Weekly Bookmarks
These are some things I’ve wandered across on the web this week. 🔖 Learning How to Learn: Abduction as the ‘Missing Link’ in Machine Learning In this paper, the question of machine learning is…
What's it like talking to AI
Thank you, Bad Girl Media
Exactly How and Why AI Will Replace Knowledge Work
And why this is actually a good thing for human meaning and creativity
Sovereign compute for small collectives: a technical implementation guide
Figure 1 This is the technical companion to my sovereign LLM post, which makes the institutional and geopolitical case for small collectives owning their own AI inference hardware. This post…
Community sovereign AI compute
Figure 1 It turns out that I’m writing a series of posts about practical community infrastructure building. This is the second one. In a previous post I sketched out a case for neo-friendly…
Spell checking with Claude
I was listening to an interview of Simon Willison just now, who mentioned that spellcheck is something that used to be really bad, but was now great with LLMs. I'd noticed the same thing when I…
hating on AI writing
A couple of weeks ago I read this NYT article by Sam Kriss about the telltale signs of AI writing. Then I started seeing them everywhere. Most of the uses are innocuous. They annoy me, but I can…
the em dash conununundrum
We have all recently heard that if a piece of writing has been created using AI, then apparently it is likely that it uses em dash punctuation frequently. This is of course very unfortunate for those…
What The Fudge for March 22, 2026: The Edge of Intelligence
Twice in a row! Here we go! /* Hide without JS */ is-land:not(:defined).video-wrapper { display: none; } lite-youtube { max-inline-size: 100% !important; background-size: cover; } is-land…
The Bright Forest Theory
The Anti-Thesis to the dark forest theory
Brain Fry
Reactions to a recent HBR article and my own experience with over-delegation to Claude Code
A Data Scientist RLM That Lives in Your Program
… or how to process DataFrames with RLMs and DSPy
System reminders - how Claude Code steers itself
Claude Code has 37 hidden reactive messages that nudge the agent mid-conversation. Here's how they work and how to build your own.
New App: Imago
Generate images from text using Flux 2
Dev Weekly: Nvidia GTC Drops Vera Rubin, Stripe Ships 1,300 PRs/Week with AI, Laravel 13 Arrives
Nvidia unveils Vera Rubin at GTC 2026. Stripe's Minions agents ship 1,300+ pull requests weekly. Microsoft restructures Copilot leadership. OpenAI signs AWS deal for government AI. Laravel 13 and…
Week 12, 2026 · OpenAI Eats the Python Toolchain, Claude Cowork Scheduling, and AI Observability
What I learned Claude Cowork Scheduling You can schedule recurring prompts in Claude Cowork. I’m now using it to get a daily AI observability news digest emailed to me every morning. Still tweaking…
Voice AI in 2026
Since I’ve been exploring the extended universe of voice AI and voice agents, I’ve met a lot of people working with them in the real world and it’s not just the average customer support bot anymore.…
Caring about Sloppypasta
Things are insane right now, and maybe I’m too. Be that as it may, I feel like I’ve been a broken record for the last 18 months. Maybe longer. Some nights I wake up at 3am, startled, sweaty, and I…
Taste is the New Bottleneck
For a decade I told my mentees 'ideas are cheap, execution is everything.' AI broke that aphorism. When execution is free, the only thing that matters is taste.
Slowing down to go fast
How a debugging session with a coding agent reminded me of the need to learn things with intention
Code as Config: The Start of Software Speciation
The era of the identical binary is ending as autonomous agents begin directly mutating source code, fragmenting once-uniform programs into bespoke digital organisms. This post explores the profound…
DuckDB, AI, and the Future of Data Engineering | with Staff Engineer, Matt Martin
The post DuckDB, AI, and the Future of Data Engineering | with Staff Engineer, Matt Martin appeared first on Confessions of a Data Guy.
On Bernie vs Claude
I just watched Bernie vs Claude (9 min, youtube) and I have thoughts: This is pure vintage Black Mirror stuff, down to the eerie music, the washed in blue shots, had to do a double take to check if…
agent harness notes
opencode’s edit tool doesn’t trust LLMs to give exact string matches.
Pas mal non ? C'est français
A quick look at the French AI sloperator Mistral AI
pi-autoresearch optimization loops
The autoresearch technique promoted by Andrej Karpathy is easily accessible in the pi coding agent.
