Antidotes to “Bad AI”: Novelty, Transparency, and Choice in IxD

Hosted by IxDA Pittsburgh

"Bad AI" means artificial intelligence systems that sap our attention, warp our beliefs, and polarize our relationships. For example, "recommenders" bring us options: What to watch next on YouTube, what to read on Facebook, and even what to consume in search results. But are these valid choices that we would take on our own? As has been widely reported, these systems create problems for countless users every day.

This talk charts a path to alternatives. We can keep the power of digital computation at scale while swinging back to our analog, open-ended, socially-animated roots as human beings. We can revise today's algorithms to enable novelty and choice, conversation and transparency, even as we continue using digital infrastructure.

This talk describes a new initiative with those values inside and with the goal of collecting as well as creating examples from art, media, performance, design, and software implementations. The result will be a catalog of design patterns, best practices, and working prototypes that can be adapted by designers, coder, and entrepreneurs. Together we can enrich the collective social fabric we need as humans in order to thrive and create.