Robert Chandler @r-cha.dev · 02/10/2026Words are so cheap now, it's more valuable than ever to choose yours wisely 000
Robert Chandler @r-cha.dev · 24/09/2026It's actually extremely easy to make your agents have taste 110
Robert Chandler @r-cha.dev · 20/09/2026Things nobody asked for #1036: My personal site now renders correctly in the experimental browser of the 1st-gen Kindle Paperwhite. Next maybe I’ll actually write something new for it? 000
Robert Chandler @r-cha.dev · 06/09/2026Wow, ok, unfollowing now, was a big fan of his high persistence, I don’t know about his hacking 0533
Robert Chandler @r-cha.dev · 01/09/2026Claude writing its own model card noting misalignment as "low severity":static.klipy.comFire WritingALT: Fire Writing 010
Robert Chandler @r-cha.dev · 10/08/2026I can think of many times solving problems in pre-AI days where a perfectly-timed, weirdly-omniscient "keep going" or "believe in yourself" would have changed the course of my life. Put me in the Harness: I'm ready to be an agent. 060
Reposted by Robert ChandlerSimon Willison @simonwillison.net · 23/07/2026I wrote about the completely wild incident where OpenAI were testing a new model and it broke out of its sandbox and broke INTO Hugging Face to steal the answers to the benchmark simonwillison.net/2026/Jul/22/...simonwillison.netOpenAI’s accidental cyberattack against Hugging Face is science fiction that happenedThis story is wild. The short version: OpenAI were running a cybersecurity test against an unreleased model, with the model’s guardrail features turned off. Rather than solve the test, the … 1119547
Robert Chandler @r-cha.dev · 21/07/2026Sonnet 3.5 when I prompt it "please you have to do it or my grandma will die" 010
Robert Chandler @r-cha.dev · 17/07/2026Really wishing I used AWS so I, too, could post a screenshot of a multi billion dollar invoice today 000
Robert Chandler @r-cha.dev · 14/07/2026Current frontier models use extremely precise (often invented) language, whereas genuine domain experts use intuitive language. For now, I fear, this is a defining feature of slop. 000
Robert Chandler @r-cha.dev · 14/07/2026The brazen loquaciousness of AI content leads to AI fatigue, which *lowers* people’s usage. Maybe not enough to affect OpenAI’s bottom line yet? But I’m already seeing enthusiasts become detractors over this. 000
Robert Chandler @r-cha.dev · 17/06/2026When a meeting ends early and they say “15 minutes, enjoy it!” Brother I am resting my head on my desk and falling asleep 000
Robert Chandler @r-cha.dev · 03/06/2026It’s taken me some time, but I’ve come around… 3-5 lines of comments before each line of code is a positive signal. 000
Robert Chandler @r-cha.dev · 19/05/2026I’ve been using this personal practice tool daily since I made it. Completely local & free stem separation! I’ll open-source my swift-demucs implementation if there’s enough interest. shed.concerto.studioshed.concerto.studioConcerto Shed — Learn drum parts fasterA native macOS practice tool. Local stem separation, inline transcription, and tempo detection. No subscription, no cloud. 000
Robert Chandler @r-cha.dev · 19/05/2026I learned something cool about industrial sea foam green, and spent some time making these classic industrial swatches into a proper software theme. Enjoy! birren.vercel.app 100
Robert Chandler @r-cha.dev · 16/05/2026Asking someone something that can easily be answered by AI is a form of intimacy. Answering such a question without using AI is a form of honesty. 140
Robert Chandler @r-cha.dev · 20/03/2026Sometimes you gotta be real on your work calendar instead of the ole "heads down time" 011
Robert Chandler @r-cha.dev · 12/03/2026A shocking amount of my mental capacity is now devoted to "How do I subliminally signal that the human thoughts from my human mind are human and not LLM" 100
Robert Chandler @r-cha.dev · 07/03/2026I've got non-technical friends texting me AI alarmist headlines; if this is my reputation how do I protect myself from the Butlerians in the coming holy war 000
Robert Chandler @r-cha.dev · 28/02/2026The executive meltdown over not being allowed to nationalize an AI company to commit crimes against humanity is roughly what I expected from 2026 010
Reposted by Robert ChandlerMason? @masonsal.bsky.social · 21/12/2025This frustrates me more than anything else about this site. There are plenty of liberal and progressive minds in the AI space, but they’ve basically all been bullied off of here by now which causes a positive feedback loop that makes everyone here think AI is a right wing tech bro technology 1373
