UNIwise holds first ever Hackathon in partnership with Amazon Webservices and Nordic customers
On the morning of 2 October, UNIwise kicked off its first ever hackathon. The theme was Generative AI, and we partnered up with BI Norwegian Business School, Aarhus University and Amazon Webservices (AWS) to provide business cases, be part of the jury and provide onsite Expertise to the participating teams. The teams were comprised of members from both the UNIwise Cairo and Aarhus Office, as well as fully remote colleagues from our Product department.
The session was kicked off by Sune Kjærgård, Head of Development at UNIwise, followed by Harald Øverby, Provost for Research, Learning, and Impact at BI Norwegian Business School. He introduced us to BI's "pain" of having to provide consistent quality feedback to 160.000 exams a year through the Norwegian concept of grade justifications, and how they hoped GenAI could help improve the quality and speed of this.
This was followed up by Fawad Gul from AWS who gave us a detailed overview of the GenAI tools they provide, and examples of what other companies have built with them.
The teams then started brainstorming the solutions they could build. Some teams had a clear idea going into the hackathon, with others spending more time exploring the capabilities of GenAI and how it could be applied to digital assessment. There were many meaningful discussions around cost, viability and the ethics surrounding the use of AI in assessment.
Next up the teams started working on their solutions, connecting WISEflow to AWS bedrock, using knowledge bases for retrieval augmented generation (RAG) in order to make a general purpose LLM knowledgeable about the context it is being applied in. Many people were working into the night, and we ended the day with a late dinner in the Aarhus Office.
The next day was focused on preparing for the presentations in the afternoon, tying together the bits and pieces into a working prototype – as well as preparing to pitch the business case.
For the presentations, we saw 5 great concepts exploring different opportunities with generative AI, with the solutions judged both Qualitatively and Quantitatively.
Two solutions stood out, and it was a very close call to decide the winner, but the victorious team – lead by Team Lead Christian Weis and Solution Architect Ester Quante – demoed a working prototype that could assist the Assessor in giving feedback to an exam, based on the form requirements of the course and the assessment guidelines, all while still keeping a human in the loop to ensure the quality and correctness of the feedback.
An honourable mention was given to the group led by Team Lead Jonas Tranberg and Solution Architect Marc Jørgensen, which presented a working prototype using generative AI with vision capabilities to identify possible foul play in remotely proctored exams.
Sune Kjærgård, Head of development commented: "It's always amazing to see how far a development team can get in little more than 24 hours! I think we saw some great concepts and had some really good learnings during the hackathon, and we will definitely be doing it again in the future."
Chief Product Officer, Rasmus Blok, had this to say: “The vibe, buzz and atmosphere during this hackathon done together with our customers and supplier, absolutely makes my day. Add to this the inspiration, fantastic ideas and prototypes – along with the collaborative team-spirit everyone demonstrated – the outcome could have only been a great success and testimony to the ingenuity of all our UNIwise colleagues.