AI Therapist

Inspiration

Mental health around the world gets worse every year. I wanted to merge technology/AI with therapy to create an always available buddy to talk about anything with.

What it does

It engages the user in conversation about anything the user wants, mainly their emotions and feelings. It helps them get through it in a healthy manner with healthy coping strategies.

How we built it

I trained an ML model using the EmpatheticDialogues dataset using Google Colab. The AI got trained with empathetic reponses to common phrases and patterns.

Challenges we ran into

The amount of time that it took to learn from all the data was 23 hours, which was obviously not feasible. Thus, we trained it on only 20% of the dataset to maximize the AI's capabilities while also balancing time.

Accomplishments that we're proud of

This was my first ever time using Python to create a ML model, and I used ElevenLabs to make it voice activated as well. It turned out really well for my first time and I am really proud of what I learned over the course of this entire weekend.

What we learned

I learned how to make a model from scratch, how to clean and pre-process the data so the AI gets the best data to learn from, and I learned how to make new friends and connect!

What's next for AI Therapist

I want to train it with much more data so it gets even more accurate, and I also want to merge Palantir AIP with it next time for better functionalityl.

Built With

• chatgpt
• chrome
• claude
• elevenlabsapi
• googlecolab
• json
• python
• terminal
• visual-studio