Stephen Downes

When Ben Williamson talks about 'constructing AI' here he is not talking about actually building AI, but rather, he is talking about how we 'socially construct it', in the sense outlined by Eynon and Young, by describing ways people talk about it and think about it. These constructions begin with framing, that is, the outlining of key terms and values informing the concept. The bulk of this describes a number of different frames from various groups involved in some way in the realm of AI and education, including (for example) government agencies, education leaders, educational advocates, AIEd evangelizers, and abolitionists. As I've said before, educational theorists love a taxonomy, and so far as taxonomies go, this is a pretty good one, though it seems to me to overly generalize what we mean by 'AI'. Williamson positions himself as 're-framing' AI in education as a 'public problem' or 'matter of concern', rather than an 'entirely positive' phenomenon, but (in my view) that just sets up a false dilemma. Some AI is a matter of concern, some AI is entirely positive, and most types of AI fall somewhere in between.