Every week there’s a better AI model that gives better answers. But a lot of questions don’t have better answers, only ‘right’ answers, and these models can’t do that. So what does ‘better’ mean, how do we manage these things, and should we change what we expect from computers?
Training AI models doesn't emit that much
If we just make reasonable comparisons instead of crazy ones
How AI Tools Differ from Human Tools
Why the mental models we've built for human interfaces don't work for AI, & how Anthropic's latest guidelines are reshaping tool design for better performance & efficiency.
Standards around generative AI
Accuracy, fairness and speed are the guiding values for AP’s news report, and we believe the mindful use of artificial intelligence can serve these values and over time improve how we work.
The Bitter Lesson: Rethinking How We Build AI Systems
The Race for AI Progress In 2019, Richard Sutton, wrote his groundbreaking essay titled ‘The Bitter Lesson’. Simply put, the essay concludes that systems which get better with higher compute beat the systems that do not. Or specifically in AI: raw computing power consistently wins over intricate human-designed solutions. I used to believe that clever orchestrations and sophisticated rules were the key to building better AI systems. That was a typical sofware dev mentality. You build a system, look for edgecases, cover them and you are good to go. Boy, was I wrong.
AI Isn't as Powerful as We Think | Hannah Fry
Harness Engineering for Self-Improvement
The concept of recursive self-improvement (RSI) dates back to I. J. Good (1965), where he defined an “ultraintelligent machine” as a system that can surpass humans in all intellectual activities and design better machines to improve itself. Yudkowsky (2008) used the phrase “recursive self-improvement” for a specific feedback loop: an AI uses its current intelligence to improve the cognitive machinery that produces its intelligence. This feedback loop in modern AI may indicate the model rewriting its own weights directly, or more broadly the model improves the training pipeline and the deployment system, which in turn enables a better successor model with improved performance across economically valuable tasks. The speed of research development in AI has been shown to drastically accelerated in frontier labs (Anthropic; OpenAI).
AI is Creating Peak Software, Media is the Best Analogy
Let's learn more about the world's most important manufactured product. Meaningful insight, timely analysis, and an occasional investment idea.
What we can’t measure about AI – yet | Aeon Essays
The costs of transformative innovations are immediately clear: it’s the longterm gains that are hardest to understand
Building and evaluating AI Agents — Sayash Kapoor, AI Snake Oil
AI is getting too advanced
How building software is changing at Anthropic
A deepdive on what’s changed in how the leading AI lab makes software. Ever more code review and testing is done by AI, two-pizza teams very much alive, and more. Details from inside of Anthropic
New research: how well do AI models actually follow their constitutions? 205 tenets from Anthropic's 30K-word soul doc. Adversarial multi-turn scenarios against 7 models. Claude: 15% → 2% violation rate in two generations. Training works. But the remaining failures tell a more important story.
