12 possible questions for investigating the feasibility of Artificial Intelligence in practice

There’s nothing new about AI, but recently AI-based components have moved from specialist to mainstream applications. Let’s assume someone has decided to use some AI, and your task is to make that work in practice.

I’ll propose 12 questions to investigate drawn from historic and current case studies including gamma-ray astronomy, chocolate manufacturing, capturing data from forms, and health care.

You’ll think about whether the questions work as a way of assessing a possible AI application: writing tests, and you’ll have the chance to propose different questions as the basis for an AI feasibility investigation.

This is NOT about how to create AI. It’s about the next step – using.

Audience Requirements

Some familiarity with the idea of creating a service that has different types of computer program and human activities combined.

Objectives of the talk:

  • A reminder of some key AI definitions such as algorithm, machine learning, explainability, training, bias
  • Learn some practical questions to use as the basis of an AI investigation
  • Critique those questions and gather ideas for improving them

Track 1
Location: Date: May 13, 2020 Time: 3:45 pm - 4:30 pm Caroline Jarrett, Effortmark