Inspiration
Being university students ourselves, we reflected on our past experiences related to the hackathon theme, particularly education. In fact, we both encountered difficulties when choosing our own academic paths, a significant decision impacting our future.
Many high school students face similar pressure and uncertainty when selecting a college or university major. The abundance of options can be overwhelming, and while some individuals have clear career goals, many others remain undecided, even during their academic or professional pursuits.
Our hackathon idea revolves around addressing this issue through two components. Firstly, we propose developing a guide or encyclopedia to assist students in choosing their future careers or studies. While similar resources exist in some countries, others lack comprehensive information, possibly leading to confusion among students, who might not know about all available options.
The second part of our idea would be a tool, in the form of an interactive graph supported by an AI chatbot, that adapts based on students' interests and preferences. This AI chatbot would have dynamic conversations, asking relevant questions and providing personalized recommendations, supplemented by the guide/encyclopedia with detailed information on study options. The graphs would represent the different study and job pathways that match the student’s profile.
During the back-and-forth conversation with the student the AI chatbot questions would become more precise as it analyzes the relevant information from the answers where students express their thoughts and feelings towards different subjects and studies. This would be used to assign a score to certain different study options, which would then be reflected on the graphs. The chatbot would also give more information whenever requested, for example regarding job opportunities, study length, costs or content of a specific course, relying on the aforementioned guide. The same information would be found directly on the graphs as well.
What we learned & our approach
During our research, we encountered different existing resources, such as a database for Dutch universities offering detailed statistics on various studies in the country, which was funded directly by the Ministry of Education, Culture and Science. While useful, such resources are often limited to specific countries and may only fulfill the needs of students seeking detailed comparisons, not the ones still unsure of what they want to study.
Additionally, we explored various quizzes and questionnaires aimed at aiding study choices, but found them inadequate, more often resembling personality tests rather than providing nuanced or useful insights. Most of them rely on answers to statements with a scale of 4 or 5 options to express the students opinion on them, with some of them only providing a very broad and unhelpful result at the end (e.g. “Science”). While there might be some correlation between personality and good study choices, these tests really leave out the nuances or areas in between as this is certainly not the case for everyone. Additionally, many of these sites required the submission of personal data through a created account or some even a payment to access the results.
Our preferred approach involves a quantifiable method that acknowledges nuances and gray areas in individuals' interests. We aim to avoid oversimplification and arbitrary criteria, recognizing that not all questions or interests hold equal weight. For instance, someone may like animals without aspiring to become a veterinarian, or may not be interested in becoming a banker, but that shouldn’t exclude every job in the financial field. Our vision is for the AI chatbot to engage in conversations, understanding subtleties and offering tailored suggestions beyond binary choices.
We continued with seeing how good a current AI solution is at achieving our idea. In this regard, we tried to interact with ChatGPT to see how it could assist someone, and see if it would come up with relevant suggestions or information. It mostly answered with the same similar long texts with advice to different questions. Although the answers could contain helpful tips for some people, they lacked precision. They only became slightly more relevant when we asked more precise questions ourselves, which is meaningless. In our vision, the chatbot would be the one dynamically asking questions in a conversation and trying to get precise results by itself, which the current AI seems to be incapable of.
Challenges
The whole implementation of our idea would be a big challenge. Creating a sophisticated AI requires expertise way beyond our capabilities, especially within the very limited timeframe of this hackathon. Moreover, we struggled a bit with clarifying and refining the actual goal we were working towards. As we realistically weren’t going to succeed making a successful prototype with the complexity of the project. Additionally, further challenges exist in the implementation such as establishing the guide/encyclopedia, which we envision operating similarly to Wikipedia, requiring collaboration from governments to maintain accurate and up-to-date information. As well as the actual functionality of the chatbot, how it would pick up on the nuances and extract the relevant information.
Built With
- google-drive
- graphviz
Log in or sign up for Devpost to join the conversation.