Agentic AI usually performs better with step-by-step instructions from a human, researchers find.

InfoWorld is a source of news, analysis, and commentary on technology trends, IT strategies, and business innovation. With a focus on enterprise technology and digital transformation, InfoWorld offers insights and guidance for IT decision-makers, software developers, and technology professionals. From  articles on cloud computing and cybersecurity to product reviews and industry trends, InfoWorld helps readers navigate the complexities of modern IT environments and make informed decisions to drive business success.

InfoWorld

New research introducing SkillsBench, a benchmark evaluating agentic AI across 84 tasks in 11 domains, finds that AI agents perform significantly better when given curated, human-provided skills (averaging 16.2 percentage points higher) compared to no skills or self-generated skills. Agents prompted to develop their own skills showed no performance improvement, underscoring that human guidance remains essential for effective agentic AI. Performance gains from human input varied by sector, with healthcare benefiting most and software engineering least.

AI agents still need humans to teach them