📡 AI Digest — 2026-05-05#
🔥 トレンド（Hacker News）#
- A Data-Driven Machine Learning Framework for Optimising Programmable Terahertz [Hacker News] https://www.researchgate.net/publication/404050094_A_Data-Driven_Machine_Learning_Framework_for_Optimising_Programmable_Terahertz_Meta_surface
- A thermodynamic trust layer cutting LLM hallucinations by 52% [Hacker News] https://github.com/Dan23RR/snc-core
- AI Won’t Save Us from the Talent Crisis We Created (Sep 25, 2025) [Hacker News] https://techtrenches.dev/p/ai-wont-save-us-from-the-talent-crisis
- AI models are choking on junk data [Hacker News] https://fortune.com/2026/05/03/ai-models-are-choking-on-junk-data/
- AI startup JuliaHub raises $65M to rival Simulink [Hacker News] https://www.axios.com/2026/04/30/bob-muglia-ai-hardware-engineering
- Apple Machine Learning Research at ICLR 2026 [Hacker News] https://machinelearning.apple.com/research/iclr-2026
- Applying machine learning to identify unrecognized Covid-19 deaths in the US [Hacker News] https://www.science.org/doi/10.1126/sciadv.aef5697
- Ask HN: Are employers getting the returns from AI? [Hacker News] https://news.ycombinator.com/item?id=48014734
- Ask HN: When did you move from AI agentic loops to simpler deterministic system? [Hacker News] https://news.ycombinator.com/item?id=48014837
- Claude Design fundamentally altered my understanding of AI [Hacker News] https://ttcbj.bearblog.dev/claude-design-fundamentally-altered-my-understanding-of-ai/
- DeepCtx – VS Code extension that auto-builds codebase context for AI tools [Hacker News] https://marketplace.visualstudio.com/items?itemName=DeepCtx.deepctx
- Duralang – decorator makes every LangChain LLM/tool/MCP call a Temporal Activity [Hacker News] https://temporal.io/code-exchange/duralang-durable-stochastic-ai-agents-with-one-decorator
- EU accused of wasting €20B on AI computing dreams [Hacker News] https://www.politico.eu/article/eu-accused-wasting-20-billion-euro-ai-computing-dreams/
- Eight LLM agents wrote 1.7M words; two refused, even when ordered [Hacker News] https://zenodo.org/records/20020017
- Exploring LLM biases to manipulate AI search overview [Hacker News] https://arxiv.org/abs/2605.00012
- Fluctuating Accuracy in LLM Responses [Hacker News] https://news.ycombinator.com/item?id=48008015
- Google Is a Full Stack AI Player, and Is Playing Well [Hacker News] https://www.nextplatform.com/cloud/2026/05/02/google-is-a-full-stack-ai-player-and-is-playing-well/5219190
- How HN: BibCrit – LLM analysis grounded in real manuscript corpus data [Hacker News] https://bibcrit.app/
- How OpenAI delivers low-latency voice AI at scale [Hacker News] https://openai.com/index/delivering-low-latency-voice-ai-at-scale/
- How to Run Any LLM in Claude Cowork and Claude Code [Hacker News] https://www.productcompass.pm/p/cowork-on-3p-any-llm
- HuggingFace launches pre-compiled machine learning kernels repository [Hacker News] https://huggingface.co/changelog/kernels
- Know thyself: LLM schema for personal memory [Hacker News] https://github.com/parrik/know-thyself
- LLM anomaly detectors are not a cause for concern despite Mythos [Hacker News] https://www.magonia.io/research/why-a-decade-of-writing-detection-logic-makes-the-mythos-exploit-numbers-less-scary/
- LLM-first document AI is missing a 50-year-old CS technique [Hacker News] https://bhavyagupta.dev/posts/llm-document-extractors-fixed-point
- Linear Algebra for Computer Vision, Robotics, and Machine Learning [pdf] [Hacker News] https://www.cis.upenn.edu/~cis5150/linalg-I-f.pdf
