Tom Chatfield, ‘How to think about AI’
Jonathon Blake, Director of Learning and Academic Enrichment, was delighted to welcome Dr Tom Chatfield to this week’s CPD session for Heads of Department and visiting representatives from the Eton Group of schools.
Tom is a British author, tech philosopher and educator. He is interested in improving our understanding of digital technology, and its uses in policy, education and engagement. He’s particularly interested in AI ethics, critical thinking and future skills. Tom took his doctorate and taught at St John’s College, Oxford, and continues to guest lecture at universities across the world.
Tom’s session focused on the opportunities and limitations afforded by Generative AI; the importance of asking the right questions and knowing the difference between a good and a bad answer. He explained that whilst it is easy to use, but hard to use well, given that, “it doesn’t live in the real world, but a world of data”, making it necessary to validate both the questions we ask it and the answers given independent of the large language model (LLM) being used.
He explained that whilst AI can be personalised, scalable, engaging, supportive and accessible; it can also be authoritarian, overwhelming, disengaging, misleading, and unethical. Institutions that utilise it effectively have transparent codes of practice, giving permission for it to be used for specific purposes in certain situations.
Whilst he explained there are no firm ‘dos and don’ts’, he recommended three thinking tools when considering how to use it optimally. These were: (1) use it as a ‘cognitive prosthesis’ to outsource a specific action, or automate what can be more efficiently by a machine to help us manage our cognitive load and support depth of engagement; (2) be mindful of the challenge of assessment, where it is not the provision of the answers which matters, but in asking the right questions. For example, in a world where answers may be generated by using the right prompts, the learning happens in the process of trying to find the right words; (3) avoid artificial idiocy, by highlighting how playing with images can reveal the limitations of a model more effectively than words, which may dupe us more easily. The example he provided was how Google wanted AI not to be racist, so inadvertently the AI rewrote history to avoid such a bias; it also had no issue with the query, ‘how to get to the moon using ladders?’, noting how the effect of our queries could lead us into the trap of confirmation bias. During the session, there were several breakout sessions, providing time for the group to discuss topics such as, ‘where can AI elevate and enhance learning?’; ‘what are the skills needed to use AI wisely?’, and ‘what do you know that AI doesn’t’?’.
In sum, he concluded that it’s wisest “to use AI to help us to structure our thinking, without thinking for us… that teachers know that students need and should use AI to fit into the jigsaw [of approaches], rather than replace them”.
This session is one of a series of ways in which staff and students at Tonbridge are exploring the ethical use of AI to support learning. The exploration of, and learning how, to use AI begins with the Digital Information Literacy sessions which boys undertake in the Novi, which guides them in the ethical use of technology, followed by the provision of a student handbook entitled, ‘Using Generative AI as a tool for learning’ to support its effective use. Tonbridge is also introducing its own ‘traffic light system’ for boys and staff to use as guidance when setting and completing work, indicating whether (and how) AI can be used.
