The NIH unveiled GeneAgent, an artificial intelligence system that mitigates hallucination errors in gene function prediction by autonomously cross-verifying outputs against curated biological databases. Unlike typical large language models that can produce factually incorrect results and circular reasoning, GeneAgent verifies gene-set analysis results using expert knowledge bases like Gene Ontology and MSigDB. Validation by human experts confirms improved accuracy and reliability for functional genomics research, representing a significant advancement in gene-set enrichment analysis methodology.