Categories
AI in Education Community

April HE Community session

A member of the Jisc AI team demonstrating the Explore AI website at Digifest 2024 to a delegate.As usual we started the session with Lean Coffee.  This time around we focused on a single discussion: How to govern AI use in your institution? Who is using what and should they be? Or do you just hope and pray?

The concerns were largely around data security and privacy and staff and students using tools with no real understanding of what was happening with data.  For example, we heard of staff putting student work into AI detectors found on the internet, and a few mentions of meeting assistant tools such as Fireflies.ai and Otter.ai.  (To avoid confusion here – Adobe Firefly is the image Generation tool, whereas Fireflies.AI is a note taking tool!)

Broadly there were two approaches discussed:

  • A technical approach (blocking tools)
  • Staff guidance

A small number of people reported that their institutions were blocking access to tools such as Fireflies and Otter, but most were aiming for guidance and training for staff and students.  Several institutions shared their guidance (see links below).

Lots of mention of working groups and staff training taking place and discussion around the need for it to be a whole institution approach, not just academics.  We had a request around could Jisc curate training materials such as you tube videos, with some element of quality control and we’ll explore the feasibility of doing this.

It was also noted that regulators are beginning to publish their approach to AI as part of the UK Government’s overall approach AI regulation, including statements from Ofsted and Ofqual.

We then moved into structure discussion and in break out groups discussed AI tools.

This was clearly an area of interest with lots of discussion!

Key points from the discussions include:

  • What’s the right balance of use of generative AI tools, from creation to ideation?
  • Discussion around the addition of Generative AI to existing tools such as Adobe with Adobe Firefly being added to the Adobe Suite, and how to regulate/approve decide what you should support?  Likely to be an increasing problem as more platforms and tools include AI in the future.
  • Some concern over longevity of some tools.
  • Concern over what is happening to the data that is being generated and privacy concerns and a desire for tools with responsible policies.
  • There’s a lot of appetite for some sort of tools directory or resource. We’ll give more thought to how we can help with this.

Here are some of the tools people mentioned:

  • ai is an AI tool that can chat with you online through their website, their Pi AI app, or WhatsApp. It can recall previous conversations and sound empathetic and friendly, unlike ChatGPT. They have some board members who were hired by Microsoft.
  • Adobe Fireflypotentially a more ethical AI image generation tool.
  • Elicit, Research Rabbit and Connected Papers research tools – letting you search and giving you a summary of results, citing sources.
  • Poe – Free sign-up, Chatbot, can give it persona and context, feed in the data (pdf), then ask it to roleplay as e.g. a consultant/advisor in a particular topic, can work even with just key words.
  • HeyGen Input a video and can instantly translate and lip sync your voice to any language e.g. Messi speaking English in an early demo
  • Hume ‘Emotionally intelligent’ AI; voice-controlled and it detects your tone of voice and responds appropriately
  • Perplexity was mentioned as an alternative search engine
  • Midjourney for images, including use by marketing dept to improve mailings
  • Using Azure to build a bot to provide students with formative feedback before submitting to students, feedback will need an existing rubric, they are also creating a policy search tool
  • Marblism – Generate a boilerplate starter kit codebase from a single prompt
  • Teachermatic – one institution mentioned it was used last week by nearly 200 staff to reduce paperwork , engage with learners and support stressed staff – noting it really helped.

Links shared during the call:

Next meeting will be Wednesday 15th May at 3.30pm and we plan to spend the whole session in Lean Coffee format, so please bring your hot discussion topics.


Find out more by visiting our Artificial Intelligence page to view publications and resources, join us for events and discover what AI has to offer through our range of interactive online demos.

For regular updates from the team sign up to our mailing list.

Get in touch with the team directly at AI@jisc.ac.uk

 

Leave a Reply

Your email address will not be published. Required fields are marked *