AI Could Ease Teachers’ Workload But Needs Close Human Supervision, Warns Exam Board

AI (Artificial Intelligence) letters and a robot miniature are seen in an illustration image on June 23, 2023. Dado Ruvic/Reuters

Artificial intelligence has the potential to help reduce teacher workload by automating simple marking processes, but needs close human supervision to be effective, said one of the leading UK exam boards.

One of the three main assessment bodies, AQA, has written to the government to inform its consultation on the use of AI in education.

AQA researchers trialled AI tools including ChatGPT, GPT4, LLaMA and Alpaca on a range of science papers. Their findings showed that AI could help teachers create tasks and informal assessments for students, such as quizzes.

For example, teachers could use AI to make a bespoke on-screen quiz on a topic, and then use the tool to mark the students’ responses and offer feedback.

ChatGPT trial showed there was a potential for AI to create curriculum summary presentations.

While workload remains one of the biggest factors in teacher recruitment and retention, researchers looked into how AI could reduce it, without compromising the quality of education.

AI could be useful in helping teachers with shorter pieces of feedback and their distribution to senior leaders and parents. An interactive chatbot, such as ChatGPT, could create a lesson with audio, video and text resources around a given topic, under the teacher’s supervision.

Alex Scharaschkin, AQA executive director of assessment research and innovation, noted the potential of AI to transform education.

“If AI can reduce workload by helping with lesson planning and marking, then the brightest people will be more likely to become teachers and stay in the job,” Mr. Scharaschkin said.

Risks and Limitations

The government’s call for evidence seeks to understand the perspective benefits, as well as risks of technology when it comes to educating young minds.

In its response to the Department for Education consultation, AQA said that certain risks were associated with AI use by teachers.

AI chatbots and other tools can base their output on wrong and misleading information, otherwise known as hallucinations. AI could be limited in the way it interprets content and perpetuates biases and prejudices if it is informed by unregulated real-world data.

“There is a chance that the AI systems simply perpetuate popular myths, as they have no real-world context to draw upon beyond ‘what’s talked a lot about on the internet’. As a result, their explanations might appear convincing while not being based on facts,” AQA said in a blog published earlier this year.

Mr. Scharaschkin has likened the AI tools to actors in the British medical drama series Casualty, where they learn medical jargon and sound like medical experts but cannot perform surgeries.

“They will always need close human supervision,” he added.

AI would struggle to explain how it reached a marking decision, something that teaches and students can request after exam results day. AQA claimed that at present all AI-generated explanations would have to be reviewed by human experts.

There is a lack of accountability when it comes to the application of AI in education, AQA said, adding that it broadly supports the government’s white paper on AI regulation.

Published in March, the white paper sets out the government’s vision for harnessing the potential of AI, while exploring regulatory frameworks to protect against its potential harms.

Among the risks posed by AI are physical harm, an undermining of national security, as well as risks to mental health. The UK has set out a pro-innovation framework for AI use by existing regulators in the sectors where AI is applied. This is in contrast to the EU, whose approach to tech regulation is more protective.

This autumn, the UK will host the first major global summit on AI safety, where together with the U.S. leadership, the government will take a “coordinated approach” to emerging tech.