How might emerging computational models in machine learning, often discussed by futurists and science fiction writers in terms of “artificial intelligence,” be fruitfully engaged as models of “artificial creativity” that afford critical reflection on analogous processes of human creativity? The intentionally provocative phrase “artificial creativity” means to question familiar cultural distinctions between human and machine. On the one hand, in common parlance, where "artificial" refers to fakes and copies, the phrase might connote imitating or “faking” human creativity. Are “machine learning” and related forms of generativity simply imitative of human learning and generativity, or are they something more and/or other? On the other hand, insofar as “artificial” refers to the work of artifice, the phrase is redundant. Is not all creativity, human and/or machine, in this sense artificial?
Consider, for example, our Emily Markovson experiment, which uses Markov chain processes to build a text bot that generates four line poems based on the complete works of Emily Dickinson and then automatically tweets them. Go here for some examples:
And here is a similar experiment, KJVBot, which generates verses based on the King James Version Bible:
And here is a tutorial -- including all the code you need -- so that you can build, experiment with, and think about your own bots:
- “Build a Bot: A DIY Toy that Makes You Think,” by Timothy Beal and Michael Hemenway. Tutorial video here and accompanying Google Colab notebook here.