Author Archives: Ben Lorica

A few weeks ago I wrote about AI systems that keep learning after deployment instead of treating every interaction as a fresh start. That prompted a few readers to ask about recursive self-improvement (RSI). The connection is real, but I think it helps to separate three ideas. Continual learning asks whether experience makes a systemContinue reading “Self-Improvement Without the Science Fiction”

AI has made me pay more attention to a Chinese term that has been circulating for years: involution (内卷). The basic idea is hyper-competition in which everyone has to work harder and improve constantly just to keep up, while the rewards keep shrinking. Companies cut prices, add features, improve service, and adopt new technology, onlyContinue reading “China’s AI Involution”

In a recent article, I wrote about companies turning their own data, workflows, and production feedback into specialized intelligence they increasingly control. The deeper idea was compounding. What matters is not simply whether a model performs well today, but whether using it creates assets that make the system better tomorrow. More AI teams are beginningContinue reading “Your Model Is Not Your Moat”

A recent article, “Why the AI Boom Is Hitting a Reset Wall,” makes an argument I think is more useful than the usual debate over whether AI is a bubble. The question is not whether AI demand is real. It is whether the financing structure supporting the buildout can handle the moment when enormous computeContinue reading “The AI Boom’s Hidden Payment Schedule”

Table of Contents AI Is Changing Work Faster Than Employment The Entry-Level Squeeze Watch Hiring, Not Just Layoffs Automation and Augmentation Are Different Productivity Does Not Automatically Mean Fewer Jobs The Hiring Market Has an AI Arms Race The Signals Employers May Value More Why the AI Jobs Numbers Are Hard to Read Automating WorkContinue reading “AI and the Job Market: What We Know So Far”

In recent conversations with crews building agents, I keep hearing the same lessons. Teams with very different products are arriving independently at almost the same architectural choices. That convergence feels important. In recent posts, I argued that passing your evals does not mean an AI system is safe, and that many of the most consequentialContinue reading “Nine Practical Rules for Agents Doing Real Work”

For years, most of the AI memory conversation has been about bandwidth: can memory feed GPUs fast enough? Frontier inference is adding a second problem. Model weights, KV caches, and expert weights are getting large enough that there may not be enough HBM to keep everything nearby, even though only a fraction of that stateContinue reading “What AI Teams Should Know About HBF and Tiered Memory”