AI conference ICLR is drowning in abstracts, with roughly 50,000 submissions before the deadline

ICLR 2027 has received approximately 50,000 abstracts before the submission deadline, a figure that dwarfs the roughly 19,500 valid papers from the previous year. Many of these entries likely represent authors hedging their bets by submitting to multiple venues, with plans to withdraw if accepted elsewhere. Corporate research spending and the general hype surrounding artificial intelligence are also driving this surge. The most significant factor is that AI tools now allow researchers to generate papers much faster. A recent analysis of NeurIPS submissions confirmed that authors heavily relied on AI to write their work. This trend complicates the peer review process, which already struggled with low-quality submissions and fabricated citations last year. Reviewers previously turned to AI just to manage the volume, creating a cycle where the tools used to produce research also help evaluate it. As the number of entries grows, concerns about trust and quality are expected to intensify.

Key details include:

• The deadline for submissions is next week.
• Final accepted numbers will be lower than the current abstract count.
• Trust in the peer review system is already eroding.