Building AI that works for people: A practical guide for developers

AI systems rarely go wrong because of errors in their coding. However, most failures happen when you have an effective working model, and you bring it into contact with real people, environments, and unforeseen circumstances. This session would provide developers with a framework for realizing the underlying risks in AI and how to ship systems that are not only effective but also reliable, predictable, safe, and ethical.
Using real-life examples, different techniques such as the PRE-DEPLOY check, the impact radius scan, and other frameworks, a brief interactive exercise will allow the attendees to see how decisions can lead to big consequences, and also how to avoid them early, without slowing down developers.