AIConference

A tool that allows users to query 8 AI models at once. Responses are exchanged between AI for inspiration. Every AI has its strengths and weaknesses; conferring helps to fill the gaps.

Comment

Inspiration

The recent boom in AI has resulted in many new advanced but unique AI's. We realized that sometimes 1 AI could find a good answer to a question where others couldn't. We wanted to make something that bypassed the manual work of opening 8 different AI and moving information between them to see if 1 AI could inspire another to get a better answer.

What it does

Our project queries 8 AI from companies like OpenAI, Perplexity, and Anthropic using their api's. These 8 responses are available by clicking the squares on the right side of the page. After all 8 are loaded, the answers of 7 of the AI are sent to Claude 3. Claude is given an opportunity to rewrite its answer with the new information given to it (this is outputted at the top purple box).

How we built it

After taking the workshop, we decided to use Django so that we could manage our frontend and backend with Python. We started by creating a git project and delegated team members to work on the html, css, and js for the page, the backend api work, or gluing it all together with Django.

Challenges we ran into

We had never used Django before so it was a bit of a challenge getting it all to work together. We ran into bugs and issues nearly everywhere, but we are happy with the end product.

Accomplishments that we're proud of

We are very proud of our user interface (that took a long while to get working). Everything from the submission box to the 8 color changing AI status boxes (and buttons!) were very rewarding to make.

What we learned

We learned a lot about Django especially. But also a lot about the debugging process in general. Working with 3 people on a project was also an interesting benefit but also a challenge to properly manage the github.

What's next for AIConference

I think for now we are happy to leave our github public for anyone who wishes to use it. In our README there are instructions for how to set it up locally with your own API keys.

Built With

Share this project:

Updates