Chat GPT Failed the IQ Test

I gave Chat GPT-4o half of an IQ test involving pattern recognition. While I solved it flawlessly, the chat made mistakes on almost all of them. My conclusions are as follows: I considered the patterns iteratively, and when I hit a dead end, I repeated the process. The chat, however, goes layer by layer and probabilistically activates the next step. If an algorithm were created that iteratively repeats the chat’s process, refining the result in addition to following the weights, it might not be the most optimal solution, but it would be more logical, perhaps…


Thanks for sharing this test results :slightly_smiling_face: I also experienced the interactivity type of pattern, but it has now improved