[CFP] ICLR 2026 Workshop: ICBINB - Where LLMs Need to Improve (Submission Deadline: Jan 31)

[Jennifer Williams]

ICLR 2026 Workshop: I Can’t Believe It’s Not Better - Where LLMs Need To Improve

Despite rapid progress, LLMs still fall short in surprising ways. Many of these failures, negative results, limitations, and boundary cases, can be difficult to publish but crucial for moving the field forward. This workshop aims to create a platform for open, honest discussion about the hurdles and roadblocks in building reliable, efficient, and safe LLM systems. By sharing these experiences, we can prevent teams from retracing unproductive paths, strengthen our understanding of failure modes and boundary conditions, and foster a culture of transparency and learning. 

We welcome papers that

• Showcase and investigate important limitations of current LLMs. This may include the evaluations on pitfalls in common approaches to alignment, reasoning, etc., and evaluations of real-world (especially safety-critical) applications.
• Attempt promising ideas to overcome common challenges but fall short of the expected gains, accompanied by analyses that clarify failure modes and boundary conditions. 
  

Submission Formats

• Short papers: 4 pages (excluding appendices)
• Tiny papers: 2 pages

Key Dates

• Submission Deadline: January, 31 2026 (11:59 pm AOE)
• Acceptance Notification: March 1, 2026
• Workshop: April 26 or 27, 2026

Awards 
Outstanding submissions will be nominated for spotlight talks and two awards:

• Entropic Award - Most surprising negative results
• Didactic Award - Most pedagogically valuable paper 

A non-archival track will always be available. Authors may choose whether their paper is archival. 

Full CFP and submission details.

We look forward to your contributions!

Sincerely,
The I Can’t Believe It’s Not Better Organizing Team