The fact that a large majority of people consistently hate word problems shows exactly the problem... They can do the thing computers are inhumanly good at, but they can't do the one thing computers are really bad at!
Programming a computer to solve the kinds of problems we normally think of as "word problems" would not be very difficult, because they tend to be stylized and follow predictable patterns. Training a typical statistical natural language processing engine on a corpus of pre-tagged examples of word problems paired with a direct translation into algebraic notation would probably do it.
What's truly difficult is encountering a new situation and figuring out which information is relevant, where to start, what the problem to be solved even is. We don't know how to teach computers how to do anything like that, but our typical program of teaching humans isn't so great at it either. Certainly "word problems" are not an effective indicator of students' ability to do real problem solving.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Re: Important things.
Programming a computer to solve the kinds of problems we normally think of as "word problems" would not be very difficult, because they tend to be stylized and follow predictable patterns. Training a typical statistical natural language processing engine on a corpus of pre-tagged examples of word problems paired with a direct translation into algebraic notation would probably do it.
What's truly difficult is encountering a new situation and figuring out which information is relevant, where to start, what the problem to be solved even is. We don't know how to teach computers how to do anything like that, but our typical program of teaching humans isn't so great at it either. Certainly "word problems" are not an effective indicator of students' ability to do real problem solving.