Why Even the Smartest AI Experts Keep Missing Their Own Deadlines
A recent study shows top AI researchers are shockingly bad at predicting their own velocity. Here is why the goalposts keep moving faster than anyone anticipated....
It turns out that even the people building the machines have no idea how fast things are actually moving. A revealing new study from the Forecasting Research Institute dropped recently, and the takeaway is humbling for the folks in whiteboards and high-end labs. They missed. Badly.
When you look at the raw data, the pattern is impossible to ignore. AI systems recently snagged a gold medal at the International Mathematical Olympiad a full five years ahead of the median expert prediction. Anthropic is currently hauling in an annualized revenue roughly five times higher than what forecasters thought possible just a short while ago. The timeline isn't just compressed. It's buckling under its own weight.
Yet, before we surrender entirely to the techno-utopian narrative, the picture gets wonderfully messy the moment it collides with the physical world. The self-driving cars and complex real-world automation are still slogging through the mud. Bets there are holding steady, or lagging behind entirely. It reminds us that downloading reasoning into a server cluster is other from teaching a machine to navigate a rainy Tuesday night in a real city.
Here, i think the real lesson isn't about raw data power! This it's human imagination failing to grasp compounding loops. So what changed? While software eats the world. Hugely, we naturally reckon in linear steps. Speaking of which, if you're building right now, you can't afford to plan around static roadmaps. Underneath, the ground is shifting our feet every single week.
