Head’s Blog: The AI Evolution in Education
Revolution is more often used than evolution to describe the rapid and transformative changes that are occurring in the economy as previously human cognitive tasks are automated. AI offers vast opportunities to accelerate processes, boost productivity, and create wealth. But it could also de-humanise interactions, destroy jobs, erode human worth, and increase inequality.
The AI revolution will make the industrial revolution look small and slow, and we should learn from the errors of the past, when rapid unregulated industrial change caused pollution, unemployment, poverty, inequality and social unrest.
Let me start by saying I am not a Luddite. The Perse has been celebrating the Nobel Prize winning work of alumnus Professor John Clarke whose research into quantum mechanical tunnelling and energy quantisation laid the foundation for superconducting qubits used in quantum computing. Quantum computing could significantly accelerate the AI revolution by training AI models to make faster and more accurate predictions. Quantum for AI could, for example, speed up pharmaceutical drug discovery and with it the prospect of extending and saving lives.
But when it comes to education, I am more cautious about AI, preferring evolution rather than a revolution, albeit evolution at fruit fly speed.
Children are a vulnerable group because, dependent on age, they cannot give fully informed consent, and rely on parents and teachers to safeguard them. In safeguarding children there must be a ‘first do no harm’ principle, so before AI tools are rolled out in education, risk assessments must occur. Ideally, these risk assessments will be informed by high quality peer reviewed studies which evaluate the pros and cons of using AI in given circumstances, and identify ways in which pros can be accentuated and cons mitigated. However, given the pace of change in AI, researchers will have to work hard to keep up.
AI is becoming ubiquitous in twenty first century life, so not introducing children to it in a measured, intentional, and ethical way will itself be harmful, as schools will not be preparing pupils adequately for their future lives.
The Perse aims to be a leader in AI in education best practice.
Our approach is shaped by two complementary objectives – to give pupils an excellent education in human capacities, such as oracy, creativity, collaboration, emotional intelligence, critical thinking and judgement, and to develop their digital skills and understanding so they are confident using AI wisely and well. This requires students to understand how AI works through statistical prediction rather than true understanding, how bias enters systems, why human judgement matters. We want our students to ask great questions of AI, for this is how you get the best outputs, and then critically evaluate those outputs to make them better. AI should be seen as the latest in a long line of tools from stone age axes to calculators that humans have used to amplify their capabilities.
The distractions and temptations of AI which adults can struggle to resist, can be irresistible to children. Schools, teachers, and parents need to set clear parameters on how/when AI can and can’t be used to speed up learning without short-cutting learning. At The Perse we do this via a system of traffic lights where teachers determine the extent of pupil AI use for any given task. There must be enough opportunities for pupils to learn concepts and practise them, free of AI, for their own cognitive and confidence development.
Used well, AI will accelerate learning. Pupils researching topics can interrogate AI to extend their basic knowledge and understanding faster than would have been possible with physical books or search engines. But having used AI to quickly gain a basic understanding of a topic, students in search of deeper learning will need to use more traditional media and discussion with teachers to check, consolidate and extend their understanding. Pupils stuck on homework no longer have to wait for a teacher explanation in the next lesson to move on as AI can provide one there and then, including consolidation exercises and feedback to improve understanding. AI allows students to have a learning assistant on tap, but the tap needs to be switched on and off. AI can assist pupils with their tasks, but it must not do them. Human brains need to be stimulated and challenged to enhance their neuroplasticity and cognitive functions. This stimulation and challenge could involve humans thoughtfully and critically using AI, but AI “thinking” should not be a straight substitute for human thinking.
Students will benefit if AI transforms teacher workflows. The aim should be to digitise basic tasks, so that teachers can concentrate on higher value activities. AI can be used to generate worksheets and presentations which skilled teachers can then enhance so that pupils receive excellent resources. At The Perse we have successfully trialled learnanything.com to mark short answers as accurately as teachers, and it does so instantaneously whilst also providing detailed and helpful feedback to pupils. It is critical that a human stays in the assessment loop for quality assurance purposes, but even so the staff time saved adds up which can be used for higher order things that only human teachers can do, such as 1:1 coaching, practical activities, or tutorial style discussions.
A well-resourced school like The Perse in an epicentre of digital innovation like Cambridge, has a duty to lead on the safe, ethical and effective use of AI in education so that we can maximise the teaching and learning gains, mitigate any downsides, and share this information with fellow teachers for the good of all pupils. This will take time, hence the need for an AI evolution in schools rather than a revolution …. albeit evolution at fruit fly speed so that children don’t unnecessarily miss out on the gains that AI will bring to education.
Ed Elliott