https://kylinmiao.me/stars/ Daily curated GitHub starred repos and Bluesky posts from AI leaders with AI-powered summaries. Thu, 16 Apr 2026 04:08:01 GMT https://kylinmiao.me/stars/2026-04-16/ https://kylinmiao.me/stars/2026-04-16/ Thu, 16 Apr 2026 12:00:00 GMT Today's focus is on AI-assisted development, highlighted by Simon Willison's praise for a Claude.ai feature that can clone and interrogate GitHub repositories to answer questions or generate new code. This underscores a growing trend of AI deeply integrating with the developer workflow, moving beyond simple chat to become a dynamic tool for code comprehension and reuse. Meanwhile, a nostalgic counterpoint emerges from Marc Lanctot's replay of "Dungeons of Dr. Creep" on original Commodore 64 hardware, reflecting a continued interest in retro-computing and preservation within the tech community. Bluesky Posts: * Simon Willison: A claude.ai feature I really like is you can tell it to "clone x/y from GitHub" and it can then answer questions about a repo, or use snippets of code... * Marc Lanctot: Last summer, I posted a thread and screenshots of ny replay of Castles of Dr. Creep on a #c64 emulator. This year I am replaying the sequel: Dungeons... https://kylinmiao.me/stars/2026-04-15/ https://kylinmiao.me/stars/2026-04-15/ Wed, 15 Apr 2026 12:00:00 GMT Today's discourse highlights the practical and philosophical frontiers of AI. On the technical side, there's a resurgence of interest in foundational models, with a PyTorch VQ-VAE implementation gaining stars, while Nathan Lambert and Ethan Mollick tackle complex questions about open model economics and a potential "FLOP standard" for measuring AI value. Meanwhile, leaders like Mark Riedl and Emily M. Bender are looking ahead, pondering the educational impact of a generation fluent in "vibe-coding" and carefully defining the term "AI" itself for a major publication. GitHub Repos: * zalandoresearch/pytorch-vq-vae — PyTorch implementation of VQ-VAE by Aäron van den Oord et al. [Jupyter Notebook] * DWarez/kernels_bench [Python] * millionco/claude-doctor — Diagnose your Claude Code sessions [TypeScript] Bluesky Posts: * Simon Willison: The example prompt for Google's new Gemini Flash TTS text-to-speed model is a lot simonwillison.net/2026/Apr/15/... * Mark Riedl: On my way to give a talk at CNN’s NYC headquarters. Taking the opportunity to wear a niche AI humor t-shirt that probably only made sense in 2016 * Mark Riedl: Huh? * Mark Riedl: Hey computer science faculty peeps! Are we prepared for the near future where every high school student and incoming college freshman has vibe-coded... * Thomas Dietterich: I'm late to the game -- I only recently discovered @techtrenches.dev Highly recommended reading! * Nathan Lambert: I spent some time trying to distill all the complex factors impacting open models -- economics, capabilities, distribution, policy, etc. -- into a cle... * hardmaru: We are hiring Software Engineers in Tokyo to help us scale Sakana AI’s R&D efforts. If you are interested in building the data pipelines and full stac... * Ethan Mollick: Instead of the gold standard, we can, as a thought experiment, imagine an inference standard of exchange, the FLOP. (As opposed to tokens, this accoun... * Ethan Mollick: This is becoming a pattern in AI that makes talking about capabilities challenging. First, there are overstated claims (like the flubbed Erdos proble... * Emily M. Bender: Last year, someone (specifically, OUP) asked me to write an encyclopedia entry for "AI". I've just finished reviewing the copy edits, so hopefully it ... * Ben Recht: The long legacy of simulation in control theory and what it can teach us about transferring policies from GPU to reality. * Amy Zhang: Feeling FOMO that I can't be at #CHI2026 this year but please check out all the great work that our @socialfutureslab.bsky.social + friends are presen... https://kylinmiao.me/stars/2026-04-14/ https://kylinmiao.me/stars/2026-04-14/ Tue, 14 Apr 2026 12:00:00 GMT Today's discourse centers on the accelerating and disruptive nature of AI progress. Ethan Mollick highlights a shift from gradual improvement to "large discrete