Not sure how many would be interested, but we recently started looking into using Inform7 as the basis for training environments for machine reading algorithms.
Our paper is here, recently published at a computational linguistics conference:
It’s based on an open-source framework developed by Microsoft (based on Inform7):
There is a recent trend of 3D game programmers working on AI related research (creating training environments), and it seems like a similar trend may emerge also for text based games. It is still early, but there is a lot of really interesting potential for collaboration, I’d be happy to hear if any others are interested in this combination.