Tag: #academic-ai6 results
Japanese firm LifePrompt released results showing ChatGPT, using OpenAI's latest model, outperforming human top scorers in multiple subjects on Tokyo University and Kyoto University entrance exams. This demonstrates strong AI performance in rigorous academic tests.
Fields Medalist Terence Tao praises Claude Code for handling all peer review comments on his paper in just 15 minutes. The AI not only completed revisions swiftly but also identified flaws in the reviewers' feedback.
Stanford researchers developed an LLM-based agent that guides reviewers to make feedback more constructive and polite. Tested on 20K ICLR 2025 reviews, 24% were revised with 80 more words on average, 68% deemed better by experts. No impact on scores or acceptance rates.
University of Washington researchers developed PaperTok, an AI-powered tool that transforms dense academic papers into short-form video content. It automates script generation, storyboarding, and author attribution.
A researcher used AI to create 42 songs about PhD struggles like rejected papers and burnout, launching 'Don't Pursue PhD Music Fest' that amassed 50M Bilibili views. Drawing from personal injury hiatus, he refined generations via trial-and-error. The project healed him and resonated with thousands via fan letters.
Nature Index publishes a special issue on the Tianli Discipline Brain, an AI system grounded in complex systems theory. It signifies a shift from experiential teaching to data-driven scholarship. Featured by QuantumBit, highlighting academic AI innovations.