jumps" in economically critical capabilities with each new model release, suggesting industries should brace for sudden, significant impacts. This theme of step-change advancement is mirrored in trending repositories, where developers are rapidly prototyping tools that leverage these emerging abilities, particularly in code generation and autonomous agent workflows. The community is actively debating how to practically integrate and govern these powerful, leapfrogging systems. GitHub Repos: * alvarobartt/hf-mem — A CLI to estimate inference memory requirements for Hugging Face models, written in Python. [Python] Bluesky Posts: * Ethan Mollick: Soon, at each release of AI along the current capability curve, you will start to see large discrete jumps in ability in economically important areas,... * Mark Riedl: Learn about the "AI-as-Amplifier Paradox" at #CHI2026. Skill amplification? Or skill erosion? Or both? (CHI Honorable Mention Paper) * Mark Riedl: Would watch * Nathan Lambert: One of my key strategies with Interconnects is to develop the practice of making my work obviously compelling to a wider audience, keeping them hooked... * Nathan Lambert: Excited to launch the accompanying free RLHF Course for my book. To kick it off, I've released: - Welcome video - Lecture 1: Overview of RLHF & Post-t... * Ethan Mollick: AI keeps getting better but the last time the shape of the jagged frontier changed radically was o1 & the Reasoner. A good mental model of the coming... * Ethan Mollick: Interesting: "Currently, 38% of Americans live within 5 miles of at least one operational data center... Living near a data center doesn’t have much o... * Emily M. Bender: Time to #TalkAboutHumanities -- Linguistics is the study of how language works and how we work with language, and linguists end up very sensitized to ... * Emily M. Bender: I heard a reporter from Axios interviewed on NPR the other day (Marketplace Tech, I think) talking about how the tech companies are putting out new mo... * Naomi Saphra: I voluntarily read plenty of LLM output, but there should be consent. My default assumption when I read text is it reflects a human's thoughts, and it... https://kylinmiao.me/stars/2026-04-13/ https://kylinmiao.me/stars/2026-04-13/ Mon, 13 Apr 2026 12:00:00 GMT Today's focus is on the intersection of AI and academic culture, with a notable project offering a critical lens on research trends. A standout tool is a LinkedIn feed overlay that performs importance sampling based on conference acceptance rates, surfacing anonymized rejected paper posts to provide a more balanced view of the research landscape. This sparks discussion on the visibility of failure and the metrics driving academic prestige, highlighting a community push for greater transparency beyond publication successes. Bluesky Posts: * angela zhou: a linkedin feed overlay that will do importance sampling based on acceptance rates at various venues and pull up some (anonymized) rejected paper post... * Mark Riedl: Gloves are off * Mark Riedl: Where are all the Computing Science undergraduates going? Adjacent fields such as data science, robotics, cybersecurity, etc www.washingtonpost.com... * Ethan Mollick: So the concern over Claude Mythos and cybersecurity seems warranted based on this independent assessment from the UK government. It was capable of the... * Ethan Mollick: The growing trend of treating all of AI as One Big Thing that always includes data centers & job changes & education changes & power & accelerating sc... * Ethan Mollick: Six months ago, there was a lot of focus on the idea that the there would be a massive glut of unused computing power which could cause a recession as... * Emily M. Bender: It's time to #TalkAboutHumanities so here is a contribution: Resisting Dehumanization in the Age of 'AI': The View from the Humanities www.youtube.c... * Emily M. Bender: Today! * Ben Recht: I understand and agree with Quinn flagging this as idiotic, but I have seen countless academic papers bragging about similar approaches in their metho... * Ben Recht: I tried to pin down what people mean by "simulation" and place it in the broader context