Expert in Platforms and Architecture for Data and AI
This session will include a moderated Q&A featuring questions from the live audience.
Most discussions on automating analytics and machine learning stop at the deployment stage when the important elements happen in operation. This session explores what it takes to automate human decision-making, and how recent approaches using machine learning techniques may offer more flexibility or power, but also come with downsides.
Operational consequences are a big part of this discussion. When you automate analytics, you need new practices and technology support to run and maintain the automated systems. This takes planning beforehand and flexibility as you learn how the automated system behaves. It’s important to guard against unwanted outcomes, which means automation entails the need for additional subsystems beyond the analytics to ensure that the decision system does what you intend.
We'll address the often overlooked question of "who are you automating?" with examples from AI systems and how to look at the larger context. The session offers some tips on how to approach automation, avoid common mistakes, and take a more human-centered approach to the solution design process. Finally, we'll provide some dos and don'ts for different stages and types of activities.