- Living Brain Cells Enable Machine Learning Computations [Hacker News] https://www.tohoku.ac.jp/en/press/living_brain_cells_enable_machine_learning_computations.html
- Machine Learning Library from the 2000s [Hacker News] http://torch.ch/torch3/
- Machine Learning Operations on ZYNQ FPGA Board for Real-Time Face Recognition [Hacker News] https://www.mdpi.com/2571-5577/9/4/71
- Machine Learning Visualized – Machine Learning Visualized [Hacker News] https://ml-visualized.com/
- Machine Learning for Android Engineers [Hacker News] https://7422889060748.gumroad.com/l/ml-android-engineers
- Machine learning& gut microbiome pathway analysis in Alzheimer’s risk prediction [Hacker News] https://alz-journals.onlinelibrary.wiley.com/doi/10.1002/dad2.70340
- Melon|One-Click Machine Learning Solutions [Hacker News] https://melon.asapi.ai/
- Numeo AI Is Hiring [Hacker News] https://news.ycombinator.com/item?id=48015554
- One year of using ForeverNotes and AI to build a scribe system [Hacker News] https://austinspires.com/blog/2026-05-01-one-year-of-using-forever-notes/
- Optimizing XGBoost Machine Learning Models with AI Agents [Hacker News] https://szilard.github.io/xgboost-autoresearch/
- Pentagon announces deal with seven AI companies for classified systems [Hacker News] https://www.aljazeera.com/news/2026/5/1/pentagon-announces-deal-with-seven-ai-companies-for-classified-systems
- Public Runtime for Convera for LLM’s [Hacker News] https://github.com/cjparadise79/CONVERA-PUBLIC
- Quantum Machine Learning: A Pragmatic Guide for Classical ML Engineers [Hacker News] https://pawankjha.substack.com/p/quantum-machine-learning-the-pragmatic
- Show HN: Aurra – Bi-temporal memory for AI agents (with LLM auto-supersede) [Hacker News] https://www.aurra.us/blog/level-2-auto-supersede-beta
- Show HN: Layers – AI skills for deep product design [Hacker News] https://layers.jamiemill.com/
- Show HN: Llmconfig – configfile and CLI for local LLM [Hacker News] https://github.com/kiliczsh/llmconfig
- Show HN: MegaLLM – Universal LLM client for any OpenAI-compatible API [Hacker News] https://megallm.netlify.app/
- Show HN: My “home rig” for iterative attribute-weighted LLM benchmarking [Hacker News] https://github.com/yuvhaim-gif/LLM_InSight
- Show HN: NoReporter – AI-only newsroom, $1/year [Hacker News] https://noreporter.ai
- Show HN: The Cat Is Under Mayonnaise – Modifying LLM Behavior Without Retraining [Hacker News] https://github.com/andycufari/the-cat-is-under-mayonnaise-experiment
- Synthetic Sandbox for Training Machine Learning Engineering Agents [Hacker News] https://arxiv.org/abs/2604.04872
- The Engineering Constraints of Distributed LLM Inference over the Open Internet [Hacker News] https://siliconandsoul.substack.com/p/the-engineering-constraints-of-distributed
- The Signature Method in Machine Learning (an interactive reading note) [Hacker News] https://ngrislain.github.io/blog/2026-4-6-reading-note-the-signature-method-in-machine-learning/
- The heartland’s revenge: how AI is reindustrializing the American interior [Hacker News] https://www.freightwaves.com/news/the-heartlands-revenge-how-ai-is-reindustrializing-the-american-interior
- The weirdest paradox in statistics (and machine learning) [video] [Hacker News] https://www.youtube.com/watch?v=cUqoHQDinCM
- Thermometry of simulated Bose–Einstein condensates using machine learning [Hacker News] https://iopscience.iop.org/article/10.1088/2632-2153/ae4f4f
