Anyone know of material addressing question formulation in AI systems? I believe there are a few architectures that can produce predictive or creative output based on patterns of input like Boltzmann machines, but I'm skeptical that they can address things like novel question formation based on a current situation. I'm very interested as to how data analytics of the future will use AI systems to find gaps in data and further how the systems might address these gaps by asking questions of human researchers.
Related: I'd also be grateful if anyone has recommendations for linguistics texts that focus on questions.
Cheers!
If I'm interpreting you correctly, you're interested in AI systems that formulate questions, rather than AI systems that answer them. I haven't seen anything on questions in particular, but for systems capable of producing novel linguistic output, some terms you might search Google Scholar on include "natural language generation", "narrative generation", and "discourse generation". Despite its age, Building Natural Language Generation Systems by Reiter and Dale may still be the most comprehensive book on the subject.