Artificial Intelligence is a hot topic and many organizations are now looking to exploit these technologies, at the same time there are many concerns around the impact this will have on society. The concept of AI is not new, but cloud computing has provided the access to data and computing power needed to turn it into a practical reality. Machine Learning technologies provide significantly improved capabilities to analyze large amounts of data in a wide range of forms. While this poses a threat of “Big Other” it also makes them especially suitable for spotting patterns and anomalies and hence potentially useful for detecting fraudulent activity, security breaches and non-compliance.
How can you ensure that your AI project does not become a liability? An improper implementation, a socially insensitive data label, or negligent data management can easily lead to an auditing nightmare. What are the best practices to safely exploit and govern AI?
This workshop will walk you through a realistic scenario where your governance of AI will be put to the test. Learn the key questions to ask when implementing an AI project so that governance and audits do not become an issue later.
After attending this workshop, you will be able to:
Many AI projects falter or fail when they encounter a governance-related problem. Don’t let your investments in an AI project go to waste because of poor governance practices, leading to project abandonment. Take advantage of the potential efficiencies that AI projects can bring by managing your AI projects responsibly.
Good AI governance needs rigorous processes and management, but also suitable tools to enable AI teams to meet the requirements of the governance structures. In this talk, I will introduce several such tools developed by IBM that aim to address several different aspects of AI governance. The most recent of these, AI FactSheets, provides a template and methodology for users to document their AI systems, the decisions that have gone into building them, and the processes used to assess their performance. FactSheets are intended to act as a declaration of conformity for an AI system, providing key information to stakeholders as well as assisting businesses to identify weaknesses in their current governance processes. I will also discuss Watson OpenScale, IBM's enterprise cloud tool for real-time monitoring of deployed AI systems that greatly simplifies the processes required in implementing the governance steps identified in a FactSheet. Finally, I will briefly cover recent advances in two crucial topics: explainability and bias. IBM Research is actively working in both these areas, and has made this work available to the broader AI community through the open-source toolboxes AI Explainability 360 and AI Fairness 360. While the problems of AI governance are far from solved, these offerings provide businesses with a useful suite of tools to deploy so that they can focus on their processes and people, rather than getting bogged down by technology.
