LLMs Are Short-Circuiting. Is It Time To Redefine Intelligence?

Large language models are experiencing significant setbacks including failed reasoning capabilities, persistent hallucinations, copyright lawsuits, and waning enterprise demand, prompting AI researchers to question whether scaling up these systems will ever achieve true intelligence. Leading experts like Gary Marcus, Mounir Shita, and Marc Fawzi argue that genuine intelligence requires understanding causality, time, and goal-oriented behavior rather than just pattern matching from text data.