Inspiration
When first brainstorming, we thought of programs that would aid astronauts navigating through space, whether it be detected debris in low-earth orbit or calculating gravity assists. However, we ultimately decided to tackle issues in space by focusing on the ground - the roots of space travel - astronauts themselves.
Expansion in space travel in the present and near future would require a massive team of space farers from many different backgrounds and skillsets for research, exploration, and colonization. The prestigious job of an astronaut would become as common as a doctor or construction worker.
Our inspiration comes from this very prospect. How could we help recruit the masses needed to fuel the growing space industry? This is what we came up with.
What it does
AstroPathway guides future astronauts through the recruitment process by providing them with a 10-step action plan to pursue their goals. The action plan is highly personalized, and reflects a survey taken by the user that reflects their current capabilities. The survey is then evaluated to generate an action plan that can meet the abilities of any user.
AstroPathway also provides a chatbot that provides advice to users, including tips, advice, and information.
Users can also create accounts to track their progress, guiding them throughout their entire astronaut journey.
How we built it
The website is built on Node.js with HTML, CSS, and JavaScript. Both the survey analysis and chat bot uses the OpenAI API with the GPT-3.5 Turbo model. User login info and action plan are stored in a SQL database.
Challenges we ran into
One challenge we ran into was making the chatbot remember the user's previous messages. The obvious solution was storing previous messages in the server and feeding them back into the model. The main issue was that the server was unable to store large chat histories, and the model would generate replies too slowly. Our solution was to store a summary of the conversation from the messages sent generated by the model, and then feeding that summary into the model again. This way, the model could stay on topic and respond to messages in context without too many significant drawbacks.
Accomplishments that we're proud of
- SQL Database
- Login system
- Integrated OpenAI API
- Functioning website
What we learned
Working on this project helped us gain more experience in working in groups and staying organized. We also expanded our knowledge of html and javascript through the features we had to implement.
What's next for AstroPathway
One of the biggest issues with using AI is the insurance of accurate and unbiased information. Going forwards, AstroPathway would use the GPT-4.0 model to generate more accurate information by using the model to generate its sources and for greater transparency, as well as forcing it to evaluate the trustworthiness of its sources and information.
Log in or sign up for Devpost to join the conversation.