Stephen Downes

This is a nice article that reviews concepts that should be familiar to authors and researchers but often are not. The core question is: when is the evidence you have sufficient to allow you to believe that something is true (or not true, which may be a different question)? Though an epistemological question, it may also have ethical implications, if you believe "it is wrong always, everywhere, and for anyone, to believe anything upon insufficient evidence" (or to disbelieve anything upon sufficient counterevidence). I think about such questions all the time, as for example when the usually reliable Miguel Guhlin posted a link to 'The Evidence That AI Is Destroying Jobs For Young People Just Got Stronger' where 'the evidence' was a survey of payroll data from a single company of young workers aged 22-25 in "highly AI-exposed" jobs. As I commented, a study like this "does not really constitute 'new evidence' of anything." I don't know why anyone would publish, let alone recommend, such research.