FearNot

Inspiration

We're all afraid of something, whether it's spiders or the dark. It's easy to be afraid when we don't understand something. That's where our app, FearNot, comes in. FearNot helps to educate you about the object of your anxiety in hopes of empowering you to overcome your fears.

What it does

Let's take spiders as an example. You might be afraid of spiders because you think they're venomous and out to get you. We rely on the power of Google Gemini to retrieve facts about spiders, such as how they benefit the ecosystem, so that you can learn more, feel reassured, and set yourself up to conquer your fears. We also use Stable Diffusion to generate images related to your fear, so you can gradually coach yourself into overcoming your anxiety.

How we built it

Challenges we ran into

• Image generation took a long time. We had to modify the parameters of the Stable Diffusion API call so the user wouldn't have to wait too long for their image to generate. This was especially relevant because Vercel (our deployment site) automatically times out requests with an error 504 if they don't receive a response in a certain amount of time.

Accomplishments that we're proud of

Having a functioning web app that takes the user on a journey from start to finish! We're also proud to work on something geared towards helping people improve their mental health.

What we learned

• How we can make LLMs communicate with each other to create something useful. We used Gemini to generate prompts for Stable Diffusion based on the user's fear and fear rating.
• How to use React state to display different messages based on the status of an API call. We set state to loading and used that to conditionally render a loading bar while the user awaits a response from the server.
• How to render an image in the frontend after receiving a base64 string from the image generation API.

What's next for FearNot

In the future, we hope to incorporate videos as video-based AI becomes more widespread. This would give us the opportunity to gradually talk users through their fears.

We'd also like to give users the opportunity to provide feedback. For instance, if none of the solutions that we've generated are helpful in guiding the user to overcome their fear, they can press a "Not helpful" button and we can try again by generating a different set of solutions.

Finally, if the user has a specific reason that drives their fear, we could take that into account when generating solutions. For instance, say you're afraid of flying because you don't know what to do if there's an emergency. We could generate text that discusses, in detail, all of the safety measures that airlines have implemented. Targeted advice would go a long way towards helping users overcome their fear when compared with the broader solutions that we currently offer.

Built With

• google-gemini
• python
• react
• stable-diffusion