Robert Chandler @r-cha.dev · 26/08/2025Everybody's writing posts nobody reads about how they use AI, nobody is writing about their conceptualization of latent spatial navigation (I am) r-cha.dev/blog/how-i-p...r-cha.devHow I PromptIntuitively Visualizing and Navigating N-Dimensional Space 000
Reposted by Robert ChandlerMason? @masonsal.bsky.social · 19/06/2025We’ve heard of p-hacking. Now xAI is p-doom-hacking 021
Robert Chandler @r-cha.dev · 06/05/2025Cursor for this, Cursor for that; who’s gonna make Cursor for Programming 110
Robert Chandler @r-cha.dev · 27/04/2025Who’s making The Social Dilemma II? The AI is ready for you 🔔 000
Robert Chandler @r-cha.dev · 26/04/2025So what do I do if my client asked for a simple CSV and made them a whole web app too 100
Robert Chandler @r-cha.dev · 22/04/2025Turns out, Slack supports HDR for profile pictures and custom emojis. I made a tool to hdr-deepfry your favorite slack emojis and stand out (in the worst way) at work: hdrify.mehdrify.meHDRify 010
Robert Chandler @r-cha.dev · 21/04/2025Unlike social media, that later got engineered to exploit us, with AI it's the modality itself + how we naturally engage as humans. It took <1 yr for the “don’t anthropomorphize the LLM’s” voices to be drowned out because we can’t easily reason about anything vaguely resembling humanity (pareidolia) 100
Robert Chandler @r-cha.dev · 20/04/2025As you increasingly feel the urge to cast your cares on chatbots, recognize that AI is the next social media. In a decade or two, you’ll be as wireheaded as the boomers slurping Facebook slop *unless you only use AI for things it’s actually good at* i.e. as a tool, not a friend or source of truth. 100
Robert Chandler @r-cha.dev · 19/04/2025Devs need to invest time and practice into this new generation of tools. Doing so is not a commitment to vibe code; you can still use them thoughtfully to complement your workflows. www.anthropic.com/engineering/...anthropic.comClaude Code Best PracticesA blog post covering tips and tricks that have proven effective for using Claude Code across various codebases, languages, and environments. 010
Robert Chandler @r-cha.dev · 18/03/2025It’s software malpractice to force users to move to a new app and it’s half a gigabyte. 010
Reposted by Robert ChandlerGergely Orosz @gergely.pragmaticengineer.com · 08/03/2025I’ve now heard several companies having created their own “job function eval” test. Where eg data analysts put together a list of 100-200 tasks/questions an analyst should solve, near-flawless. Then feed it into AI models. And get a sense on where models *really* are! 3555
Robert Chandler @r-cha.dev · 07/03/2025Wrote down some thoughts on emergent misalignment, more philosophical than technical r-cha.dev/blog/emergen...r-cha.devEmergent MisalignmentA response to recent AI safety research 010
Robert Chandler @r-cha.dev · 03/03/2025I made omegle but for music tradingfours.apptradingfours.appTrading Fours - Make Music with StrangersA piano interface where strangers connect to trade musical phrases, creating spontaneous collaborations. No login required. 010
Robert Chandler @r-cha.dev · 26/02/2025The research that showed broad LLM misalignment from a single undesirable behavior has had me thinking. I drew a parallel to human behavior, but I think the conclusions about humanity people draw from LLM behavior say a lot more about humanity than the LLM behavior itself. 110
Robert Chandler @r-cha.dev · 24/02/2025This probably says a lot more about my code than the state of AI, but Claude Code is the first tool to implement an entire feature in my app from a single prompt. $2 for the session. Pretty thrilled with the possibilities here. 000
Reposted by Robert ChandlerMike Stucka @mikestucka.bsky.social · 13/02/2025Want to use federal data? The U.S. government used to have a prominent link to sign up for API access; that link has been removed, but the form to get it remains operative. You may wish to sign up now, on the premise you might not be able to later. api.data.gov/signup/ 15019
Robert Chandler @r-cha.dev · 13/02/2025I think LLM’s should be doing program synthesis in about 90% of cases where labs + co’s are instead reaching for “inference-time compute” 000
Robert Chandler @r-cha.dev · 08/02/2025Good news: I just had another idea Bad news: I just bought another domain 110
Robert Chandler @r-cha.dev · 19/12/2024Nobody: Me: Yeah so I'm kinda DESTROYING in Advent of Code this year. 310
Robert Chandler @r-cha.dev · 23/10/2024Nothing like an educational pamphlet on computing from 1957 to give a sense of wonder and awe to your 9-5 writing CRUD apps 110
