AI
177 items
104 posts, 62 tools, 11 guides
Terry Tao on coding agents: how he ported two dozen 1999 Java math applets to JavaScript in hours, built new visualization tools, and weighs risk and trust.
AI 2040 Plan A is the AI Futures Project scenario for a US-China accord that delays superintelligence to 2040. What it proposes and what HN thinks.
Ghost Font hides text in video using motion, so humans can read it but frame-by-frame AI models cannot. How it works and how HN commenters broke it.
pgvector, Pinecone, Qdrant, Weaviate, Chroma, Milvus, and Turbopuffer compared on hosting model, filtering, scale, and cost for RAG.
An 82M parameter text-to-speech model that runs on CPU and produces high-quality speech across multiple languages - no cloud APIs or GPU required.
Mistral Robostral Navigate is an 8B robotics model that navigates with only a camera and natural language commands. How it works and what the benchmarks mean.
Ilya Sutskever's rumored 30-paper ML reading list, now on 30papers.com. What the papers cover, the compression thesis behind them, and what HN thinks.
Small AI models run on handheld devices where networks fail, like an on-device pill scanner. HN debates offline emergency kits and domain-specific models.
A Dartmouth study of an AI tutor in intro statistics links full engagement to a 0.71-1.30 SD exam gain. What it measured, and why selection bias still matters.
Comparing LLMs by token pricing alone can lead you to choose worse, more expensive models. Cost per task tells the real story.
Event sourcing for AI agents: a new paper makes the append-only log the source of truth for agent state. What it enables, and whether HN thinks it is just CQRS.
Agents 101 companion guide: how to build and deploy AI agents fast on the Vercel agentic infrastructure stack, and where to go next to learn each piece.
Ornith-1.0 open-source coding models explained: what "self-improving" really means, the four model sizes, hardware needs, HN reactions and whether to try it.
A developer fed 266MB of DICOM MRI data to Claude Code for a second opinion on a shoulder diagnosis. It disagreed with the doctor; HN radiologists weighed in.
GLM 5.2 vs Claude Code on Semgrep IDOR benchmarks: the open-weight model scored higher at about one-sixth the cost per finding. Setup, results and HN reactions.
Filippo Valsorda argues LLMs have made vulnerability reports less special. His case against old coordinated disclosure norms, plus the HN pushback.
Unlimited OCR is Baidu's open-source vision-language model that parses 100+ page documents in one pass without memory blowup. How it works and when to use it.
Cloudflare's wrangler deploy --temporary flag gives AI agents an ephemeral account to deploy Workers with no OAuth or browser, plus a claim link to keep it.
LLM architectures got complicated: MoE routing, mixed attention variants and integrated vision encoders replaced the simple transformer. What changed and why.
The Transformer co-creator leaves Google DeepMind for OpenAI just two years after Google paid $2.7 billion to bring him back from Character.AI.

Get Smarter About AI Dev
New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.