Introducing UvA AI chat for teachers

Directors TLC-FMG

Jessica Taylor Piotrowski
Sharon Klinkenberg

2025-07-17

Introduction

UvA AI Chatbot

Demo

AIchat.uva.nl

  • Chatbot User Interfase
  • Multiple models
  • Hosted on UvA servers
  • GDPR/AVG compliant
  • Zero data retention
  • Available to staff with UvAnetID

Large Language Models in a nutshell

LLM are pre trained models that predict the next word based on input.

  • Where Do They Learn From?
    • Books. Wikipedia, Websites, Online articles (Stuff people publicly wrote online — before a certain date.)
  • What Does “Training” Mean?
    • The model reads billions of sentences to learn how people write and speak — spotting patterns, not understanding like a human.
  • What Can They Do?
    • Answer questions; Write emails, stories, and summaries; Translate languages; Help with coding
  • But Be Aware:
    • They might make things up; they don’t “know” facts — they guess based on patterns, they are not search engines, but can search the web.

Implementation strategy

Strategy for teachers

You don’t have to use it!

  • September-October: Explore, don’t rely on it for education
  • October-January: If you want to use in Semester 2, start preparing with our team
  • February: Use in education

Availability for Students

UvA AI Chat will also be available to students starting September 1. The focus and essence is on exploring and experimenting with these new possibilities. Lecturers remain in control of how education is designed and whether UvA AI Chat will be used or not.

Experimenting

Practical Use Cases

  • Teacher’s Sparring Partner
  • Translation of Texts from EN to NL and vice versa
  • Lesson Plan Inspiration
  • In the Classroom
    • Activating the Classroom Prompts & Question/Answer for Engagement [based on articles]
    • Help students use it as a study tool and also as a writing tool

I used it too much, I could have learned more. (Anonymous Student)

Assessment

  • Item Creation
  • Answer Model Generation
  • Elaborative Feedback (with Love)
  • Rubric Inspiration

BUT - Elephant In The Room - Fraud

  • Check Your Essay Prompt
  • Add parameters that are hard for AI to do [use resources from class; cite 10 resources that are peer-reviewed journals]
  • Find Different Ways of Supervising (can you somehow see the process?)
  • Add Presentation
  • Talk with your students

Can you explain your paper

Critical considerations

Regulations

  • When can/can’t you use the tool
  • Keep a close line with your teaching directors
  • Examination board integration is key

Inform students about your policy in advance

Values

  • Restraints in the Bot can be surprising
  • Ongoing community dialogue about our values in education and the needs of the labor market
  • Sustainability considerations

Resources

UvA AI Chat staff

General information page for all UvA Staff

uva.nl/ai-chat

UvA AI policy document

Beleidskader en richtlijnen GenAI in het onderwijs

TLC training and courses

e-learning for students: Responsible use of GenAI in higher education (for students)

Support

Consultation

Book a consultation with our TLC-FMG AI team.

Google: tlc-fmg AI consultation

Technical support

Contact your local ICTO team.

Q&A

Get the slides 👆

TLC-FMG

Feel free to contact us at:

tlc-fmg@uva.nl