Mohamed Aghzal

🚨New Paper at ACL'26🚨 LLM web agents remain far from human reliability, and we try to explain why. We built a hierarchical framework to find out where and why they fail by breaking agent behavior into 3 levels: high-level planning, low-level execution, and replanning. 🧵

[1/6] LLMs/VLMs aren't reliable planners—can they evaluate plans? 🤔
Our #CVPR2025 paper tests this in path planning.
We find that VLMs show weak low-level perception & hallucinated reasoning.
📄 arxiv.org/abs/2411.18711
📊 huggingface.co/datasets/maghz…
📅 Fri Jun 13 4-6 PM @ ExHall D