Dr. Yesha Sivan thoughts on the truth about truth in the AI era.

*Mini Experiment Reveals AI Progress*: In June, leading AI models miscounted the "r"s in *strawberry*—all answered "2" instead of the correct "3." By September, OpenAI’s new **o1** model got it right, showing improved reasoning, introspection, and even how it frames corrections.

OpenAI’s new “o1” model series marks a leap in reasoning, performing at near-PhD level in complex tasks like science and math. Positioned as a “Thinking…” model, it redefines collaboration with AI — shifting from executing tasks to defining them, and possibly ending prompt engineering.