Christian Schaller describes using Claude (an LLM) to automate much of the drudgery of reverse-engineering undocumentated hardware by building a reusable "hw-harness": run the vendor Windows driver in a VM with device passthrough, capture bus traffic from the Linux host with usbmon, repeat with a Linux driver, and diff traces so the AI can spot changing bytes, propose meanings, and generate tests and patches. That workflow produced working results: a SANE backend for the Plustek OpticFilm 8200i (Genesys GL128, USB 07b3:1825) with color/infrared modes, up to 7200 dpi and infrared dust removal; Negative Vibes, a Rust GNOME Flatpak app that preserves 16-bit linear/raw + IR, does density-space inversion, dye-crosstalk correction, roll/frame overrides, IT8 calibration and multiple export formats; and targeted fixes for a ThinkPad X1 Carbon Gen 14 MIPI camera (Sony IMX471 + Intel IPU7).
He gives concrete technical fixes and measurements: kernel fixes for crop-selection, test-pattern register width and analog gain cap; libcamera fixes for missing lens-shading correction and mis-scaled white-balance statistics. In a low-light test brightness rose from 31.1 to 90.9 and color ratios moved toward a webcam reference. Lessons: AI excels at parsing captures, proposing constants and tests, and eliminating tedium, but humans must run hardware verification and make design choices; the harness observes device behavior rather than copying driver code, and reference captures are crucial.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Terence Tao discusses the allure of AI and large language models in advancing mathematics and science, comparing it to Borges's imagined Library of Babel containing all possible knowledge. He highlights how AI's potential to generate vast amounts of research could revolutionize scientific discovery, despite the challenges of sorting through misinformation. (terrytao.wordpress.com)
Companies rushed to adopt AI technology, leading to increased errors, low-quality work, and security risks. Some firms, like Meta, are now pulling back on AI integration to address these issues and reduce costs. (theguardian.com)
Accenture has limited employee spending on the AI tool Claude to $29 per month, down from as much as $1,000. Employees must now find ways to fund additional usage if they exceed the cap, reflecting broader industry cost-cutting measures on AI tools. (404media.co)
Muse Spark Open is the most capable open-weight AI model to date, optimized for desktop use on Apple M5 Ultra Macs. It excels at complex reasoning, coding, and agentic tasks, and is available for download and integration through multiple platforms. (research.meta.ai)
Flux 3 Image Klein is a 5.6 billion parameter model designed for text-to-image generation and image editing, accepting prompts and reference images to produce or modify images. It uses the Qwen3-VL-4B-Instruct text encoder and supports various applications, including fine-tuning and research. (huggingface.co)
Muse Spark Open is the most capable US open-weight multimodal model with 816 billion parameters, optimized for desktop use on Apple M5 Ultra Mac Studios. It excels at coding, complex reasoning, and agentic tasks, and is available through multiple platforms for customization and deployment. (research.meta.ai)
Today's best Hacker News stories, summarized and screenshotted, one email a day.