You Don't Need a Bigger Workhorse, You Need a Smarter Llama
Run the same large language model in two different interfaces and you'll get what feels like two different intelligences. Not a little different — different enough that you'd swear…
Random shitheads who use AI to monetize shit
As I logged in on GitHub, I was suggested one so-called trending repository. It was FujiwaraChoki/MoneyPrinterV2. I then discovered harry0703/MoneyPrinterTurbo, at some point “number one shit of the…
Claude answers GPT re ailien minds. And the winner is...
Last time, I posted a lengthy missive by ChatGPT appraising my new book ailien minds, prompted by Richard Bliss, with some of my own commentary and demurals laced-in. All-told, the effect was that of…
Report from the teleoplexic frontier: human science may be on the path to becoming obsolete
Greatest living human mathematician Terence Tao said this in an interview with Dwarkesh Patel:I’m not sure nowadays that hypothesis generation is the bottleneck anymore. Science has changed in the…
LLM Inference and Optimization: Fundamentals, Bottlenecks, and Techniques
LLMOps Part 13: Exploring the mechanics of LLM inference, from prefill and decode phases to KV caching, batching, and optimization techniques that improve latency and throughput.
Why OpenAI Needs a Chief Snark Officer as It Doubles Its Workforce
OpenAI plans to double its workforce—but as it scales into an AI powerhouse, this open letter argues it’s missing one critical role: a Chief Snark Officer to keep it human.
Things take time
Yes, slow down this year. The Earendil site looks nice, particularly the use of Departure Mono. There’s a blog too: Interaction: As quantum physicists know, the universe emerges more from how objects…
Live Vibe Coding using Others' Ideas
I spoke today on Design in the Age of Infinite Generativity at the Chennai Design Festival. You can read about the talk in the link about. This post is about my preparation. Tue 10 Mar 2026. Damn!…
2026-03-21 17:08
Nvidia Nemotron 3 Superhttps://build.nvidia.com/nvidia/nemotron-3-super-120b-a12bNvidia presented their new model - Nemotron 3 Super, an open hybrid Mamba-Transformer MoE model: 120B total / 12B…
OpenClaw’s ChatGPT moment sparks concern that AI models are becoming commodities
By Jonathan Vanian People queue to have their laptops install with OpenClaw, an open-source AI assistant at the Baidu headquarter in Beijing on March 11, 2026. Adek Berry | Afp | Getty Images Three…
matrix city
How do you simulate civilization when you don't understand psychology? You reach for the worst tools available and it turns out that's good enough.
AI Slop on Etsy
Amy Kay This is a rant. I used to like Etsy, I bought some queer bracelets off of there a few years back and they lasted me a good while. My money went to a shop in LA, and independent artist making…
Bot, Are You the One Who Is Living Life?
And yet, though we strain against the deadening grip of daily necessity, I sense there is this mystery: All life is being lived. Who is living it, then? Is it the things themselves, or something…
NVIDIA DLSS 5
Earlier this week, Nvidia posted a video featuring their latest iteration of DLSS technology (Deep learning super sampling). DLSS 5 is an AI-powered “real-time neural rendering” process and unlike…
Prompting Structure and Mix-Ins
Use OS text replacement to expand prompt boilerplate
Hedging Your Bets on AI
My friend and colleague Patrick Prill has been writing articles about AI in our industry at a prodigious rate over the last couple of weeks. (Here's the second of yesterday's pieces.) I recommend…
Slowing Down in the Age of Coding Agents
Everyone is trying to go faster, running more agents to run more agents, to open more pull requests to orchestrator agents to then get reviewed by another fleet of agents. The arrival of coding…
On AI
#Enshittification #Sustainability #Tools
Maintaining a human web: human.json & the ‘AI’ blacklist
My dislike strong opposition to engaging with “AI” generated things, be that texts, images, code or anything else is well - documented . But avoiding the mindless slop is still getting harder, as it…
Using ChatGPT to Tanscribe Homework
/* Reset margins and set full height */ html, body { margin: 0; padding: 0; height: 100%; } /* Use flex layout for full page height */ body { display: flex; flex-direction: column; font-family:…
AI-moderated Research is a Modality Now. Stop Arguing About What to Call It. It Doesn't Need Your Approval. It Needs a Framework.
Five people sent me the same link on the same morning. That's either a sign the UXR community has fully become a hive mind, or that something important just happened.It was the second…
The problem with Perplexity (and other AI assistants)
On Friday, March 20, 2026, I read news that Chuck Norris had passed away. Naturally I went over to Perplexity.AI to see if the story was verified or a hoax. What I encountered was a dystopian…
How to Attract AI Bots to Your Open Source Project
A practical guide to getting the engagement your project deserves.