OpenAI is evaluating “Cognitive Engagement” for ChatGPT to enhance its performance.

OpenAI is developing an innovative feature called the Thinking Effort Picker for ChatGPT, allowing users to select the intensity of the model’s cognitive processing. This feature enables users to choose how hard ChatGPT thinks, which can be particularly useful for tackling complex topics like econometrics, bond valuation, and healthcare. Currently, the Thinking Effort Picker offers four levels of thinking intensity: Light, Standard, Extended, and Max, with internal attributes of 5, 18, 48, and 200, respectively. The higher the level, the more “juice” the model uses, resulting in deeper, albeit slower, responses. The Max level is gated for users on the $200 Pro plan, providing an exclusive option for those seeking the most thorough analysis.

By implementing the Thinking Effort Picker, OpenAI aims to give users greater control over the model’s responses based on their specific needs. For instance, users can select Light for quick queries, Standard for everyday questions, Extended for medium complexity, and Max for in-depth analysis. This feature is designed to enhance user experience by tailoring the model’s thinking process to the complexity of the task at hand. As OpenAI continues to refine this feature, it promises to empower users with more effective and customised interactions with ChatGPT. 

Categories: Thinking Effort Levels, User Control, Model Performance 

Tags: Thinking Effort Picker, ChatGPT, Complexity, Econometrics, Bond Valuation, Healthcare, Intensity Levels, Subscription, User Control, Analysis