I think the argument in this article is probably sound, though we need to be careful about drawing too many conclusions from it. The assertion is essentially that the research to date, and especially the meta-analyses,are not reliable. There are many reasons this is the case, including bad methodology, poor comparisons (eg., "AI + Something versus nothing), unrepresentative data, changes in AI technology, etc. But we should also be cautious about the assessments. We have an MIT researcher, for example, saying "her team has collected almost 900 studies on AI in education and determined that only 20 of them were of high enough quality." Well, anyone could do that (I sort of do it every day). But what counts as quality here? That is an open question in many ways - it's not like medicine, where both human bodies and treatment interventions are describable and quantifiable. So take all of the research, including the criticisms, with a healthy dose of scepticism.
Today: Total: [Direct link] [Share]
