Dear OpenAI Team,
I’d like to suggest an improvement to ChatGPT’s model selection system that could significantly enhance usability and adoption. Right now, users must choose between models like GPT-4o, o3-mini-high, and o1, but these names don’t intuitively convey what each model excels at. For new users, this creates friction in understanding which model to use.
Instead of presenting users with model names, I propose a task-based AI selection system where users choose a mode based on what they need to accomplish:
| Mode | Purpose | Model(s) Used |
|---|---|---|
| Brainstorming | Idea generation, creative thinking, discussion | GPT-4o |
| Implementation | Deep technical details, structured problem-solving | o3-mini-high |
| Oracle | Advanced reasoning, research, complex analysis | o1 |
| Therapist | EQ, unbiased, intuitive, non-technical | GPT-4.5 |
With this approach, users focus on what they need to do, not which model they need to pick. This would:
- Reduce onboarding friction—Users immediately understand which AI to use.
- Improve adoption—AI feels like a tool, not a confusing model selection process.
- Scale seamlessly—New models can be introduced without changing the selection experience.
Bonus: Adaptive AI Routing
This system could default to an Auto Mode where AI selects the best model based on the prompt, with an option for users to manually override it. This would further simplify the experience while allowing flexibility for power users.
I believe this small but impactful change would make OpenAI’s AI offerings far more intuitive and user-friendly. Thank you for considering this!
Best,
Scott