4-page short papers
Submitted as single PDFs up to 50 MB; references and appendices do not count toward the page limit.
AI/CORE Unranked
AID-Wild 2026
ACM CAIS 2026 Workshop on AI Agents for Discovery in the Wild
AID-Wild 2026 is an inaugural ACM CAIS workshop focused on AI agents that perform discovery in real-world settings—beyond benchmarks—under constraints like expensive evaluations, noisy feedback, and human oversight. It brings together research on agents for scientific discovery, infrastructure optimization, and deployment challenges in science, engineering, and systems.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
9-page long papers
Submitted as single PDFs up to 50 MB; references and appendices do not count toward the page limit.
Research areas in scope
01
Search and Optimization with LLM Agents
Agents that explore design spaces, generate candidate solutions, and optimize over code, configurations, or experimental parameters — including evolutionary, iterative, and tool-augmented approaches.
02
Agents for Systems Infrastructure
AI agents applied to scheduling, capacity planning, database tuning, network architecture, cluster management, performance debugging, and other infrastructure operations.
03
Agents for Scientific Discovery
Agents that assist with hypothesis generation, experiment design, proof construction, data analysis, or literature synthesis — across domains such as mathematics, natural sciences, and engineering.
04
Evaluation Under Real-World Constraints
Methods for validating agent behavior when ground truth is limited, experiments are expensive, feedback is noisy, or reliability requirements are strict.
05
Deployment Reports and Failure Analysis
Case studies from applied settings, including adoption challenges, failure modes, operational lessons, and what didn't work.
06
Human-Agent Collaboration and Oversight
Workflows where domain experts direct, audit, or intervene in agent-driven processes — including trust calibration, delegation boundaries, and expert-in-the-loop design.
Policies worth checking twice
- Submissions are double-blind with at least two reviews per paper.
- Review PDFs must not reveal author identity, including linked or supplementary material.
- References and appendices do not count toward the page limit.
- Concurrent submissions and recently accepted work are welcome if venue policies permit.
- Submissions and reviews will not be public; only accepted papers will be made public.
- All accepted papers will be presented as posters; a small number will be selected for oral presentations.
- The workshop will present a Best Paper Award.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026