How can we reliably assess knowledge following the advent of Generative AI?
Presented by Manchester Lit & Phil
Recent advances in Generative Artificial Intelligence represented by new tools such as ChatGPT, have caused much excitement and some alarm in education. Mark Johnson’s talk is about what underpins both the excitement and the alarm: the reality that an automaton can select words which are as meaningful to humans as those which might be selected by humans themselves.
Given that education has traditionally associated the assessment of knowledge with the ability to select words in writing, this technological development presents a number of fundamental questions including:
What is knowledge beyond the selection of words?
How is the human selection of words different from that of an AI?
How can deeper, and often tacit, knowledge be taught and assessed?
Beginning with philosophy of language and an account of the mechanics of AI, Mark will present an analysis of these questions in relation to practical experiments. He will argue that understanding what is happening ‘under-the-bonnet’ of AI helps us to see the critical differences between human word-selection and artificial word-selection.
This presents some reassurance as to the uniqueness of human action, but some urgent critical challenges for the future of educational practice. Some examples of innovative educational practice with AI will be presented, drawing on work in the UK, China and in European Universities.
Book Knowledge, Teaching and Artificial Intelligence at Cross Street Chapel