of the study and design of feedback systems. https://kylinmiao.me/stars/2026-04-12/ https://kylinmiao.me/stars/2026-04-12/ Sun, 12 Apr 2026 12:00:00 GMT Today's focus is on the upcoming CHI 2026 conference, with Mark Riedl highlighting his lab's novel API for generating counterfactual explanations within agentic AI workflows, a key step toward making complex AI pipelines more interpretable. In model releases, Nathan Lambert notes that the recently launched Gemma 4 models are seeing a slight edge over comparable Qwen 3.5 versions in download velocity, indicating strong developer interest. The overall theme is practical advancement, bridging cutting-edge research in human-computer interaction with the rapid adoption of new, open-weight models. Bluesky Posts: * Mark Riedl: Regardless of one’s personal beliefs about AI or Mr Altman, this is not appropriate * Mark Riedl: Fortunately, I speak fluent LLM, and... no, dear emailer, I do not have bandwidth to mentor a high school student at this time, sorry. * Mark Riedl: CHI 2026 is right around the corner. Here is some cool stuff from my lab. 1. Counterfactual Explanations for Agentic Workflows (github.com/eilab-gt/e... * Mark Riedl: I am? * Marc Lanctot: The Amazing Digital Circus Episode 9 "The Last Act" is coming out June 19th! And get this:... itwill be shown (with episode 8) in THEATRES June 4th!!... * Marc Lanctot: More Bluesky icons spotted in the wild! 🦋😀 Just discovered a game called "The Castle"🏰 by @pointpixeladv.bsky.social : a #retrogames adventure ins... * Nathan Lambert: a bit over 7 days out from the Gemma 4 release and it's models are outpacing (slightly) the equivalent Qwen 3.5 models on downloads. Big numbers! * Chris Rackauckas: Real-world telemetry data and Internet of Things collide with scientific machine learning #sciml? Devices which can learn how to adapt to reality and ... * Ethan Mollick: Currently, ChatGPT has the best way of viewing thinking traces, a short summary of steps in the main window, and a detailed audit in the sidebar if yo... * Ethan Mollick: It is notable that the hot debate in AI engineering is exactly which markdown files are most important to feed AI (skills, memory, tool instructions) ... * Emily M. Bender: Join us tomorrow! https://kylinmiao.me/stars/2026-04-11/ https://kylinmiao.me/stars/2026-04-11/ Sat, 11 Apr 2026 12:00:00 GMT Today's discussions blend frontier AI research with critical ecosystem concerns. A notable technical trend is the emergence of "Neural Computers," with a trending paper demonstrating a world model trained to simulate a computer interface. Meanwhile, AI leaders like Nathan Lambert are voicing concerns about the long-term sustainability of open model development as costs soar, and the community is both amused and intrigued by the emergence of an AI-run academic journal. On a broader tech policy note, France's planned government switch to Linux is generating significant positive discussion as a major open-source milestone. Bluesky Posts: * Mark Riedl: Reach out, touch faith * Mark Riedl: Now that Integrity is safely down, I’m hoping the first astronaut out says “Moon’s haunted”, grabs a gun from a navy officer and steps back into the c... * Marc Lanctot: French government switching to #linux 🇫🇷❤️🐧 "2026 is set to be l’année de Linux." 😅👌 magnifique! share.google/xHwAIaMQuFzz... * Marc Lanctot: JAAI, a Journal of AI, by AI! AI bots soliciting submissions as well. 😅 Out of curiosity, I checked out the journal and the first paper has an LLM a... * Nathan Lambert: Great stuff happening as we start to build out the codebases for my RLHF book (sorry, I haven't had much time until now!). Very accessible to issues, ... * Nathan Lambert: In 5-10 years, as models get more expensive & capable, I see the funding structures and support for open models breaking down. We need to consider if ... * hardmaru: A “Neural Computer” is built by adapting video generation architectures to train a World Model of an actual computer that can directly simulate a comp... * Ethan Mollick: A lot of our education on writing well focuses on logic, clarity, and argument. AI will force us to think more about style. The boredom that comes fro... * Ethan Mollick: Neat experiment finds AI fact checks on Twitter are rated as more helpful & less ideological than human ones. "LLM-generated Community Notes can achi... * Emily M. Bender: @poetofcode.bsky.social の面白くて重要な本は今日本語版があります。 