Cross-Dataset Bloom Question Classification: Supervised Models and Prompted LLMs

Collection of recent AI/ML papers (arXiv) covering: cross-dataset evaluation of Bloom’s taxonomy question classification and LLM prompting (LLMs generalize better than supervised models); PictoPercept — an open-source visual toolkit to measure human and model bias (GPT-5 shows stronger biases than humans); an EduNLP systematic review highlighting misaligned EdTech incentives and under-served teachers; empirical work on ESG narrative scoring showing limited incremental value from reasoning-heavy LLMs relative to cheaper ensembles; position and theory pieces advocating planet-centered AI and re‑