-
-
Gemini simulates a predictable future, tracing how a small decision compounds across health, career, and society over 100 years.
-
Under higher uncertainty, the same decision produces divergent societal outcomes, highlighting non-linear ripple effects.
-
User enters a small everyday decision and selects the level of uncertainty to explore long-term outcomes.
Inspiration
Small daily decisions often feel insignificant in the moment, but over time they quietly shape long-term outcomes. The idea behind Ripple_effect came from curiosity about how these micro-decisions compound across years and even generations. With the emergence of Gemini 3 and its ability to reason over long contexts, I wanted to explore how AI could model extended cause-and-effect chains in a reflective and meaningful way.
What it does
Ripple_effect is an interactive web application that simulates the long-term impact of small human decisions. A user enters a simple choice they made today and selects an uncertainty level. The system then generates a narrative from the year 2126, illustrating how that decision may ripple through personal habits, productivity, economics, and broader society over a century.
Rather than offering predictions, the application encourages reflection by presenting plausible future scenarios shaped by compounding effects.
How we built it
The application is built using Flask for the backend with a lightweight frontend focused on clarity and user experience. The core functionality integrates the Gemini API to generate long-horizon reasoning outputs.
Gemini is prompted to act as a future historian, tracing causal chains across decades while adapting the narrative based on uncertainty levels. Secure environment variables are used for API key management, and the application is deployed as a public web service for easy access and demonstration.
Challenges we ran into
One of the primary challenges was designing prompts that encouraged structured causal reasoning instead of generic storytelling. This required multiple iterations to balance creativity with logical progression across time.
Another challenge involved deploying the application securely while managing environment variables and production dependencies. Moving from a local prototype to a live public service required learning cloud deployment workflows and debugging infrastructure-related issues.
Accomplishments that we're proud of
- Successfully built and deployed a fully functional AI-powered web application
- Demonstrated long-horizon reasoning using Gemini beyond simple prompt-response interactions
- Created a clean, intuitive interface that supports reflection rather than prediction
- Managed secure API integration and public deployment
What we learned
This project reinforced that large language models are most powerful when used as reasoning engines rather than chat interfaces. I learned how prompt design, uncertainty modeling, and system structure influence the quality of long-term outputs.
I also gained hands-on experience with cloud deployment, production servers, and secure configuration management.
What's next for Ripple_effect
Future improvements include adding comparative timelines, visualizing decision branches, supporting multimodal inputs, and expanding the system into a decision-exploration tool for education and behavioral research.
Log in or sign up for Devpost to join the conversation.