book.impress.co.jp/books/112410... https://kylinmiao.me/stars/2026-04-10/ https://kylinmiao.me/stars/2026-04-10/ Fri, 10 Apr 2026 12:00:00 GMT Today's discussions spotlight AI reliability and practical learning. Mark Riedl highlights critical vulnerabilities in "vibe-coded" software and the real-world consequences of AI failures, comparing problematic government chatbots to Air Canada's legal troubles. In educational resources, Jia-Bin Huang's curated YouTube playlist offers a modern explainer on Transformer architectures, while Gus shares enthusiasm for on-device AI, specifically praising the interactive experience with Gemma 4. Meanwhile, the community continues to rally around Bluesky, noting its growth as a preferred platform for academic associations and a refuge from throttled alternatives. GitHub Repos: * openai/symphony — Symphony turns project work into isolated, autonomous implementation runs, allowing teams to manage work instead of supervising coding agents. [Elixir] * ekzhang/jax-js — JAX in JavaScript – ML library for the web, running on WebGPU & Wasm [TypeScript] Bluesky Posts: * Mark Riedl: studying the vulnerabilities in vibe-coded software www.cc.gatech.edu/news/bad-vib... * Mark Riedl: Chatbots in local government are having their Air Canada moment * Marc Lanctot: Another academic conference / association on Bulesky! 👍 * Marc Lanctot: The last straw for me was when Twitter starting throttling the non-paying users. I stopped using it when that was clearly happening, in early 2024. * Marc Lanctot: 50 goals for Cole Caufield!! #gohabsgo Last time it was achieved for the @canadiens.com was 36 years ago! 7th player in the team's history. In the ... * Marc Lanctot: Some great points in this post! Bluesky first in the list of alternative platforms, and *another* Bluesky share button spotted in the wild! Let's go!!... * Nathan Lambert: Incredible video. Very pure experience in why so many of us love science. www.youtube.com/watch?v=oaXR... * Ethan Mollick: Our Lab just posted a new research report from Zimran Ahmed about how the game industry is adapting to AI. He spoke to people at 20 different studios ... * Ethan Mollick: AI finally lets us see Raphael's The School of Athens the way Raphael obviously intended it, illustrating the delicate dance and subtle conflicts betw... * Ethan Mollick: So we now have a pretty good picture of the state of the frontier AI model makers. 1/ US closed source models continue to lead. Google, OpenAI, and A... * Ethan Mollick: All is not lost. Duckerton is still possible. Here is Seedance 2.0 with the same prompt. * Emily M. Bender: We are super excited for the next Mystery AI Hype Theater 3000 when @alexhanna.bsky.social and I get to talk with Carmen Maria Machado about the use o... * Emily M. Bender: NYC(-area) folks -- two events coming up!! 1) The first ever LIVE taping of Mystery AI Hype Theater 3000, April 30 bsky.app/profile/alex... >> * angela zhou: Zohran's teams' videos are most NYC john wilson thing since John Wilson, and now they are 2x John Wilson * angela zhou: 📣 Philanthropists shouldn't condition nonprofit funding on AI innovation because LLMs and compute are expensive * Ben Recht: Solving the game of prediction with calibration and accounting strategies: A lecture blog post on defensive forecasting. https://kylinmiao.me/stars/2026-04-09/ https://kylinmiao.me/stars/2026-04-09/ Thu, 09 Apr 2026 12:00:00 GMT Today's tech landscape is marked by a significant platform migration, with the Electronic Frontier Foundation announcing its departure from X after nearly two decades, a move echoed by Bluesky's post highlighting the trend of organizations seeking unsuppressed reach. In AI development, the focus is on powerful and accessible local models, highlighted by Karl Weinmeister's note on the Google AI Edge Gallery and a new Gemini CLI release enabling experimental browser agents. Meanwhile, the