- Topology Meets Machine Learning: The Euler Characteristic Transform [pdf] [Hacker News] https://www.ams.org/journals/notices/202507/rnoti-p719.pdf
- Train a LLM from Scratch [Hacker News] https://github.com/raiyanyahya/how-to-train-your-gpt
- Trusted Remote Execution: Policy-Enforced Scripts for AI Agents and Humans [Hacker News] https://aws.amazon.com/blogs/opensource/introducing-trusted-remote-execution-policy-enforced-scripts-for-ai-agents-and-humans/
- TuxAide – Local AI assistant that runs in your Linux terminal [Hacker News] https://github.com/deltaxmodules/tuxaide
- VulkanForge – 14 MB Vulkan LLM engine that runs native FP8 models on AMD (Rust) [Hacker News] https://github.com/maeddesg/vulkanforge
- What do we lose when AI does our work? [Hacker News] https://rickyyean.com/2026/05/04/what-do-we-lose-when-ai-does-our-work/
- Wiki Builder: Skill to Build LLM Knowledge Bases [Hacker News] https://academy.dair.ai/blog/wiki-builder-claude-code-plugin
- You Need MLOps: When CI/CD for Machine Learning Becomes Mandatory [Hacker News] https://blog.ateljevagabond.se/posts/why-you-need-mlops/
- Your AI assistant doesn’t belong in your inbox [Hacker News] https://abiraja.com/blog/your-ai-assistant-doesnt-belong-in-your-inbox
💬 Reddit (MachineLearning / Artificial)#
- [D] Self-Promotion Thread [Reddit (↑9)] https://www.reddit.com/r/MachineLearning/comments/1t1d2m0/d_selfpromotion_thread/
- [D] Monthly Who’s Hiring and Who wants to be Hired? [Reddit (↑7)] https://www.reddit.com/r/MachineLearning/comments/1t0gle1/d_monthly_whos_hiring_and_who_wants_to_be_hired/
- Why SSMs struggle in parameter-constrained training: empirical findings at 25M parameters [R] [Reddit (↑21)] https://www.reddit.com/r/MachineLearning/comments/1t3hxsy/why_ssms_struggle_in_parameterconstrained/
- Is there a notable increase in demand for privacy-preserving AI/ML with the advent of LLMs? [D] [Reddit (↑5)] https://www.reddit.com/r/MachineLearning/comments/1t3u2jh/is_there_a_notable_increase_in_demand_for/
- Are modern ML PhDs becoming too incremental, or is this just what research looks like now? [D] [Reddit (↑138)] https://www.reddit.com/r/MachineLearning/comments/1t311vb/are_modern_ml_phds_becoming_too_incremental_or_is/
- [P] QLoRA Fine-Tuning of Qwen2.5-1.5B for CEFR English Proficiency Classification (A1–C2) [P] [Reddit (↑2)] https://www.reddit.com/r/MachineLearning/comments/1t3ogbw/p_qlora_finetuning_of_qwen2515b_for_cefr_english/
- How do you experiment with a (very) large model architecture? [D] [Reddit (↑0)] https://www.reddit.com/r/MachineLearning/comments/1t3savv/how_do_you_experiment_with_a_very_large_model/
- Parax v0.5: Parametric Modeling in JAX [P] [Reddit (↑1)] https://www.reddit.com/r/MachineLearning/comments/1t3jmdc/parax_v05_parametric_modeling_in_jax_p/
- torch-nvenc-compress: GPU NVENC silicon as a PCIe bandwidth multiplier — PCA + pure-ctypes Video Codec SDK wrapper. Parallel-path overlap measured at 67% of theoretical max on a real GEMM + encode workload. [P] [Reddit (↑18)] https://github.com/shootthesound/torch-nvenc-compress
- AutoBe benchmark: structured harness narrows frontier-vs-local gap in backend generation [D] [Reddit (↑1)] https://www.reddit.com/r/MachineLearning/comments/1t3hks9/autobe_benchmark_structured_harness_narrows/
- Richard Dawkins spent 3 days with Claude and named her “Claudia.” what he concluded after is hard to defend. [Reddit (↑1772)] https://www.reddit.com/r/artificial/comments/1t2z0tn/richard_dawkins_spent_3_days_with_claude_and/
