Speaking to ADAPT before Data Edge, Edward Santow, Director of Policy and Governance at University of Technology Sydney, discusses organisations’ risks when using AI and data science.
He mentions three categories of risks: legal, financial, and reputational. He also discusses algorithmic bias, its complexity, and how it can be addressed.
He emphasises the need for organisations to involve multiple stakeholders in governing bias and fairness when using AI and data science.
Regarding legal risks, Santow notes that regulators are starting to focus on applying existing law to the use of AI and data science.
Organisations need to be aware of potential legal issues and ensure that they comply with relevant regulations.
In terms of financial risks, Santow points out that if AI tools go wrong, they can lead to unfairness and discrimination, which can harm both customers and the organisation’s bottom line. Reputational risks arise when organisations become known as being untrustworthy or not using technology effectively.
Santow also discusses the complexity of algorithmic bias, noting that it can range from something that looks dodgy to unlawful discrimination.
He cautions against relying too heavily on automated tools to solve the problem and emphasises the need for a more analogue approach to investigating red flags.
He suggests that organisations involve multiple stakeholders in governing bias and fairness when using AI and data science, rather than relying solely on technical or technological tools.
ADAPT will gather 130 leading Australian Chief Data Officers at Data Edge.