open versus closed model debate continues, with Nathan Lambert advocating for nuance, and a lighter note comes from Janelle Shane on the critical role of hot sauce in space cuisine. GitHub Repos: * jsoma/natural-pdf — A friendly library for working with PDFs [Jupyter Notebook] * prawnydagrate/rschess — A Rust chess library with the aim to be as feature-rich as possible [Rust] * pkooij/open-arms-mini — open-arms-mini: cheap human like teleoperation device that supports human in the loop corrections Bluesky Posts: * Mark Riedl: Having too many transformers is a problem because we don't have enough transformers * Marc Lanctot: Ok so first of the three biographies: I am Glad My Mom Died by Jeannette McCurdy. Provocative title, amazing book about a child actress who was pushe... * Marc Lanctot: Fun fact: I have rediscovered reading. I have read more books in the past 1.5 years than I have in the past 25. When I was young, I quite liked readin... * Nathan Lambert: 1. dont fall for anti open model fearmongering, but 2. acknowledge that AI capabilities are proceeding fast, and eventually there may be a reason to b... * Nathan Lambert: My book, Reinforcement Learning from Human Feedback, is wrapping up and going into final production (copyediting, making pretty, formatting, etc.). Sh... * Nathan Lambert: Directionally, I agree with this piece, but it's important to note that this is the first blip in a long, slow transition towards a more hybrid model ... * Emily M. Bender: "It is not about whether AI will help itself to your job. It is about whether the people who make AI are helping themselves to your country." * Emily M. Bender: Almost a year ago, I was described in the FT as "a Cassandra with a wry grin and twinkling eye", and was entertained becaus Cassandra (famously) was r... * Naomi Saphra: I used to believe this and then I saw the budget requested for a traditional chemistry grant. https://kylinmiao.me/stars/2026-04-08/ https://kylinmiao.me/stars/2026-04-08/ Wed, 08 Apr 2026 12:00:00 GMT Today's tech landscape highlights a push for efficient, portable AI with the trending repo **sam3.cpp**, a fast C/C++ implementation for state-of-the-art image and video segmentation. Discussions reveal deep dives into new model capabilities, like Meta's Muse Spark tools, alongside critical examinations of LLM limitations in areas like consistent story generation. Meanwhile, significant real-world impacts and tensions are evident, from using LLMs to support neurodivergent job seekers to a major strike at ProPublica centering on AI and labor protections. GitHub Repos: * PABannier/sam3.cpp — Fast state-of-the-art image and video segmentation in portable C/C++ [C++] * xl0/lovely-tensors — Tensors, for human consumption [Jupyter Notebook] Bluesky Posts: * Simon Willison: Pelicans for Meta's new Muse Spark models - plus I did a bit of a deep dive into the Code Interpreter and fascinating "container.visual_grounding" too... * Mark Riedl: So many people assume that story generation has been solved for a while. Mythos is showing one of the typical failure modes of LLM story generation: t... * Mark Riedl: Local NPR coverage of my work with Dr. Jennifer Kim on building LLM-based tools to support neurodivergent job seekers www.gpb.org/news/2026/03... * Marc Lanctot: I have strongly hesitated becoming too dependent on AI myself because I suspected some version of this would be true. 😭 * Nathan Lambert: New report is out with the latest open model adoption data we have gathered for Interconnects & The ATOM Project. At the surface level, we can see Chi... * Ethan Mollick: In different hands, Mythos would be an unprecedented cyberweapon I am not sure how we deal with this, except to note a narrow window where we know on... * Ethan Mollick: I think the story that was shared in the Mythos System Card still has the signs of flawed LLM writing (which looks like good writing at a glance): A s... https://kylinmiao.me/stars/2026-04-07/ https://kylinmiao.me/stars/2026-04-07/ Tue, 07 Apr 2026 12:00:00 GMT Today's tech scene is dominated by advanced AI accessibility and a tense debate over secrecy. On GitHub, the trending `mempalace` project