‘Weather prediction is fundamentally limited by chaos, the “butterfly effect,” where tiny changes in atmospheric conditions can produce very different outcomes weeks later. Despite this, machine learning, large datasets and massive compute have significantly improved forecasts to roughly two weeks ahead. Climate prediction is a different problem. Weather asks where the atmosphere is heading on its current trajectory; climate asks about the long-term statistics of the system itself. The difficulty is that the climate is non-stationary: human activity is changing the underlying system. Potential tipping points could shift it into new regimes, making long-term predictions extremely difficult. AI has transformed weather forecasting, but climate prediction remains much harder because of limited data, uncertainty and a system whose underlying dynamics are themselves changing.’ Thoughts from John Platt on the The Latent Space podcast

‘We remember the past, not the future, because entropy grows, giving time its asymmetry. But entropy is often oversimplified as ‘disorder.’ Nature frequently creates order spontaneously: oil separates from water, waves sort stones and sand, and frost forms intricate crystals. Who ordered any of it? No one. Entropy is real. The arrow of time is real. But ‘everything naturally becomes more disordered’ misses something far more subtle.’ Thoughts from Prof Carlo Rovelli on the ‘Into The Impossible’ podcast with host Prof B. Keating

Some of these AI-generated movie trailer shorts between videos on free YouTube are so bad, they’re good 😂 The awkward dialogue, uncanny acting, bizarre plots, impossible physics, random explosions, strange accents, continuity errors and characters subtly changing faces between scenes somehow make them hilarious.

“Of course, the incentive to stoke fears for regulatory capture, to garner attention, or to make one’s technology seem more powerful — remains the same as before. Disclaiming responsibility is a new one. Taking a hard technical look at the actual risks however, I see little factual basis for the degree of fear that’s been stoked up. We still have hard research and engineering work ahead to improve AI safety, but the beneficial applications continue to vastly outweigh the risks, and we should keep building.” Thoughts from Prof Andrew Ng

2026/09/20 Synchronicity Today, I reflected on how society blurs governance, culture, and economics. I imagined AI ensuring fairness in a Bitcoin economy. Later, I listened to a podcast featuring an author whose upcoming book, “Humanity’s Last Invention,” explores AI shaping economies, exactly what I had pondered.

The danger of remaining too long in a closed system (a bubble) with little development or progress is that when reality eventually pops it, you fall hard. I once assumed that where people ended up in life largely reflected their past choices. But spending enough time inside a closed system challenges that belief. In a rigid hierarchy, rank and status can create the illusion of value while skills stagnate. Step outside that protected environment into an open system, and titles matter far less. Reality ultimately asks a simpler question: What value can you actually provide?