The Future of Education: Embracing ChatGPT or Facing Chaos?

I was recently asked to provide a short interview on the impact of ChatGPT in education.

I'm posting here my responses for anyone interested in how this ground-breaking technology is going to impact young people in their school or multi-academy trust.

Thanks to Derek Hills and Zine Eddine for the inspiration to write this post!

What impact will ChatGPT will have on education, schools and multi-academy trusts?

In short, I think it's going to have a huge impact and I don't say that lightly - I'm not usually one to jump on the latest bandwagon and say that the latest thing is going to change everything. But I think this will.

Why do you think that the impact of ChatGPT will be so significant?

The difference with this technology is that it strikes right at the heart of the very definition of 'education'. Before easy access to information, we considered someone to be well-educated if they knew a lot of 'facts', and our examination system was tailored to this apparently universal truth. People who could remember a lot of stuff tended to do better than those who didn't.

With the invention of calculators and the internet, this definition of education started to look a bit silly, because everyone had easy access to facts. What became more important was how you could assimilate and structure these facts into writing that conveyed a point of view, or a series of logical leaps to answer a multi-stage mathematical puzzle. Education shifted away from repetitive rote-learning to idea synthesis, elevating students from mere sponges to inventive value-creators.

And is that now under threat?

Yes - idea synthesis was supposed to be the final safe haven for the definition of 'education'. Gone were the days of old-fashioned rote-learning with exams in timed conditions and here to stay was the modern world of coursework, giving students time to research a diverse array of sources, form their own ideas and generate long-form content that could demonstrate the very best of their abilities. The problem is that ChatGPT blows this idea out of the water.

Could you describe how?

Any student can now create a very respectable essay, with a few semi-competent prompts and a bit of fact-checking. There are already concrete examples of this directly affecting students’ grades. As far as I can see, each school or MAT is having to decide their own stance on the use of AI, and this is creating a huge differential between students from the most lenient and the strictest institutions. I bet that you will already be able to see this effect in the data in this summer's results; more pronounced for coursework-heavy, written subjects.

So why don't we just ban it?

I don't think it's that simple. Firstly, ban what? If you use it to generate 10 bullet points to write your essay around and then do the rest yourself - is that 'cheating'? What about if you use it to generate a brilliant essay, then go through and make it sounds slightly more like something you would write, with a few bits of rubbish grammar and a dodgy conclusion. Instead of your usual 4, you get a 6. How can the teacher distinguish this from something you've poured hours into and tried your very best at? What teacher is going to risk calling out a student for cheating anyway, where there's nothing they can do to prove it? This runs far deeper than simple plagiarism.

What is the answer to this imminent problem?

The answer I think is more fundamental than we want to believe. The current approach to education and examination isn't compatible with a world where ChatGPT exists. We need to redefine education again.

Is this a doom-and-gloom scenario?

Not at all! In fact, I think the complete opposite is true. With some adjustments, I think we can build a better system that actively involves tools like ChatGPT, enabling students to reach levels that were previously unattainable, but in a way that is a fair and accurate reflection of their true ability. This needs active work though - if we sit back and wait to see what happens, I fear that chaos will ensue next year. We can't afford to get this wrong.

Is there already guidance in place for use of ChatGPT in assessment?

The Joint Council for Qualification (JCQ) has released guidance on AI 'malpractice', but this is practically unworkable in my opinion as it relies on AI detection tools that are deeply flawed and simply do not work well enough. It treats AI as if it were a 'source' to be quoted (like a book) and therefore plagiarised, rather than a tool (like Grammarly) that elevates the quality of the work.

So what practical steps can we take to take advantage of the technology?

Here are 5 things that we can do (and need to do) right now, to set us on the right path.

1. Update the JCQ guidance to be pro-AI and talk about it as a tool, not a source. For example, there is currently no mention of how ChatGPT can be used by a student who is dyslexic to quickly get their ideas down on paper. At the moment, that student would be punished for 'paraphrasing AI-generated content', which feels counterproductive.
2. Provide curriculum lessons for students on how to prompt large language models like ChatGPT to learn about a topic and dig deeper. It can be used as a personal tutor if prompted correctly, helping students to break through mental barriers about a particular topic. This can be a brilliant aid for so many students that are embarrassed to ask for help from their teachers if used correctly.
3. Mark the student, not the student's work. For example, a written assessment outside of timed conditions should really be accompanied by a 'viva' style defence of the writing in front of peers or teachers. I think this is a good idea regardless of ChatGPT, as it is a far truer reflection of the real world, where we need to defend our ideas in front of those who want to challenge us and deeply understand our position.
4. Utilise AI tools to support teachers. ChatGPT can be used to help teachers with tasks like grading, providing feedback, and writing personalised learning plans. By freeing up teachers' time, they can focus more on engaging with students 1-2-1 and overseeing the overall learning process.
5. Educate students on AI dangers. There are lots of them, but we need to trust the younger generation with the tools that will define their future, just like we do with the internet.

Are these significant changes to how we view education?

None of these are monumental changes, just a small shift in how we define education and learning. Overall, I'm hugely optimistic above the potential that this technology can unlock - if we lay out a positive, pro-AI message early and educate students and teachers on how to use it constructively, I think will look back and see this one of the most remarkable technologies ever introduced in schools, in the same league as the internet itself.