If I had conducted the interview with Professor Ben Shneiderman I would have liked to talk more about the strengths and weaknesses of artificial intelligence and machine learning. In particular, I would have liked to hear more about the chess and music examples. Recently, people have been using artificial intelligence to compose music, and while I believe this may be possible for pop music due to its formulaic nature, people are trying to apply this to classical music too. Artificial intelligence and machine learning have been used to analyze hundreds of works from a specific composer with the goal to "finish" their unfinished works by replicating the composer's writing style, an example is Schubert's unfinished symphony. In fact, the work by the artificial intelligence is quite impressive in the way that it emulates specific writing styles and composes melodies referencing previous works. So, what are the limitations of artificial intelligence in its capabilities? Artificial intelligence is able to beat humans in a game of chess where the outcome is definitive, but is it possible that artificial intelligence can compose "better" music, which is judged subjectively?
Dear Daniel, I too am curious about the limitations of artificial intelligence with respect to music. I feel that, especially for pop music, R&B, and hip-hop, sometimes the voices of the singers and rappers actually more than the lyrics being sung/rapped. I wonder what Ben Schneiderman would say about this idea, and I am interested at what he thinks restricts the abilities of machine learning to successfully compose pop/R&B/hip-hop music.
Good question—what are the LIMITS—and how do we find out??