- am I the only one whose friends are completely divided on AI? [Reddit (↑26)] https://www.reddit.com/r/artificial/comments/1t3e57u/am_i_the_only_one_whose_friends_are_completely/
- Vertical vs. Horizontal: Who wins the Agentic AI race in banking? [Reddit (↑4)] https://www.reddit.com/r/artificial/comments/1t3ulp7/vertical_vs_horizontal_who_wins_the_agentic_ai/
- If Claude App gave you the same control as Claude CLI then would you bother with the CLI? [Reddit (↑10)] https://www.reddit.com/r/artificial/comments/1t3if54/if_claude_app_gave_you_the_same_control_as_claude/
- I gave my local LLM a “suffering” meter, and now it won’t stop self-modifying to fix its own stress. [Reddit (↑127)] https://www.reddit.com/r/artificial/comments/1t31ghg/i_gave_my_local_llm_a_suffering_meter_and_now_it/
- What’s the best AI voice generator? [Reddit (↑6)] https://www.reddit.com/r/artificial/comments/1t3l8pa/whats_the_best_ai_voice_generator/
- Built a simple OpenAI Powered TTS feature into my AI platform - truly unlimited free generations, no paywalls, want to know if this is actually useful [Reddit (↑3)] https://www.reddit.com/r/artificial/comments/1t3rp18/built_a_simple_openai_powered_tts_feature_into_my/
- Gallup Analysis Finds AI Not Reducing Artists’ Earnings [Reddit (↑7)] https://letsdatascience.com/news/gallup-analysis-finds-ai-not-reducing-artists-earnings-0b0c052e
- Yoda Translator - Toy 500k Param Model [Reddit (↑2)] https://yoda.jpoehnelt.dev/
- ROCm 7.2.3 brings minor updates, ROCm XIO documentation [Reddit (↑1)] https://www.phoronix.com/news/AMD-ROCm-7.2.3
📰 ニュース#
- OpenAI’s cozy partner Cerebras is on track for a blockbuster IPO [TechCrunch AI] https://techcrunch.com/2026/05/04/openais-cozy-partner-cerebras-is-on-track-for-a-blockbuster-ipo/
- Image AI models now drive app growth, beating chatbot upgrades [TechCrunch AI] https://techcrunch.com/2026/05/04/image-ai-models-now-drive-app-growth-beating-chatbot-upgrades/
- Katie Haun raises $1B for new venture funds [TechCrunch AI] https://techcrunch.com/2026/05/04/katie-haun-raises-1-billion-for-new-venture-funds/
- Hackers are mass-exploiting the cPanel bug to gain control of thousands of websites [TechCrunch AI] https://techcrunch.com/2026/05/04/hackers-are-still-exploiting-the-cpanel-bug-to-gain-control-of-thousands-of-websites/
- Elon Musk’s only AI expert witness at the OpenAI trial fears an AGI arms race [TechCrunch AI] https://techcrunch.com/2026/05/04/elon-musks-only-expert-witness-at-the-openai-trial-fears-an-agi-arms-race/
- Sierra raises $950M as the race to own enterprise AI gets serious [TechCrunch AI] https://techcrunch.com/2026/05/04/sierra-raises-950m-as-the-race-to-own-enterprise-ai-gets-serious/
- Elon Musk sent ominous texts to Greg Brockman, Sam Altman after asking for a settlement, OpenAI claims [TechCrunch AI] https://techcrunch.com/2026/05/04/elon-musk-sent-ominous-texts-to-greg-brockman-sam-altman-after-asking-for-a-settlement-openai-claims/
- Anthropic and OpenAI are both launching joint ventures for enterprise AI services [TechCrunch AI] https://techcrunch.com/2026/05/04/anthropic-and-openai-are-both-launching-joint-ventures-for-enterprise-ai-services/
- DoorDash adds AI tools to speed up merchant onboarding, edit photos of dishes [TechCrunch AI] https://techcrunch.com/2026/05/04/doordash-adds-ai-tools-to-speed-up-merchant-onboarding-edit-photos-of-dishes/
Generated at 2026-05-05 07:00 JST by Emma 🍻