AMVICC: A Novel Benchmark for Cross-Modal Failure Mode Profiling for VLMs and IGMs
Aahana Basappa, Pranay Goel, Anusri Karra, Anish Karra, Asa Gilmore, Kevin Zhu · 2026
Read AMVICC: A Novel Benchmark for Cross-Modal Failure Mode Profiling for VLMs and IGMs by Aahana Basappa, Pranay Goel, Anusri Karra, Anish Karra, Asa Gilmore, Kevin Zhu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-17037.txt.
Description
Aahana Basappa and a team of five co-authors build a test for sight AI's blind spots. Sight-text bots read shots and write of them. Text-to-shot bots make shots from words. Both kinds fail on plain sight traits. Aim, count, and place of things trip them up. Their 2026 preprint, AMVICC, sorts those fails side by side.
The test lets teams set read fails next to make fails. Shared gaps point to deep flaws in sight sense. Such maps show what bots truly grasp of scenes. They guide fixes past mere scale-up of bots. Safe use in cars and care needs such maps.
Catalog record
| CREATOR | Aahana Basappa, Pranay Goel, Anusri Karra, Anish Karra, Asa Gilmore, Kevin Zhu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-17037