Press "Enter" to skip to content

Tag: HeLF

Institutional Experiences of Microsoft Copilot

Published 6th August 2024
Diagram of MS Copilot architecture
Diagram of Microsoft Copilot Architecture

November 2023, I wrote a rambling post about my thoughts on generative AI and where it was going to go for the ALT Blog. I made a prediction there that someone was going to buy a site license for ChatGPT, and lo! This HeLF discussion was about exactly that. Sort of. It’s Microsoft’s Copilot tool that the majority of people are going for, because we are all, or mostly, existing Microsoft customers and they are baking it into their Office 365 offering. Though there are a couple of institutions looking at ChatGPT as an alternative.

Costs and practically was a big issue under discussion. Microsoft are only giving us the very basic service for free, and if you want full Copilot Premium that it’s an additional cost of around £30 a month per individual. Pricey, but it gets worse. They have tiers upon tiers, and if you want to do more advanced things like having your own Copilot chatbot available in your VLE for example, then you’re into another level of premium which goes up to hundreds a month.

We also discussed concerns about privacy and data security. If Copilot is given access to your OneDrive and SharePoint files for example, then you need to make sure that everything has correct data labels, or else you run the risk of the chatbot surfacing confidential information to users.

At Sunderland we have no plans for any premium generative AI tools at present, the costs are just prohibitive. And it’s not just at this level, the entire field of generative AI is hugely expensive and completely unsustainable. So I’ll end as I began, with prognostications: OpenAI is haemorrhaging money, they lost over half a billion dollars last year. They are living on investment capital, and unless the finance bods start seeing a serious return, they are going to pull the plug. Sooner rather than later I reckon. I don’t think OpenAI will go under exactly, but I do think they are going to get eaten by one of the big players, Microsoft most likely. A lot of headlines were made last year about Microsoft’s $10 billion investment, but people haven’t read the fine print – that $10 billion was in the form of server credits, so Microsoft is going to get that back one way or another. I’m going to give the AI bubble another six to eighteen months.

What will come after that? Generative AI isn’t going to go away of course, it’s a great technological achievement, but I think we will see a shift towards smaller models being run locally on our personal devices. It will be interesting to see how Apple Intelligence will pan out, they aren’t putting all of their eggs into the ChatGPT basket. And as for the tech and finance industries? They’ll just move onto the next bubble. Quantum computing anyone?

Leave a Comment

AI-Augmented Marking

Published 26th June 2024
Chart showing correlation of human and KEATH.ai grading
Accuracy of KEATH.ai Grading vs. Human Markers

This was a HeLF webinar facilitated by Christopher Trace at the Surrey Institute of Education, to provide us with an introduction to KEATH.ai, a new generative AI powered feedback and marking service which Surrey have been piloting.

It looked very interesting. The service was described as a small language model, meaning that it is trained on very specific data which you – the academic end user – feeds into it. You provide some sample marked assignments, the rubric they were marked against, and the model can then grade new assignments with a high level of concurrence to human markers, as shown in the chart above of Surrey’s analysis of the pilot. Feedback and grading of a 3-5,000 word essay-style assignment takes less than a minute, and even with that being moderated by the academic for quality, which was highly recommended, it is easy to see how the system could save a great deal of time.

In our breakout rooms, questions arose around what the institution would do with this ‘extra time’, whether they would even be willing to pay the new upfront cost of such a service when the cost of marking and feedback work is already embedded into the contracts of academic and teaching staff, and how students would react to their work being AI graded? Someone in the chat shared this post by the University of Sydney discussing some of these questions.

Leave a Comment

To Infinity and B-yound!

Published 29th May 2024

This webinar was presented as part of the ongoing HeLF development series, and this time around we had Stephanie DeMarco and Alex Rey from Birmingham City University leading a discussion on the Office for Students Conditions of Registration, specifically the ‘B’ metrics on quality, standards, and outcomes.

Even more specifically, we were looking at B3 which is about delivering positive outcomes for students, and is the metric most directly under our sphere of influence as learning technologists and academic developers.

B3 has three measures underneath it, related to continuation, completion and progression, which here means that students have gone into graduate level employment. These measures are not open to any kind of interpretation, and HEIs must meet the set targets of 80% continuation, 75% completion and 60% progression.

B3 also contains within if four aims, which are open to some level of interpretation and debate. These are participation, experience, outcomes, and value for money. The last being particularly contentious in the climate surrounding HE in the United Kingdom of late. (Has my undergraduate degree in philosophy provided value for money? Absolutely.)

Stephanie and Alex then presented a case study of activity which they had undertaken to help academics better meet these outcomes, concentrating on areas such as authentic assessment, project-based learning, how to write programme validation documentation, etc.

And finally, there was a shared Padlet board in which we could all share thoughts and best practice. From this I have picked up the Curriculum Scan model, development by Alexandra Mihai, which can be used for auditing modules. This reminded me of storyboarding process done as part of instructional design before a module goes live, but for auditing and checking a module which is ongoing.

Leave a Comment

Exploring Modality in the Context of Blended and Hybrid Education

Published 2nd February 2024

It’s come to my attention, because I’ve just been writing about this for my CMALT portfolio review, that I don’t always record HeLF webinars on my CPD record, so here I am, doing just that. The ‘Heads of eLearning Working in UK HE’ forum facilitates regular CPD webinars for its members, and this one was exploring different kinds of attendance in a post-pandemic context.

Simon Thomson, of the University of Manchester, began with a discussion on how they have previously used the TPACK Framework in academic development, but found that people often got too caught up in the technology aspect to the exclusion of other factors. Simon has therefore adapted this model, replacing ‘technology’ with ‘modality’ to create the ‘Subject, Pedagogy & Modality’ Framework, or SPAM, instead. The models are captured in the first screenshot taken from the presentation, above. This led into a discussion on the rationale and value of specific modalities, and confusion over terminology. From the second screenshot, the idea of student choice resonated with me. I think it is very much the wrong tack when institutions, or worse, the government, dictate how students should be learning for non-pedagogical reasons. (Like checking visa compliance for example…!)

Sue Buckingham, from Sheffield Hallam, picked up on the confusing terminology in their part of the presentation. How many students would be able to confidently define ‘HyFlex’ learning for example, or explain the different between blended, hybrid, and hyflex? Could you? Could I!? HyFlex is exactly what I’ll be doing when my own module starts up again next week. It’s all been planned and designed to be in person, but I’m also going to stick a laptop at the front, pointed at me and the board, and have a concurrent Teams session running too. Students in 2024 have rich, complex lives. Jobs, school runs, caring commitments, so give them a choice as a reasonable accommodation and act of compassion.

Leave a Comment