Can generative artificial intelligence tools actually support learning?

Kortext's James Gray introduces a new initiative to harness the power of generative artificial intelligence (genAI) to support learning

James is the founder and CEO of Kortext

Most thinking about generative AI in higher education has been focused, perhaps inevitably, on plagiarism.

It’s been a pattern throughout the information revolution – you may be old enough to remember the similar conversations when mass access to the nascent world wide web became possible in the late 1990s.

A new way of accessing and using information emerges, therefore there is a concern that students may use it to cheat on assignments. Though it may be true for some, it is hardly a fault in the technology itself.

Beyond the danger zone

Focusing just on the dangers of new technology is, in itself, dangerous. Just as current technology has brought new ways of working that have changed the very nature of employment, it is fair to assume that generative AI will have an impact on the way we work in the future.

As Tom Chapman, Principal Teaching Fellow, and MSc Marketing Programme Leader at Southampton Business School, told me:

It’s like a firework has gone off, sending lots of light and energy in numerous directions. Businesses are coming up with all sorts of bright ideas to integrate genAI into everything, but a lot of these lights will go out. For me the immediate future will see integration into current systems

On assessment, he sees genAI as a fact of life:

I’m not penalising students for the use of genAI tools this year – for a start it’s undetectable. The approach to students using these tools should be based on good assessment design, based on authentic assessment approaches. Marketing professionals are using genAI so our students studying marketing need to learn how

Already we see genAI built into software and platforms in ways that can be genuinely useful. But genAI is not just the chatbots and image generators we have been used to playing with. And that’s why we’ve built a higher education specific implementation of genAI into our Kortext platform that goes live today 5th October 2023 with the launch of Kortext Premium.

If you are used to things like ChatGPT, you will be aware that the current state-of-the-art in genAI is limited and shaped by the kind of training it is given. In most of the text-based tools we have seen so far, the training materials are an enormous corpus comprising (as a starting point) pretty much the entire internet. This is clearly both a strength and a weakness – the strength is that just about everything is on the internet, the weakness is that not everything on the internet is accurate or useful.

And the chat interface – while undeniably impressive – is also surprisingly difficult to get good content out of reliably. There is an emerging artistry in prompt engineering, and this is a skill that will no doubt become ever more valuable in years to come.

As Chapman put it:

gen AI tools are exceptionally useful for idea creation, planning, and the interrogation of documents. They are inherently un-useful for producing academic writing

Training plan

But what if you trained a genAI (in this instance actually based on ChatGPT’s GPT heart) on university content and academic materials – While you wouldn’t be able to ask it any question (and yield the hallucinations and fake references you get from chat bots) you would be able to ask, with some degree of confidence, about the subject matter of the content used.

After all, we can be broadly confident that academic content is accurate and reliable – and as students already use these materials in supporting their learning we are not introducing any new complications.

What our tools offer, therefore, is bespoke summaries, study notes, and revision questions based on specific sections of books or on the concepts they cover. We’ve been testing these with students over the spring and summer, and we’ve also spotted that staff are keen to use them to generate summaries to help them in the course preparation.

Chapman suggested that:

Bringing in an AI layer to the Kortext platform is a bit like adding a virtual assistant. It will allow us to raise the bar and stretch students further. I’ve particularly found that using it for summarising text – even as feedback on my own text – is very helpful

We’re fascinated to see how students start to use these new functions – and it will even be possible for institutions to track such use against attainment through our data platform. One concern some have expressed is that the process of note-taking etc is a valuable part of the learning process, and is there therefore a risk in ‘lost learning’ when AI is assisting in this task? However, generated notes may also lead to more efficient revision, especially where student time is constrained – and the process of checking such material with the source text is another valuable way to drive engagement with the material.

Whenever any new technology emerges into the education space we deal with a whole range of responses – from those who believe it will change everything forever, to those who struggle to see a use for it. In reality, what tends to happen is a technology can do a few things really well and is widely used for that purpose. We think that genAI can be really helpful in undertaking some of the heavy lifting – generating summaries, study notes and revision help – and we can’t wait to see what staff and students do with these tools.

To find out more about the sector leading genAI study tools now embedded into the Kortext platform go to:  

One response to “Can generative artificial intelligence tools actually support learning?

Leave a Reply