An AI methods conference track proposed for practical computer vision solutions

Michael Black proposes an AI methods track at computer vision conferences for practical solutions built with frontier models. Submissions would explain the problem and share prompts, code and a demo, potentially earning poster or demo space through an abstract and link rather than a full review process. Black argues that solving an open problem can be worth sharing without technical novelty. Ani Kembhavi supports the idea and emphasizes non-obvious uses of AI, nontrivial problems and a short process that keeps the work relevant. This is a proposal, not an announced conference track.