3 ms·
That's only useful in cases where human intuition matches reality. In a lot of cases it does, in many more cases it doesn't. Where it doesn't rote is the only t
by candiodari 10y ago
That's only useful in cases where human intuition matches reality. In a lot of cases it does, in many more cases it doesn't. Where it doesn't rote is the only thing that may work and intuition without rote will actually prevent you from understanding, and lead you to believe falsehoods. In statistics, this is quite common. You should never use intuition, always use the math, always work it out, even when "it's obvious".
By the way, I have a friend who's a TA in statistics and this is a big point of frustration for them. Undergrads are starting to do just what you said : unquestioningly follow intuition. It's a disaster. Not that there wasn't always 50% of the class doing this, it's just that it's now 90%.
- seanmcdirmid 10y agoI don't think we disagree that much. My problem is simply training humans to become better pattern marchers via rote practice is simply less useful now that computers are getting better (much much better) at that anyways. There is not much point for a person to do model training by hand, for example. So what is left? Abstract reasoning, problem solving rather than number crunching. Being able to understand the math so that it becomes intuition. Many classes are still stuck in teaching students to pattern match rather than really understand what they are doing, and why. Practice on a bunch problem sets without does nothing to fix that. Rote learning is just not useful for the jobs that are humans are going to do next, where computers are going to be able to pattern match much better than us anyways.