Self-driving cars, automated personal assistants and domestic violence, and how they each relate to data and artificial intelligence, were just some of the topics up for debate at this week’s Ethics of Data Science conference.
There’s no denying that data and artificial intelligence are well on their way to impacting every facet of our daily lives. Already, industries and governments are relying on machine learning to make important decisions that will have a real effect on the lives of consumers and citizens.
Hosted by the University of Sydney between 27-29 March 2019, the forum brought together world-renowned experts to address the current crisis in confidence around algorithms, including Australian Human Rights Commissioner, Edward Santow and Commonwealth Bank of Australia’s Head of Data Science, Dan Jermyn.
Algorithms are a fundamental tool in everyday machine learning and artificial intelligence, but experts have identified a number of ethical problems. Models built with biased and inaccurate data can have serious implications and dangerous consequences, ranging from the legal and safety implications of self-driving cars and incorrect criminal sentencing, to the use of automated weapons in war.
The conference's speakers, who spanned diverse disciplines including ethics, law, and artificial intelligence, discussed the current research and practice relating to the ethics around algorithms, and identified solutions for creating a new generation of ethical data science techniques.
It is important to note that algorithms are not unethical, it is the bias in sampling created by some implementations of them which is an issue.
Centre for Translational Science data expert, Dr Roman Marchant, believes there needs more concerted effort between government and private institutions who can use data to better understand criminality and put an end to domestic violence.
“We are at a point in history where we have a world of data at our fingertips however the complexity of issues such as criminology will always be bigger than any amount of data that we have.
“To tackle criminal behaviour and understand the drivers behind it we need to build truly multidisciplinary partnerships between data scientists and experts in criminology,” explained Dr Marchant.
“To reduce or eliminate crime, we need to focus our efforts on understanding the problem. Like with any crime, there are specific levers and influences which lead to a person committing domestic violence.”
Director of the Centre for Translational Data Science Professor Sally Cripps believes data experts must understand how to quantify uncertainty to prevent bias.
“It is important to note that algorithms are not unethical, it is the bias in sampling created by some implementations of them which is an issue,” she explained.
“If an algorithm finds that a subgroup of the population is more likely to experience domestic violence, and on that basis continues to sample from that subgroup, then it is a self-fulfilling prophecy. To guard against this, a deep understanding of uncertainty and how to quantify it needs to be incorporated into algorithms.”
The Centre for Translational Data Science uses data science to preserve natural resources, build intelligent systems, improve digital health and explore the human condition. Hosting experts from across a diverse range of disciplines, the centre tackles important research questions, applying innovation and translation, application and foundations to find solutions and ensure real-world impact.