Scott Porter, VP, Methods, and Aliza Pollack, VP Qualitative, led a session at the Planningness 2015 unconference. Scott works in marketing analytics, and regularly uses Artificial Intelligence algorithms to sort through large quantities of data to find plausible causal models for the interrelation of drivers, outcomes, and intermediate or mediating variables. In facilitating discussions between marketers and modelers, Scott has realized that the process of gearing up to work with AI can help us think better. Computer algorithms have the advantage when sorting through large amounts of data, but they have their limitations. With current technology, artificial intelligence has to sort through the data it is given--it generally can’t intuit missing data that would be important. However, as humans, we excel at this sort of intuition. Or, at least we can... we have to overcome our human tendency to stop when we've uncovered the first plausible answer. We shared a structured approach to brainstorming that forces us to push wider to additional context that might be important. These exercises (looking for multiple causes, looking for side effects, looking for missing causes) are the steps we would need to go through in order to select the right data for a computer to have sufficient information to build a quantified model. However, the steps are useful regardless of whether or not we later quantify the model, because the techniques help us push beyond where we would normally stop because we found a single reasonable explanation. After going through an overview of the theory and the process, we put it into practice. We took turns leading small groups in structured hypotheses sessions to systematically unpack potential complexities of real client challenges shared by members of the session, and brainstorm what information we (or algorithms) will need to better understand potential opportunities. Speaker note annotations are available within the deck if you download the pdf (orange boxes at the top left of each slide).