This is a good article that really makes clear a lot of the institutional constraints to using AI. It's presented in a '10 lessons learned' format, and while some of the lessons are useful (such as, "culture is infrastructure") the greater value is found in the contrast between what we might call an 'institutional understanding' and the wide open context that is the English language as understood by an AI. In an institution, even simple words, like 'enrollment', 'course' or 'undergraduate' have a specific meaning (and indeed may mean different things in different offices) while to an AI any old off-the-shelf meaning will do. As well, there is a distinction between 'facts' and 'truth' as the institution understands them - there's an 'official' story for everything - and what the wider community may view as factual and truthful, which may include some unofficial perspectives. And there are some things for which there should be no official answer - such as 'which professor gives the best grades?' - even if there is a fact of the matter that people want to know.
Today: Total: [Direct link] [Share]