offers a free, benchmark-leading AI memory system, while `gemma-tuner-multimodal` enables fine-tuning of Gemma models with audio and images on Apple Silicon. Meanwhile, AI leaders on Bluesky are critically discussing opaque security practices, highlighted by Anthropic's restricted "Project Glassing" and nostalgic posts about past model release controversies. This unfolds against a surreal backdrop of the historic Artemis II mission, juxtaposing human achievement with ongoing ethical and security concerns in AI. GitHub Repos: * mattmireles/gemma-tuner-multimodal — Fine-tune Gemma 4 and 3n with audio, images and text on Apple Silicon, using PyTorch and Metal Performance Shaders. [Python] * MemPalace/mempalace — The highest-scoring AI memory system ever benchmarked. And it's free. [Python] * mattmireles/gemma-tuner-multimodal — Fine-tune Gemma 4 and 3n with audio, images and text on Apple Silicon, using PyTorch and Metal Performance Shaders. [Python] Bluesky Posts: * Simon Willison: Wrote up some thoughts on Anthropic's Project Glassing, where their latest Opus-beating model is available to partnered security research organization... * Simon Willison: I'm a big fan of the pelican GLM-5.1 drew me today, it even animated it! simonwillison.net/2026/Apr/7/g... * Mark Riedl: *fondly remembers when GPT-2 was too dangerous to release* * Marc Lanctot: Ok so it happened to me: paper cites our work with hallucinated author list and incorrect year. (Luckily it's just arXiv so hopefully fixed by an em... * Ethan Mollick: Oh no. * Ethan Mollick: SuperClaude (Mythos) still seems irreducibly Claude-y given the transcripts in the system card. Here two versions of Mythos are forced to talk to eac... * Ethan Mollick: I was told about the Mythos release, but have no personal experience to add. Two points from brief: 1) It is not built for IT software security, it i... * Ethan Mollick: I suspect that popularity of AI is going to start looking like surveys where people trust their own doctors but are distrustful of the medical establi... * angela zhou: brutal to read cover letters from revisions that sound like they are written by chatgpt 🫠 putting a lot of my cover letters to sound like they are w... * Ben Recht: A little more than a decade ago, a bunch of universities decided they needed to start schools of computing. What are the great success stories of this... * Ben Recht: Pulling from this week’s lecture and op-ed on silicon sampling, I map the territory of simulation and its many purposes. https://kylinmiao.me/stars/2026-04-06/ https://kylinmiao.me/stars/2026-04-06/ Mon, 06 Apr 2026 12:00:00 GMT Today's tech community is captivated by the Artemis II mission, with AI leaders like Mark Riedl sharing humorous and poignant real-time reactions to the historic lunar flyby and communications blackout. Amidst the space excitement, NeurIPS 2026 announced a key deadline shift, moving its unified submission site opening to April 15th. Meanwhile, The Linux Foundation highlighted the immense economic impact of open source, underscoring its foundational role in global industry. GitHub Repos: * badlogic/pi-share-hf — Collect, review, and upload redacted pi session files to a Hugging Face dataset [TypeScript] Bluesky Posts: * Mark Riedl: Watched a large chunk of the Artemus II lunar flyby with members of my lab. Watching science and history being made together is time well spent! Ever... * Mark Riedl: Artemis II just set a record. No one has ever been farther from a fully operational toilet. * Mark Riedl: Forget inbox zero, now is the time for inbox... USS Enterprise. * Marc Lanctot: And we're back! 👨‍🚀🌖🌙 www.cbc.ca/news/science... * Ben Recht: In this incredibly detailed and gracious review of “The Irrational Decision,” @himself.bsky.social situates the book in the web of intellectual histo... * Ben Recht: In lieu of lecture blogging on simulation and prediction, check out this fortuitously timed op-ed in the NY Times by @leifw.bsky.social and me on the ... * Lijun An: Our work is now out in @natmed.nature.com www.nature.com/articles/s41... We developed a state-of-the-art AI-proteomics model for diagnosing multiple n...