Am Neumarkt 😱

#ai J. Simon et al., “There will be a scientific theory of deep learning,” arXiv [stat.ML], Apr. 23, 2026. doi: 10.48550/arXiv.2604.21691. Available: http://arxiv.org/abs/2604.21691. [Accessed: May 01, 2026]

#ai PaperOrchestra https://yiwen-song.github.io/paper_orchestra/

#tools llm-wiki https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f

#ai It might be hard to distinguish deceptive behavior and ignorance due to being incapable. ----- Opinion: AI systems can easily lie and deceive us – a fact researchers are painfully aware of https://www.westernsydney.edu.au/news-centre/stories/2025/opinion-ai-systems-easily-lie-and-deceive-us-fact-researchers-aware-of

#misc Claude Code source code leaked, at least part of it. https://x.com/Fried_rice/status/2038894956459290963

#ai Wow, the compression is amazing. Seriously, is there a similar correspondence like the Ads/CFT correspondence for transformers? https://swarmnotes.kausalflow.com/

#vibecoding I tested this set of skills for AI coding agents and it works really well. One of the pain points of building frontend is to specify precisly what we need from the agents, but without UI/UX knowledge and expertise, it is not easy. This set of skills preset a lot of prompts and we just need to call those skills instead of breaking our head for consistencies. https://impeccable.style/

#misc The Slow Collapse of MkDocs — Florian Maas https://fpgmaas.com/blog/collapse-of-mkdocs/

#ai https://productivityradar.com/how-to-use-ai-as-a-manager/

#ai https://www.figure.ai/news/helix-02-living-room-tidy

#ai What, is this real? Look at the authors... https://www.researchgate.net/publication/401488012_Integrating_AI-Driven_Predictive_Analytics_for_Proactive_Cybersecurity_Risk_Mitigation_in_Large-Scale_IT_Program_Management

#misc https://www.anthropic.com/research/labor-market-impacts