Sean McGregor, Underwriter Laboratories, Rancho Mission Viejo, CA
Adopting artificial intelligence systems capable of operating at fantastical speed and no marginal cost is a business imperative. However, every successful AI system is built on a mountain of failed attempts. Making a system safe for the complexities of the real world without first adding to the mountain of doomed projects requires learning from past successes and failures. This presentation details the elements of successful and failed machine learning programs as experienced by the startups and corporations producing them.