Artificial Intelligence (AI) applied to Education has become one of the most hotly anticipated applications of AI.

With ChatGPT (Chat Generative Pre-Trained Transformer) being on many students' speed dial and allowing for groundbreaking ways for teachers to deliver an enriching curriculum, it’s an exciting time to be in the Education space.

Within this article, I’ll share insight into how generative AI is paving the way into the education sector and explore examples of both challenges and opportunities in this fast-growing area. 

The education landscape

Before moving into the nuts and bolts of AI within Education, let’s holistically look at what’s happening within Education across the UK. Education at all levels - whether it be at primary, secondary, or university level - is undergoing significant reform.

For example, T-Levels are being gradually rolled out across England. Broadly equivalent to three A-levels, these programs are designed to provide core knowledge but with the added extra of an industry placement to ensure students can apply their knowledge in a workplace. [1]

In addition to T-levels, there’s been an increase in Masters Conversion Degrees across both AI and Data Science, with over 2,500 (and counting) courses being delivered across English universities: a core element of supporting the growth of the UK in becoming an AI powerhouse as part of the UK’s National AI Strategy. [2] 

AI in education deep dive 

Educational reform aside, we now turn to a closer look at AI applied to Education. A highly charged application in excitement, concern, and pace but significant in terms of market opportunity. With a market size worth $2.5 billion back in 2022, the catalyst of Generative AI is on its way to propel the sector to be worth over $23 billion by 2023. [4] 

There’s a broad range of use cases in how AI can be brought into Education at all levels. A couple of examples of how teachers can utilize ChatGPT within lessons were covered in a recent article: Transforming industries and everyday life with AI applications, but I’ll outline some holistic examples below. First up, is critical thinking. 

Example 1: Critical thinking and AI 

As students' education progresses, the increased importance of critical thinking emerges: the ability to think about arguments and ideas presented and not just recall facts and transcribe verbatim. Take a scenario where students are presented with two papers on what Hallucination in AI systems is: one written by a human and another written by ChatGPT.

This exercise isn’t just checking the student understands what Hallucination means, but also can, more importantly, be used to facilitate discussion on why one interpretation might be better explained than the other. This example isn’t just showcasing the value of utilizing and comparing humans vs. machines, it’s showcasing how the responsible use of chatbots can facilitate learning. And that’s pretty powerful. 

Example 2: School correspondence 

Another angle of use for ChatGPT for teachers is for generating correspondence: not lessons, but letters. Let’s imagine a teacher is planning to take pupils on a school trip but has assignments needing to be marked and returned by the following day. Here goes (slight redaction due to response length): 

In other words, there’s so many opportunities to utilise ChatGPT in education: If you’re unsure whether ChatGPT can do something, give it a go! 

Ethical considerations 

It’s important to recognize that AI’s impact on the sectors it has made its mark hasn’t come without tradeoffs, and it’s no different with Education. Before getting to the elephant in the room, one of the major challenges faced by Generative AI is the pace: AI is growing at such a rate that it’s becoming difficult for policymakers, educators, and even industry to keep up with the advancements. 

With tightened school budgets and curriculum reforms in full swing it’s proving to be a tough balancing act: making sure the next generation of data practitioners are taught skills the industry needs but at the same time ensuring there’s people with the relevant experience to teach the material in the first instance. 

Pace aside, turning to something that’s under increasing spotlight: Academic Integrity and AI. It could be argued that the rise of Generative AI has created a significant shift in how plagiarism is thought of. With ChatGPT’s 570GB of textual training data [5], concerns have been raised about balancing the use of the tool and maintaining the authenticity of the content. 

My take on balancing the plagiarism and Generative AI stance is to educate to educate: educate students about what plagiarism is and why it’s important, and in addition to this, share examples and consequences if plagiarism is detected in pieces of work: this is key in helping students appreciate the importance of authenticity in their work.

On another angle, if there’s elements of the course where ChatGPT could add value to the students learning, share it! A blank canvas is all well and good, but if you feel ChatGPT could help your students in more challenging parts of the course and help them appreciate the use of it whilst not abusing it, then create some guidelines for them to work with. 

In addition to the above guidance, many educational institutions use plagiarism tracking software such as Turnitin, but an area of increasing interest is how well current platforms will be identifying AI-generated content. For example, being able to identify discrepancies in vocabulary or factual errors. This will become an increasingly important area to watch with the recent identification of ChatGPT-4 showing signs of hallucination [6]: phenomena not uncommon in AI systems, but an important aspect to keep in mind as Generative AI systems continually advance.

Conclusion 

Generative AI is here to stay, so I’d encourage educators to work with the technology and not just push against it. Reflecting back on my Mathematics degree, ChatGPT reminds me of a calculator: for basic algebraic manipulation it helps as a safety net, but at the same time the over-reliance on it risks distorting the understanding of the topic.

On the flip side, though, imagine doing work in Bayesian Statistics: the supplementary benefit of ChatGPT allows for improving understanding which is the best part of learning. Industry-academia relations are more important than ever in understanding the maze of Generative AI: take advantage of these links, whether it be through University Industry Advisory Groups or local meet-ups and the result could be transformative. 

Bibliography 

[1] T-Level Qualifications: Department for Education (2020). Introduction of T Levels. [online] GOV.UK. Available at: https://www.gov.uk/government/ publications/introduction-of-t-levels/introduction-of-t-levels. 

[2] AI in Education Market size: Future, M.R. (2023). AI in Education Market Size To Reach USD 23.82 Billion by 2030 with a CAGR of 38% – Report by Market Research Future (MRFR). [online] GlobeNewswire News Room. Available at: https://www.globenewswire.com/en/news-release/2023/04/06/2642270/0/en/AI in-Education-Market-Size-To-Reach-USD-23-82-Billion-by-2030-with-a-CAGR of-38-Report-by-Market-Research-Future-MRFR.html [Accessed 29 Aug. 2023]. 

[3] Digital Skills GDP Growth: The UK workforce digital skills gap Why closing it matters and a roadmap for action. (n.d.). Available at: https://futuredotnow.uk/wp content/uploads/2023/07/FutureDotNow-roadmap_final-digital.pdf [Accessed 29 Aug. 2023]. 

[4] National AI Strategy: National AI Strategy. (2021). Available at: https:// assets.publishing.service.gov.uk/government/uploads/system/uploads/ attachment_data/file/1020402/National_AI_Strategy_-_PDF_version.pdf. 

[5] ChatGPT Textual Training Data: Ruby, D. (2023). 30+ Detailed ChatGPT Statistics — Users & Facts (August 2023). [online] DemandSage. Available at: https://www.demandsage.com/chatgpt-statistics/#:~:text=ChatGPT%20is%20trained%20on%20300 [Accessed 30 Aug. 2023]. 

[6] ChatGPT-4 Technical Report: OpenAI (2023). GPT-4 Technical Report. [online] Available at: https://cdn.openai.com/papers/gpt-4.pdf.