Data and algorithms are increasingly making our decisions for us, and their use is inevitably spreading into public sector policy development and decision-making.
Despite this, public services are often unaware of the biases inherent in data and its limitations, as well as how to effectively combine technology with people skills.
Professor Leah Ruppanner, from the University of Melbourne, who will be one of the presenters in ANZSOG’s 2023 Public Leadership Masterclass series, with a masterclass on Understanding and Correcting Data Bias says that the public service needs to change to take control of its use of data.
In her masterclass, Professor Ruppanner will introduce participants to the ways data drives big and small decision-making and how bias can influence the data in overt and covert ways, using gender bias as an example.
“From your Netflix to Spotify, to the push notifications you get, to issues around who to hire, it’s all being guided by your data. To some degree, people know the obvious ways, but there are other less obvious ways in which data is influencing the decisions that impact your life,” she said.
“Right now, you are providing data to help build the next generation of technology which will again inform how you make decisions and live your everyday life.”
“Even some of the tech companies which are building the technology that’s guiding our prescriptive decision-making don’t fully know how it works, because its unexplainable AI, it is doing machine learning and it’s making decisions in a way that is often opaque.”
Professor Ruppanner said that the positive view of public service data – that governments would be able to use data to make better decisions because they would know more about their communities – ignored many issues around the quality of the data being collected, and how it was used.
“Outcomes driven by data are only as good as the quality of the data, and only as good as the decisions you make based on that data,” she said.
“There are four sets of issues you need to think about: the extent to which you may have chosen the wrong data to answer your questions, the extent to which the data is wrong, biased or incomplete; the methods you are applying and whether they are right or wrong; and the extent to which good decisions can be made based on your data.”
“Data is really good for certain groups, but in some instances, it is not good for marginalised groups, such as older people, people with English as a second language, or unemployed people, yet these are in many cases the people that public services need to be most concerned with.
“I am working on a project on genderised hiring algorithms to determine what extent gender matters for the big companies that are using algorithms to rank CVs. We are pretty easily finding that women are experiencing discrimination from these algorithms because, even with two candidates with the same qualifications, experience etc, the algorithm prefers the male candidates.”
“We also know that because of the bias in data collection, and the absence of women in data, science and medicine often assume that women’s bodies can be treated as men’s bodies or behave the same as men’s bodies, which leads to negative outcomes such as women being less likely to be diagnosed with heart disease.”
“Yes, we have more data points that can help us make decisions, but I think we often assume the data is neutral and overestimate the value of the machine or algorithm and underestimate the value of the human who is there to arbitrate, or to decide whether the decisions from the data are actually accurate.
The recent Robodebt scandal, which saw over-reliance on data and an algorithm cause the federal government to falsely accuse thousands of welfare recipients of claiming money they weren’t entitled to, illustrates the danger of relying too heavily on technology.
Professor Ruppanner said it showed that public services need to think carefully about how they operate in a world that is technology-influenced even if not yet technology-controlled
“We need to be cautious about the move to automated decisions that have serious real-world consequences for people’s lives. You can see why that is really appealing in a public sector that is overworked and overstretched, but we need to be careful about what we are asking technology to do, when it comes to outcomes that effect people’s lives, whether they get jobs, money, housing or healthcare.
“We need to be consciously thinking through where we need human intervention or oversight in the process.
She said that governments were not moving quickly enough to put ethical barriers and safeguards around their own use of data, or the private sector’s
“Data and technology is moving so fast that it is predominantly regulating itself. We don’t have frameworks to put around technology because so much of the data and the algorithms is proprietary. There are also power disparities between global tech companies and governments. When the Australian government tried to regulate Facebook, they were like ‘we’ll just pull out of your market’
Professor Ruppanner said that the future of work, in both public and private sectors was not just about technology and data but about the coming together of data and human skills.
“We will have the best outcomes if we have people with lived experiences, including in the social sciences, and technology working together both in the design of the technology and the rollout. We are getting sold a lie if we are told the future of work is just technology. It is this intersection of tech and creativity, adaptability and human interaction.
“There is a knowledge gap, and a potential skills gap between what the Australian workforce thinks they need for the future of work and what they actually will need, and as technology accelerates that gap will widen.
Some of the work we are trying to do is make sure that people who are not traditionally in STEM fields have some of the knowledge about what the technology is, where the changes are coming and what skills you will need. There is a big segment of the population that doesn’t understand what the technological changes will be and where we want to increase the base knowledge and make sure that they can be part of the conversation.
“This will hit all sectors but think it will be harder for the public service because the tech sector and the private sector know where they are going but the public sector is still at an early stage of thinking through these issues.”
ANZSOG’s Public Leadership Masterclass series (PLM), will re-energise and educate hard-working and passionate emerging and current leaders and expose them to fresh ideas. The thirteen masterclasses cover a range of themes and provide an invaluable opportunity for self-reflection and professional growth for you and your team. PLM is a ‘choose-your-own adventure’ style series which puts you in control of your online learning experience. Choose from various packages which feature masterclasses led by leading domestic and international thinkers on leadership and public management from the public, non-profit and private sectors.