By Marlon Kruizinga
These meetings have become the central forum to discuss the research methodology for AI MAPS. We reflected on our visit to the flower auction in Aalsmeer and we also sought together to come to a working definition of ‘Artificial Intelligence’. The purpose of this was first to recall and record important insights from interactions at the flower auction, and second to develop an AI definition which would allow us to pick out which technologies are relevant to the research in a given case.
We noted with interest the argument made at the auction which suggested that broad data collection may be necessary at first to ‘know what to look for’ in future data collection through AI. Though we also noted the possibility for this thinking to lead to indiscriminate data collection. We concluded that for such a broad and exploratory data collection phase in development of monitoring software, strict guidelines must be in place which ensure responsible data collection and storage (e.g. destroying data after the exploratory phase). As for our working definition of AI, we had a great presentation from Michael, TU Delft’s PhD candidate, which illuminated some popular misconceptions about what AI is, such as that it is identical with data or algorithms. Ultimately, we decided on a pragmatist working definition which prioritised those technologies which are impactful enough in taking over human tasks to warrant our research focus. In addition, we decided to create a taxonomy of systems we (and other stakeholders) consider AI in the near future.