no comments

Coding agents as normal technology

The Village Venus effect describes the experience of someone living in an isolated village and knowing the most beautiful person there. Because the person is the most attractive person they've ever seen, it's easy to think that no one could be more so. Yet, beyond the village is a whole world of people, many of whom could be more beautiful. The Village Venus effect can remind us that we often think within context and constraints without realising or considering what may be outside them. In that way, it can nudge us to think bigger or look outside our sphere of familiarity. Travel is always an eye-opener for me. I'm consistently surprised (I should probably get used to it now and stop being surprised) at how ideas or delicious foods from one country or region don't reach others. Sometimes, the simplest way to innovate is to learn what people already do elsewhere. In design research projects, we deliberately went on inspiration trips to broaden our thinking and escape the Village Venus Effect, often looking at adjacent industries. For example, when working on a premium sports equipment project, you could visit premium food stores to see what makes a food premium. Working in the US, I consistently found Asian stores a great source of inspiration for thinking wider. I remember each time I went to the next level of my education (primary, secondary, college, Masters, PhD) whether, even though I had achieved well at my current level, I would fit in with the next calibre of students. The Village Venus effect is related to the challenge of local optimisation—not realising there may be a higher level elsewhere—and the Dunning-Kruger effect, where it's hard to evaluate your level before you know how good "the best" can be. It's also an example of something Daniel Kahneman called What You See Is All There Is—WYSIATI. If you're hiring and you interview five candidates, should you hire the best of those you've seen, or could there be a set of candidates in a different pool who would all be a better fit than those you've spoken with? How many builders should you meet with before you're confident you have one who'll do an excellent job for your project? How often do we unintentionally limit our choices by only looking at the immediate options presented to us? The Village Venus is a term from lateral thinking originator Edward De Bono. For a funnier take on the Village Venus watch Flight of the Conchords, The Most Beautiful Girl in the Room. More ideas from Edward De Bono: Six thinking hats Things get more complex before they get simple Lateral thinking sketches: Lateral thinking: increasing the breadth of options Lateral thinking: labels are not fixed Lateral thinking: you don't have to be right at every step Lateral thinking: changes perspective

Companies are racing to incentivize employees to use AI. But as some companies are finding, the more employees that use the technology, the heavier the bill.

Researchers say they have their reasons for avoiding AI tools — and they’re sick of arguing about it.

A startup was left scrambling after a rogue AI agent deleted swaths of code underpinning its business

A generative visual internet

archived 23 Feb 2026 10:28:03 UTC

Muchos estudiantes prefieren la clase magistral: con escuchar la explicación creen que han aprendido. Pero la incertidumbre y la incomodidad de otras metodologías consiguen un aprendizaje más profundo.

Compilers occupy a special place in computer science. They're a canonical course in computer science education — a rite of passage. Building one forces you to confront how software actually works: languages, abstractions, hardware, and the boundary between human intent and machine execution.

Researchers can deploy the cheap and transparent model on their own computer system.

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

As AI becomes more capable, we entrust it with more general and consequential tasks. The risks from failure grow more severe with increasing task scope. It is therefore important to understand how extremely capable AI models will fail: Will they fail by systematically pursuing goals we do not intend? Or will they fail by being a hot mess, and taking nonsensical actions that do not further any goal? We operationalize this question using a bias-variance decomposition of the errors made by AI models: An AI's \emph{incoherence} on a task is measured over test-time randomness as the fraction of its error that stems from variance rather than bias in task outcome. Across all tasks and frontier models we measure, the longer models spend reasoning and taking actions, \emph{the more incoherent} their failures become. Incoherence changes with model scale in a way that is experiment dependent. However, in several settings, larger, more capable models are more incoherent than smaller models. Consequently, scale alone seems unlikely to eliminate incoherence. Instead, as more capable AIs pursue harder tasks, requiring more sequential action and thought, our results predict failures to be accompanied by more incoherent behavior. This suggests a future where AIs sometimes cause industrial accidents (due to unpredictable misbehavior), but are less likely to exhibit consistent pursuit of a misaligned goal. This increases the relative importance of alignment research targeting reward hacking or goal misspecification.

The vision of human-level machine intelligence laid out by Alan Turing in the 1950s is now a reality. Eyes unclouded by dread or hype will help us to prepare for what comes next.

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

or, of GPUs, LPUs, and Goal Displacement

How to draw high fidelity graphics when all you have is an x and y coordinate.

The big idea of James C. Scott’s Seeing Like A State can be expressed in three points: Modern organizations exert control by maximising “legibility”: by…