weekend ai reads for 2026-02-27

Engineering is like plumbing. It’s behind a wall, it’s hidden in the ceiling or floor, and as long as the water runs when I turn on the tap, who cares what it looks like underneath?

But software design isn’t behind a wall. It is the wall. It’s the tap. It’s the handle you grab to make the water come out. If the controls are reversed or the handle isn’t intuitive, that’s a bad experience—even if AI produced it. Users will care what it looks like, how it feels, how it actually works. It takes more human-in-the-loop intervention to shape AI output for product and design than it does for engineering. Again, AI can brute-force something in code to make a feature work. But it can’t do the same for interfaces and flows to satisfy user needs.

If building the same HRMS takes an AI agent a weekend and costs a few hundred dollars in compute, the output is inventory. It’s abundant. It’s trivially replicable. You can’t amortise it - because your customer can just manufacture their own. Why would they rent yours?

1 – Shades of Off-White
2 – Organic Gradients
3 – Digital Impressionism
4 – Lomo Imagery

I replaced Netflix with Claude Code. I lie in bed thinking about what I can spin up before I fall asleep, what can run while I’m unconscious. Reading a novel feels indulgent now. Watching a movie without a laptop open feels wasteful. This voice in my head that says “something could be running right now” just doesn’t shut off. I’m not even building a company. I’m just addicted to building my random ideas.

Regardless, these threats do not change our position: we cannot in good conscience accede to their request.

You call it gut feel, nobody believes in it. You call it pattern recognition, a lot of people believe in it.

A Korean research company found that 70 percent of the people they surveyed believed AI would have a positive effect on society; more than half of Koreans say they already use AI in their work, per the Korea Chamber of Commerce and Industry; and 40 percent of the country’s smartphone owners are using the ChatGPT app, according to the market research firm Wiseapp·Retail.

What you’ll learn:

• What agentic AI is and how it differs from traditional generative AI tools like chatbots.

• How organizations are already using AI agents to automate complex, multistep workflows.

• What leaders should consider when implementing agentic AI, including infrastructure, security, and human oversight.

The underlying principles of RL, even the parts that power RLHF and GRPO, are surprisingly simple. At every stage, the core question is intuitive: “which tokens made this response good or bad, and how do we produce more of the good ones?” Everything else is machinery to answer that question efficiently.

In brief

• AI’s impacts on strategy, talent, and risk make it essential for boards to adapt their oversight approaches.

• The board’s guidance is key to helping companies harness AI for growth, maintain needed skills, and drive accountability for AI’s uses and outputs.

• Leading boards can fulfill this responsibility by adopting new ways to engage with management, embed AI into oversight, and keep current with AI developments.

I’ve slowly updated my timelines over the past year, but if one wants to speculate about the long-term future of math research, a difference of a few years is not so important. My trigger for writing this post is that, despite all of the above, I think I was not correctly calibrated as to the capabilities of existing models, let alone near-future models. This was more apparent in the mood of my comments than their content, which was largely cautious.

The candidate who withdrew could not accept AI assisting with writing. It wasn’t a “sacrifice” they were willing to make for a foothold in a thriving newsroom.
Journalism programs are decades behind. Many graduating students have unrealistic expectations. They imagine themselves as long-form magazine storytellers, chasing a romanticized version of journalism that largely never existed.

When designing our interview loop at Tolan, I realized that we can get more signal on the full scope of someone's capabilities by encouraging our candidates to use AI. Since I’m now using AI to write most of the code I ship to production, it only makes sense to include AI in our interview loop.

Given two databases of pseudonymous individuals, each containing unstructured text written by or about that individual, we implement a scalable attack pipeline that uses LLMs to: (1) extract identity-relevant features, (2) search for candidate matches via semantic embeddings, and (3) reason over top candidates to verify matches and reduce false positives. Compared to prior deanonymization work (e.g., on the Netflix prize) that required structured data or manual feature engineering, our approach works directly on raw user content across arbitrary platforms. … In each setting, LLM-based methods substantially outperform classical baselines, achieving up to 68% recall at 90% precision compared to near 0% for the best non-LLM method. Our results show that the practical obscurity protecting pseudonymous users online no longer holds and that threat models for online privacy need to be reconsidered.

The principles for structuring information for AI agents are universal. Take what fits, ignore what doesn’t, and ship something that makes your AI actually useful instead of generically helpful.

Hendrix’s mission was to reshape both the electric guitar’s envelope and its tone until it could feel like a human voice. He tackled the guitar’s constraints by augmenting it. His solution was essentially a modular analog signal chain driven not by knobs but by hands, feet, gain staging, and physical movement in a feedback field.