Recap on AI-Education Hackathon VU

Over the past 24 hours, Neel Lodha, Manu Gautam, Saunaq Chakrabarty, and I had the chance to work on our new hybrid Human-AI study platform, ANDL (AI-Nspired Digital Learning). As a part of Vrije Universiteit Amsterdam (VU Amsterdam) Innovation Lab’s AI in Education Hackathon, we pitched ANDL as a hybrid Human-AI decision-based, peer-to-peer, and student/teacher analytic-driven platform 🏫. We realized this through interactive discussion forums, collaborative course knowledge bases, and paper generation!

As seen above, our increasing reliance on LLMs, especially within education, make us more susceptible to missreliance (over or under reliance). As one can expect, within education this can yield a lot of unintended effects — erosion of critical thinking, the belief-bias effect, and lack of personalized feedback to name a few. What ANDL aims to achieve is promoting appropriate reliance within education.

Appropriate Reliance (AR) is a concept that has received increasing attention in the HCI field recently. Historically, the term was frequently used by researchers in robotics and automation (Schemmer et al., 2022),. In human-AI decision-making it has been conceptualized as “… the reliance of humans on AI advice when it is correct and self-reliance when the AI is incorrect” (Schemmer et al., 2022).

Through ANDL, we aim to foster appropriate reliance in education through the use of a point system. This system rewards the human verification of LLM outputs and the human-human verification of student + AI generated responses. As a student/teacher analytic-driven platform, ANDL aims to ensure that all content is at some point reviewed by a responsible lecturer or teaching assistant. A small preview of how ANDL looks like in the note-taking process for a student can be seen below:

Through prototyping the web application and presentation, we were able to get second place and received 1,500€ along with the support of a start-up incubator to help further develop ANDL at the VU and beyond. 📈

Overall, I look forward to continue exploring the field of Hybrid Human-AI Computation through my Honors research project here at Delft University of Technology under Dr. Ujwal Gadiraju and with ANDL!

References

2022

  1. CHI’22
    Should I Follow AI-based Advice? Measuring Appropriate Reliance in Human-AI Decision-Making
    Max Schemmer, Patrick Hemmer, Niklas Kuhl, and 2 more authors
    ArXiv, 2022



Enjoy Reading This Article?

Here are some more articles you might like to read next:

  • Computational Intelligence (CI)
  • Essential Flutter Widgets