Information overload, lossy models, telephone games, the lure of loops, pretty-but-mostly-worthless-cascades, and the quest for the one roadmap view to rule them all.
I feel I disagree with the email blast example. It is easy to use AI to sift through a large volume of data to find patterns in the data. You might be surprised at what it finds. If you change the free format answers into a structured survey, you severely constrain the possible outcomes, possibly limiting the results to what you expect to receive.
I feel I disagree with the email blast example. It is easy to use AI to sift through a large volume of data to find patterns in the data. You might be surprised at what it finds. If you change the free format answers into a structured survey, you severely constrain the possible outcomes, possibly limiting the results to what you expect to receive.
I'm wondering how much extra context AI would need. In our experiments, it falls flat without an ontology / knowledge graph.