In a massive study of 22,000+ papers, humans actually preferred AI-generated peer reviews over human ones.
April 16, 2026
Original Paper
AI-Assisted Peer Review at Scale: The AAAI-26 AI Review Pilot
arXiv · 2604.13940
The Takeaway
Peer review is supposed to be the ultimate human expert task, but AAAI-26's pilot study shows a major shift. Authors and chairs found AI reviews to be more technically accurate and provided better research suggestions than their human counterparts. This isn't just about speed; it's about the quality of the technical critique. This result is a massive 'vibe shift' for academia and high-end professional services. If AI can outperform PhDs at technical peer review, it can likely outperform most humans at any high-level document auditing task. It marks the moment AI moved from 'writing assistant' to 'expert technical judge' at scale.
From the abstract
Scientific peer review faces mounting strain as submission volumes surge, making it increasingly difficult to sustain review quality, consistency, and timeliness. Recent advances in AI have led the community to consider its use in peer review, yet a key unresolved question is whether AI can generate technically sound reviews at real-world conference scale. Here we report the first large-scale field deployment of AI-assisted peer review: every main-track submission at AAAI-26 received one clearly