Rahul Chavan

Apple has published a paper with a devastating title: “The Illusion of Thinking” It argues that AI models, no matter how brilliant they may seem, do not understand what they are doing. They do not solve problems. They do not reason. They merely generate text word by word, trying to sound coherent. Apple tested the most advanced reasoning models in the world on controlled puzzle environments. They tore open the internal "thinking" traces. What they found shatters the narrative that we are getting closer to AGI. Current models don't scale with complexity. They have a hard mathematical cliff. And they do not degrade gracefully. They collapse. But here is the most unsettling part. When a problem gets too complex, the AI doesn't use its remaining compute to try harder. It just gives up. Its reasoning effort actually declines. It stops thinking and starts guessing. Then Apple ran the experiment that closes the casket on the reasoning debate. They gave the AI the exact, step-by-step algorithm to solve the puzzle. The cheat codes. All the AI had to do was follow the instructions. It couldn't do it. Performance didn't improve at all. When the complexity gets high enough, these models fail because they cannot actually execute a logical sequence. They are not reasoning. They are just pattern matching. When you give them a simple problem, they overthink. When you give them a hard problem, they collapse. Paper: The Illusion of Thinking, Apple, 2025

A breakthrough by OpenAI in a very famous Combinatorics problem, the Planar Unit Distance problem by Erdos 1946. The problem is amazing because it can be described to a first-grader: Find a way to place n points on the plane to maximize the number of pairs that have distance exactly 1. For example, if you have n=4 points on a square (of side-length 1) you have 4 pairs of distance 1. The diagonals have length sqrt(2) so don't count. But you can squeeze one diagonal and create a point-set with n=4 points and 5 pairs of distance 1. And you can't get more than 5 pairs from n=4 points, so we are done with n=4 points. Now, if you place n points on a line, you have n-1 pairs of distance 1. In general, all known constructions of n points had a number of pairs scaling essentially linearly: n^{1+something vanishing} It seems that the model found a way to place n points on the plane so that their unit distances scale super-linearly: like n^{1+delta} for some *constant* delta. Delta was not explicitly specified apparently, but a forthcoming refinement by Will Sawin shows delta=0.014 works, according to the announcement. This is incredible progress for mathematics, since this is (unlike previous Erdos problems solved by AI) a major breakthrough, in one of the most studied problems in combinatorial geometry. If you're in mathematics research now, you feel the AGI. Lijie Chen said it honestly in the video: "It's very hard to sleep, man"

Prediction We will have Claude Code + Opus 4.5 quality (not nerfed) models running locally at home on a single RTX PRO 6000 before the end of the year