AI agents still need humans to teach them

"AI agents need skills - specific procedural knowledge - to perform tasks well, but they can't teach themselves, a new research suggests."

"The researchers looked at each task under three conditions: no skills (agent receives instructions only), with curated skills (provided with a directory, code snippets and resources to help it) and self-generated skills (agent has no skills but is prompted to develop them). Typical tasks included conducting a security audit of npm dependencies for vulnerabilities, or analyzing differential protein expression in cancer cell line data."

SkillsBench evaluates agentic AI performance on 84 tasks spanning 11 domains including healthcare, manufacturing, cybersecurity, and software engineering. Each task is assessed under three conditions: no skills (agent receives instructions only), curated skills (agent is provided a directory, code snippets, and supporting resources), and self-generated skills (agent is prompted to develop its own skills). Results indicate that agentic AI requires explicit procedural knowledge to perform tasks effectively and that agents struggle to self-teach such skills. Representative tasks include conducting security audits of npm dependencies for vulnerabilities and analyzing differential protein expression in cancer cell line datasets.

Read at Computerworld