Can AI agents conduct open-ended AI research?
A new study introducing 'shadow evaluations' found that frontier AI agents given six days and thousands of dollars of compute could complete all engineering tasks for two unpublished NeurIPS 2026 submissions without huma…