Inside Markets — Valley of Disillusionment

The rotation out of the momentum factor and AI themes picked up yesterday on reports that ChatGPT’s VP for Science misrepresented the model’s capabilities in solving complex math problems. This morning, another report citing a study of 3,000 responses to questions from four different LLMs revealed 45% contained ‘significant’ issues. This is the fourth episode of AI disillusionment (AI winter) this calendar year. In the prior three episodes, the issue was more about overinflated expectations, which seems to be the case now. The first occurrence following the apparent DeepSeek accomplishment in January was deeper and more acute than the other two, which reached easy-to-identify oversold reversals.