from Dean Baker Productivity growth is an old concept; we’ve been seeing it at a substantial pace for more than 200 years. Nonetheless, many elite intellectual types like to claim they know nothing…

If you don’t know what success looks like, neither does your AI. Almost every single failure I see mentioned on social media around AI, from deleting a company’s entire database to…

An EBM finally crossed 800M parameters without collapsing. Nobody has independently reproduced the 35% scaling claim. Both halves matter.

And the American consumer is struggling! The post I thought Elon hated Anthropic appeared first on Cautious Optimism.

Nostalgia for the pre-AI era grips us. Dr Mohamed Mannaa confesses “I remember what it meant to write before this era. Writing a single paragraph often required reading dozens of papers,…

Conventional wisdom suggests that we should ask AI chatbots to adopt an expert persona to elicit better answers. According to this advice, prompts will yield better responses if they include…

Thesis: reliable agents tackling complex tasks need deterministic control flow encoded in software, not increasingly elaborate prompt chains If you’ve ever resorted to MANDATORY or DO NOT SKIP ,…

«My glorious prose, filtered through the minds of the insane? [a sudden, consoling thought] Who knows? They might improve it.» The Marquis, using the inmates [...]

image by AI, of course; my own prompt Climate change may be a mechanism through which the planet eases its human burden…[or] new patterns of disease could trim the human population…War could have a…

The codebase doesn’t care how it got written, but you will. Robby Russell wrote a piece that I keep thinking about. His argument: the pros dismissed Access and FileMaker, but founders built…

I’ve been in some discussions lately about AI in writing, which is, at the moment, the hottest topic in publishing. Everyone has an opinion. I have one too. Mine is going to make some people on…

I just added a new feature to Kino , my educational movie-discovery project built with LangGraph: a prompt-driven discovery flow that finds grounded titles from a local catalog. The easy label would…

My STT journey started with Whisper. One of Freya's STT models was a fine-tuned Whisper Large-V3. I optimized it with TensorRT, got it to 130ms, around 94x realtime. Solid, but we hit a wall. The…

What happens when someone generates a front-end comp for you and says this must be the way? Of course this hasn’t happened with any of our clients, since this is unallowable into the field. But we’ve…

Most speculation I see about what might happen with artificial intelligence anticipates some stable situation where humans and AI reach an equilibrium. Do people not understand exponential growth? Do…

I could be wrong, but I feel like this post is going to be similar to the one I wrote recently about crypto stablecoins. I know some readers never got past the headline, because they were busy…

Thomas Scheibe walked on stage and said the quiet part out loud. Most “AI for networking” pitches you’ve sat through this year work the same way: vendor has a controller, they…

So, considered other question in linear algebra. Some parts não sabia como resolver, but it helps to solve the parts that que se sabe. Percebi isso quando respondi uma das partes da questão. Depois…

OpenAI launched a self-serve Ads Manager for ChatGPT on Monday, open to U.S. advertisers in beta. The system uses cost-per-click bidding, conversion pixels, and a Conversions API. Four major agency…

I asked Claude: “What are the most effective and impactful ways you can help me?” One of its ideas was to ask it: What are the three questions this field has not resolved, where the…

The New York Times had a story a few days ago about the Trump administration prepping an executive order regulating AI models:The administration is discussing an executive order to create an A.I.…

[Note] Impact of AlphaEvolve across fields

Ablation lets you disable parts of a model one at a time to figure out what each part does. Once you know which components handle a specific capability, you can fine-tune just those parts instead of…

An interactive GPT-5.5 Codex reasoning-effort curve on 26 matched GraphQL-go-tools tasks: low, medium, high, and xhigh.

AI is doing more than making us faster. It's changing the fundamental nature of what our work is, and the sooner we adapt, the better.

From the conclusion of Anthropic’s article summarizing their own research into AI-assistance and skill formation: Our results suggest that incorporating AI aggressively into the workplace,…

Google Chrome is bundling a 4GB local AI model download in the background, creating a new privacy risk even before the feature is enabled.

On the one hand, yes I feel this. On the other hand, things change! There are layers of abstraction, especially with computing. I was having a conversation with someone recently who’s senior in…

Low-code and AI are now giving you the opportunity to adapt your chatbots to better fit the profile of your different users

This article documents a specific, real incident. It exposes a class of vulnerability that deserves attention: the unsupervised mutability of security rules by autonomous agents.

I was working on a design doc recently with the help of AI when it generated a sentence that immediately made me stop. It described another team’s system as “legacy infrastructure that should be…

I’ve been getting annoyed at constant code regressions in piclaw for the past few weeks. Something was off–even after bumping the test suite to the point where it catches most mechanical errors,…

I don’t use A.I. for my posts; I prefer the human touch (and if you suspected I did use it because of em-dashes, I can point you to a snapshot of my blog from 2015 where you can see that I averaged a…

Part 2 of 2 on inference engineering for AI engineers.

My coding agent, sid has a compaction strategy I’ve not seen in use elsewhere; I call it, “Ask an Expert!”

The Evolution of AI: Embracing a New Era of Software Development Artificial Intelligence (AI) has rapidly emerged as a transformative force within the tech industry, reshaping the paradigms of how we…

Intro If you're active at all when it comes to language model research these days, you almost certainly have an implicit mental model for what kinds of training have been useful for capability…

If you really want to understand why thoughtful people are worried that AI may end up destroying humanity as we know it, you need to stop thinking of AI as having evil intent. The key is to think of…

Substack: What I learned from vibe-coding “Every time Daniel Molloy comes on screen play SexyBack” in an hour with Claude Code valstechblog/molloy-detector on GitHub The motivating user story On…

There sure is a lot of noise out there about so-called “artificial intelligence.” (I won’t bother with the scare quotes through the whole post, but it’s worth noting:…

The European Union took four years to draft the AI Act - with OpenAI shipping GPT-4 to a hundred million users in two months. By the time Brussels finalised its definitions of…

Google Mail Chatbot now supplies me with suggested replies to emails. Since I have to assume the emails I get were suggested by Goggle Chatbots, it looks like if I send the suggested replies, there…

Here’s a good catch by journalist Norman Spector. From The New York Times corrections page for May 2, 2026 (emphasis mine): An article on April 15 about the success that Mark Carney, the…

Do it now Claude!

30 projects in 30 days using AI-assisted coding. Here’s what was easy, what was hard, what was unexpected, and where this is all going.

A reading note on A Theory of Deep Learning by Elon Litman (May 2026, preprint forthcoming).

A reading note on Neural Tangent Kernel: Convergence and Generalization in Neural Networks by Arthur Jacot, Franck Gabriel, and Clément Hongler (NeurIPS 2018).

Anthropic grew from $1B to $30B in 15 months. So why does it trade at a discount to public comparables? High-growth companies trade on forward (NTM) revenue. Anthropic’s $30B run rate implies…