Defining Artificial Intelligence (AI)
The conversation about AI’s uses in education was largely speculative until the public launch of OpenAI’s ChatGPT in 2022, which brought generative AI into the mainstream conversation. Generative AI concerns the creation of new content, particularly images, music or written material. It stands in contrast to classification AI, which focuses on categorising or labelling data based on predefined classes or categories, and predictive AI, which uses algorithms to project forward-looking correlations. These kinds of AI have been widely used by finance, marketing, and healthcare analysts for decades, and more sparsely in EdTech for purposes such as predicting grades.
The most pervasive form of modern AI is the generative ‘large language model’ (LLM) – a machine learning tool that can comprehend and generate human language text or audio to respond to an input of the same kind. OpenAI’s ChatGPT, Google’s Gemini and Meta AI are all examples of LLMs. Though they are not designed specifically for the education sector, their cheapness and ease of use have made these AI programs ubiquitous across the working world.
This KnowledgeBank contains a wealth of resources exploring the uses of multiple kinds of AI and the impact they are having on the world of education today. Here, we explore some of the most common areas where these technologies have found use cases.
For Teachers
Workload Reduction
Generative AI, like most EdTech solutions, primarily appeals to educators because of its potential to automate or otherwise simplify mundane tasks that would otherwise require a significant time investment.
With a little creativity, any LLM can be harnessed to take over essential tasks that require little manual input. Prompts as simple as ‘Write five examples of metaphors and five examples of similes’ or ‘Suggest classroom activities about the Blitz appropriate for a Year 6 History class‘ are a fast way to find material for lesson plans, with results often sourcing examples of best practice shared online by other teachers. Shortcuts of this nature are already well known to thousands of teachers who have been introduced to AI.
AI-powered tools created with educators in mind are especially useful for reducing their prep time, as you might expect. The ‘e-spaces’ platform is an example of this[1], with teacher-tailored features including automated creation of question sets, marking and predictive grades. A human hand on the wheel is still necessary to ensure that the AI’s results are fit for purpose, but the end result is less time spent on rote tasks and reduced stress for teachers.
Assessment
Many educators have identified assessment as an area that is ripe for AI involvement.[2] AI models’ potential for providing personalised assessment materials that adapt to students’ proficiency levels and perhaps even their personal interests.
What this may look like in a formal setting is not yet clear, but individual teachers have already found success in using AI to create questions, compile assessment data and translate materials into newer and more accessible formats.[3] Currently available AI tools have also been put to use in automated grading systems, though this is a contentious area owing to issues of privacy and the low accuracy of some AIs (see Obstacles to Progress below.
For Students
SEND
Some of the most important EdTech applications help young people with SEND to overcome the obstacles to learning that they face. Audiobooks, ‘speak selection’ and speech-to-text software are tried and tested pedagogical tools that many students with SEND find invaluable. The introduction of AI is a natural step in their evolution.
One such example of this is Dragon Speech Recognition[4], which allows dyslexic students to dictate text during classroom and homework activities without the need to struggle with a mouse and keyboard. Powered by predictive AI, the program focuses on quickly and accurately translating students’ thoughts to paper, eroding the barriers faced by students with dyslexia and ADHD.
Subject-Specific Guidance
Generative AI in particular has a wealth of uses when it comes to engaging students in the classroom. Image generation allows scenes from literature to be recreated, or for famous pieces of art to be translated to different styles; history students are even able to role-play by instructing LLMs to assume the role of historical figures.
Citizenship teacher Martin Ridley has described his success in having students use AI to explore topics of national importance[5], and mathematician Lew Ludwig has published a series of articles[6] on his use of AI to teach students about complex subjects from calculus to programming. Many further examples can be found in the resources contained in this KnowledgeBank.
Obstacles to Progress
In spite of their many uses, currently accessible AI platforms have a number of problems in common that keep them from truly becoming indispensable as a pedagogical aid. Beyond the attainment gap caused by lack of access to AI tools – which goes for every example of EdTech – these factors should be understood by any educator looking to make use of AI as part of their practice.
Incorrect Information
LLMs are trained on interactions between humans, and so are biased towards providing helpful answers whenever asked – regardless of whether or not they have the ability to do so. This leads to ‘hallucinations’[7] where incorrect or misleading information is presented as fact. Asking an AI for sources relating to a particular topic is likely to return a list of made-up titles and non-existent authors that the program has hallucinated. In some cases, AI can be bullied into accepting that 2+2=3 or that historical information is false.[8]
This is a pervasive problem that even the latest LLM iterations have failed to address, making AI-returned information fundamentally unreliable as a source of knowledge. Teachers should therefore be sure to check information gained from an AI against reputable sources before presenting it as fact. When conducting research (particularly into complex topics), it is often both faster and easier to forgo the use of AI altogether. It is also crucial that students are taught about these shortcomings and the importance of independent study.
Privacy
It is important to remember the importance of data privacy when using AI programs. Sensitive information, especially relating to students, should never be entered into an LLM or any other platform whose terms and conditions allow data to be retained.
Concerns regarding privacy will need to be addressed before AI models are integrated into any formal assessment process.
Overreliance
With LLMs’ ability to explain concepts, explore topics in greater depth or generate whole essays from simple prompts, it is hardly surprising that more than 67% of secondary students already use AI regularly in their education.[9] However, a good amount of this use ultimately boils down to work avoidance rather than engagement in learning.
Pedagogues like Anthony Seldon caution that unrestricted use of AI in schools will have a deleterious effect on students’ development of learning skills[10], with the tools ultimately infantilising those who rely on them. Whatever long-term norms spring up around AI, educators using it in the classroom today must be ready to set clear expectations for how students will engage with it.
Looking Ahead
There are many more facets to consider before we can urge the widespread adoption of AI as a pedagogical tool, but we can be certain that the models available today are only the tip of the iceberg. AI will inevitably grow more sophisticated as it is iterated upon, growing more entrenched in EdTech. Today’s students already expect to interact with AI in their future employment[11] and want to get an early start in learning about it.
To discover more about the success stories and ethical considerations of AI as it is used in teaching today – as well as projections about the future of the technology – we urge educators to take a look at further listed in this KnowledgeBank.
- https://www.teachingtimes.com/ai-technology-reduces-stress-for-teachers-and-students/
- https://www.teachingtimes.com/assessment-and-ai-the-benefits-and-risks/
- https://www.teachingtimes.com/making-use-of-ai-in-the-classroom-today/
- https://www.teachingtimes.com/intellectualpotential/
- https://www.teachingtimes.com/teaching-citizenship-in-a-time-of-ai/
- https://www.mathvalues.org/masterblog/there-and-back-again-a-mathematicians-tale-of-ai-exploration?rq=lew%20ludwig
- https://zapier.com/blog/ai-hallucinations/
- https://www.theverge.com/2024/5/15/24154808/ai-chatgpt-google-gemini-microsoft-copilot-hallucination-wrong
- https://www.teachingtimes.com/two-thirds-of-secondary-school-students-use-ai-to-do-their-school-work/
- https://www.teachingtimes.com/sir-anthony-seldon-bursting-the-ai-dam/
- https://www.teachingtimes.com/early-talent-career-influences-in-the-ai-age/