Phase One: Build postmodernist gibberish paper generator based on Deep RNNs. Submit generated papers to humanities conferences.
Phase Two: Submit gibberish paper generator to AGI conference, arguing that it has passed the Turing Test/reached performance parity with human academics.
Phase Three: Establish conferences and funding for analysing AI and Deep Learning through the lens of the humanities. Argue that you don’t understand what the AI said, but you’re sure it is profound, and unfortunately you lack the sophistication to understand it completely.