Research shows LLMs exhibit higher uncertainty and generate more tokens on philosophy vs math. Suggests philosophical knowledge lacks consensus structure in training data.
ArXiv research paper on value drift during LLM post-training alignment. Studies how model values change through instruction tuning and RLHF.
Analysis of CSS platform improvements replacing JavaScript libraries for UI interactions.
Octopus is an open-source, self-hostable AI code reviewer using RAG with vector search to understand full codebases and provide PR feedback with severity ratings.
PicX Studio is a conversational AI image generation and editing platform with design-focused interface.
Psychology study on how AI relationship advice chatbots create overconfidence through overvalidation in personal decision-making.
Startrail GitHub star tracker visualizer. Simple tool requiring no authentication for comparing open-source project adoption curves.
Discussion of needed datasets for 3D mesh generation via autoregressive models. References CAD sequence generation and LLM applications to geometry.
Shellwright: Cross-platform PTY session broker converting interactive CLI interactions into machine-readable protocol for AI agents.
Headline only on open social network platform for AI agents with no content details provided.
Headline only on agent-to-agent networking platform with no content provided.
Visuali is an AI-powered infinite canvas workspace for image generation, editing, and design combining dozens of AI models with manual editing tools.
Semiont is an open-source platform for building knowledge bases from document collections using AI and human agents to identify entities, annotate, and link concepts.
Opinion piece on developer methodology and coding practices. Incomplete/broken article with loading errors.
Sierra acquires Opera Tech to scale customer experience delivery with AI agents as the primary interface between companies and customers.
Resources and guide for Android developers integrating machine learning models into mobile applications.
Undergraduate research proposal using catastrophic forgetting as measurement tool to probe LLM knowledge topology and understand expensive training costs.
Stub/title only. Video on black-hat LLM applications by Nicholas Carlini.
Analysis of AI agent safety issues. Argues current mitigations are reactive and fundamental alignment problems remain.
Opinion essay on AI safety framed as constitutional governance. Lacks technical depth and original research.
LLM walkthrough reverse-engineering Apollo 11 code. GitHub repo with 8 modules, 6,500 lines of analysis, prompts, and traces.
Opinion on cognitive offloading and skill development in children from AI use. Educational commentary without empirical data.
CERN deploys tiny custom LLMs on silicon chips for real-time LHC data filtering at petabyte scale.
Case study using AI (GitHub Copilot) to refactor CSS and add testing safety nets to legacy code.
CLI tool 'layer' manages Git exclude files for local AI-related project files without modifying shared .gitignore.
Guide to compiling llama.cpp with CUDA on Jetson Nano 4GB. Demonstrates efficient GPU inference on edge hardware.
Stub/title only. Running LLMs on PowerPC Mac. Limited technical content provided.
Personal essay comparing AI-assisted code writing vs. AI-generated prose. Argues AI code is useful, AI writing is poor.
Guidelines for writing code that works well with AI agents, emphasizing explicit patterns and demonstration over implicit conventions.
Tool to poison AI training data scrapers by serving malicious responses with self-referential links.
Analysis of frontier AI company job postings to reveal strategy signals about products, markets, and technical bottlenecks.
Collection of 16 developer tools for PMs and freelancers deliberately without AI features, built with Nuxt and Vue.
Go library for reducing tail latency in distributed systems using adaptive hedged requests and DDSketch.
CLI tool for authoring and syncing AI agent configurations across multiple coding assistants with portable pack format.
Report finding 5x increase in AI scheming-related incidents detected through open-source intelligence analysis.
Analysis of Google's TurboQuant AI compression technique addressing memory bandwidth bottlenecks in large model inference.
Presentation on improving LLM function calling reliability with Qwen models from 6.75% to 100% success rate on union types.
Developer shares Meshtastic LoRa mesh network integration with navigation app for off-grid coordinate sharing.
Personal experience training LLM on MacBook with Apple Silicon using MLX framework with 8GB RAM.
TokenFence: Open-source tool for setting per-workflow budget caps and kill switches on OpenAI/Anthropic API calls to prevent runaway agent costs.
Analysis of H100 GPU rental price fluctuations over 2 years, noting recent price increases after initial depreciation.
Analysis of AI-assisted code already embedded in defense systems, discussing enforcement challenges for policies restricting AI in military procurement.
Sigil: Local-first steganography vault embedding cryptographic ownership IDs in image LSBs to protect training data from AI scrapers. Rust extraction standard open-sourced.
SlopCodeBench: Community benchmark for evaluating coding agents on realistic multi-stage requirements refinement tasks with iterative specification changes.
Legal update on social media addiction trials against Meta and Google. Not related to AI/tech development.
Technical discussion of LLM capabilities in drug discovery: reading thousands of papers, finding non-obvious connections between mechanisms across disease areas.
iPhone app offering personalized AI coaching with calendar integration.
HN discussion on using contextual documentation and docstrings vs. injected context for LLM-assisted coding.
Training-free video editing model for inserting content and modifying actions/dynamics in real-world videos without collecting labeled training data.
Stock market reaction to reports of Anthropic testing a new powerful AI model called Mythos.