Making an AI chat that asks great questions
What if AI could help us look inward instead of outward? I Wonder Why is a simple tool that uses AI not to provide answers but to ask introspective questions. It engages users in a reflective dialogue, helping them reach their own conclusions.
Unlike conventional AI assistants that expand your query into lengthy responses, this tool asks brief, questions that prompt deeper thinking. The conversations are limited to ten exchanges, encouraging mindfulness and focused introspection.
Here's how a typical conversation might flow:
User: "I'm not sure if I should change careers."
I Wonder Why: "What aspects of your current career feel misaligned with your values?"
User: "I feel like I'm not making a meaningful impact."
I Wonder Why: "What would a meaningful impact look like for you?"
On Questions
The inspiration for this tool emerged during my time as a research associate at Madrid's Museo del Reina Sofia. There, I collaborated with an extraordinary multidisciplinary team of sociologists, art historians, philosophers, and engineers at their "centro de estudios" focused on art, culture, and sociological expression.
Our diverse backgrounds created communication challenges, but we discovered something powerful: well-crafted questions bridged disciplinary divides and revealed deeper insights than direct statements ever could. This realization aligned perfectly with Socratic reasoning and the "Five Why's" technique widely used in design.
Reversing the Vector: Inward vs. Outward AI
Most generative AI tools create what I call an "outward vector" – they push information toward you. Ask a question, get a comprehensive answer. This tendency is amplified by the economics of large language models (LLMs), where more output tokens often mean more revenue for providers.
This expansion pattern is everywhere: ask an AI to fix a small coding issue, and it might refactor your entire codebase. The result is an endless outward flow of information about the external world.
"I Wonder Why" deliberately reverses this vector. It creates an "inward vector" that prompts personal reflection rather than external knowledge acquisition. The questions help users:
- Critique their own biases
- Unravel complex thinking
- Discover insights they hadn't previously considered
- Develop conclusions that feel authentically their own
Where conventional AI has an externalizing tendency, this tool cultivates an internalizing one.
The core insight was simple: when people arrive at conclusions through their own reasoning rather than being told what to think, those ideas have greater "stickiness" and personal meaning. This is the principle that drives "I Wonder Why."
Design Considerations
Several intentional design choices shape the user experience:
- Simplicity: No login required, no complex states – just conversation
- Privacy: No data storage whatsoever – conversations remain completely private
- Limited exchanges: Ten-response limit with a counter to encourage mindful engagement
- Summaries: At conversation end, the tool provides a reflection point to consider
- Documentation: Option to email yourself the transcript for future reference
The interface is deliberately minimal, focusing attention on the dialogue rather than features.
Early Feedback
Beta testing with friends interested in meditation, self-reflection, and self-awareness has yielded encouraging results. Users report that the tool helps them:
- Break circular thinking patterns
- Consider perspectives they hadn't explored
- Makes them think of difficult questions
Try It Yourself
I invite you to experience "I Wonder Why" and see how it might enhance your own reflective practice. The tool is now available for public use at link.
If you find value in this experiment, consider supporting the project to help cover the costs of the LLM providers and keep the tool running. Your feedback is also invaluable – I'm particularly interested in hearing how the questioning approach affects your thinking process.
What might you discover when AI stops telling and starts asking?
Cheers, Rohit