A class at Georgia Tech did an experiment where artificial intelligence (“Watson”) was used to “enhance human creativity”. It sounds like a cool class:

Following research on computational creativity in our Design & Intelligence Laboratory (http://dilab.gatech.edu), most readings and discussions in the class focused on six themes: (1) Design Thinking is thinking about illstructured, open-ended problems with ill-defined goals and evaluation criteria; (2) Analogical Thinking is thinking about novel situations in terms of similar, familiar situations; (3) Meta-Thinking is thinking about one’s own knowledge and thinking; (4) Abductive Thinking is thinking about potential explanations for a set of data; (5) Visual Thinking is thinking about images and in images; and (6) Systems Thinking is thinking about complex phenomena consisting of multiple interacting components and causal processes. Further, following the research in the Design & Intelligence Laboratory, the two major creative domains of discussion in the class were (i) Engineering design and invention, and (ii) Scientific modeling and discovery. The class website provides details about the course (http://www.cc.gatech.edu/classes/AY2015/cs8803_spring)

Here’s how they actually went about using the computer:

The general design process followed by the 6 design teams for using Watson to support biologically inspired design may be decomposed into two phases: an initial learning phase and a latter open-ended research phase. The initial learning phase proceeded roughly as follows. (1) The 6 teams selected a case study of biologically inspired design of their choice from a digital library called DSL (Goel et al. 2015). For each team, the selected case study became the use case. (2) The teams started seeding Watson with articles selected from a collection of around 200 biology articles derived from Biologue. Biologue is an interactive system for retrieving biology articles relevant to a design query (Vattam & Goel 2013). (3) The teams generated about 600 questions relevant to their use cases. (4) The teams identified the best answers in their 200 biology articles for the 600 questions. (5) The teams trained Watson on the 600 question-answer pairs. (6) The 6 teams evaluated Watson for answering design questions related to their respective use cases.

The value of the computer seems to be in helping the humans sort through and screen and enormous amount of literature in a short time that otherwise could take years to go through. This theoretically could accelerate progress by allowing us to make connections that otherwise could not be made. There are going to be some brilliant ideas out there that are stuck in a dead end where they never got to the people who can use them. And there are going to be many more brilliant ideas that emerge only when older ideas are connected.

These students seem to have restricted themselves to a research database in one field (biology). But I think it could be very valuable to cross disciplinary boundaries and look for analogous ideas – let’s say, in thermodynamics, ecology, and economics. Or sociology and animal behavior. These are boundaries that have been crossed by just a few visionary people, but are often ignored by everyone else. If making connections was more of a standard practice, many more brilliant ideas would escape the information cul-de-sacs.

This reminded me of the novel Stand on Zanzibar, where “synthesist” is a job. The world is not doing so well, and governments are seeking out unconventional thinkers to try to synthesize knowledge across multiple fields and try to come up with new problems. There is also an artificial intelligence in the book as I recall, but I don’t remember it being involved in the synthesis. I don’t have a copy of the book, and this particular piece of human knowledge and creativity is walled off from me by “intellectual property” law, so I can’t benefit from it or connect it to anything else right now.