I have no doubt there is a tidal wave coming of GPT-based academic articles coming, but until then, journals are still publishing old-school articles like this. I'm not saying it's a bad article; it's actually quite good. But it's old school in the following senses: first, the problem of 'drop-outs' presumes a specific and arguably outdated model of online learning; second, the mechanism of a decision tree and risk-level classification system harkens back toi the days of rule-based AI, an approach now completely overtaken by deep learning; and third, the idea that there can be an 'explanation' and therefore 'intervention' reflects a similarly dated model of causal reasoning. There's a lot of good thinking here, but it all feels to me like a gears-and-pistons model of learning and cognition.
Today: 5 Total: 5 [Direct link] [Share]
