra trick boosting inference efficiency — stochasticchasm · 2026-09-22
- Conference fixes for AI paper flood: per-author submission caps and faster fake-profile detection — sarahookr · 2026-09-22
- Benchling benchmarks Claude and ChatGPT on wet-lab protocols: helpful, not solved — nlarusstone · 2026-09-22