Prof Julian Thomas:
I’m Julian Thomas and I’m the Director of the new ARC Centre of Excellence for Automated Decision Making and Society. And we thought we might just talk a little bit about what automated decision-making is, and what the new centre’s all about. So, when we talk about automated decision making, we’re really talking about the proliferation of a whole group of new technologies and systems, where we find computers are starting to take over the role of humans and institutions in making or informing key decisions about our lives across all sorts of areas of activity. So, we’re thinking about for example, how machines are beginning to make decisions about whether people are eligible for insurance or a particular social benefit from the government. They’re making decisions about what sorts of grades students get, they’re making decisions about what sorts of news items you see when you open your favourite social media platform. So, there’s been a proliferation of these kinds of what we call ‘socio-technical systems’ in the last few decades. They’re having considerable impacts on how we work on how we live, on the way we experience culture, and everyday life.
So automated decision-making promises extraordinary benefits, but it also carries with it really considerable risks. And it’s very important that we do what we can to mitigate those. We know what some of those risks are. There’s the problem of data discrimination where the data sets that our artificial intelligence systems are learning from turn out to be biased. There are systems which we find systematically discriminate against underprivileged or disadvantaged groups. So this is a significant problem, and we need to deal with it, but we can’t just solve these problems in terms of the technologies themselves. They are social as well as technological systems, so they require us to look at the whole political and social environment in which these systems work. They require us to look at the people who design and build these systems, and the people who work with them and whose lives are affected by them. They’re affected by the kind of data that these systems draw upon in order to make predictions and decisions. They’re influenced by the institutional setting in which these systems operate, whether it’s government departments or systems of rules developed within a professional context. Universities, businesses, any of those kinds of spaces. So there’s a lot of complexity about the ecosystems in which automated decision making systems work. And in order to understand those you need to bring a number of academic research and professional disciplines together. You need to work across the humanities, the social sciences and the technological sciences, and so you need a research centre that can operate at scale. Not all of the expertise is located in one University, so we need a national centre which brings together capability from Universities across the country, and makes links with significant centres of expertise overseas, and that’s what the ARC Centre of Excellence for Automated Decision-Making and Society has been equipped to do.
The Centre’s trying to achieve a number of important things in order to address the risks of the proliferation of automated decision making in the first place. We’ve developed a research program which will provide us with the first integrated understanding of how automated decision making works, its dynamics and distribution in Australia, in countries around us, and comparably across the world. So in the first place, the centre’s designed a research program which aims to produce the first integrated account of the distribution dynamics and consequences of automated decision making in Australia, and in our region. That’s a very important part of what we’re doing. We’re also training a new generation of researchers in the necessary cross-disciplinary skills, to work in this space. We see them going not just into Universities to teach subjects around this kind of problem in the future, where we think it’ll be very important, but also going into industry and government and the community sector. We see the Centre making a significant difference to debates around policy and practice in this space we see the Centre providing examples of how we can make sure that automated decision making as we go along, is better than it has been. More responsible, more ethical, more transparent, and more inclusive. So they’re our key aims for the Centre. But the last thing which is very important, is contributing to that public debate about where Australia goes, what our automated future is going to look like, who is going to be included, who is going to be supported, and how are we going to make those critical decisions about what our future looks like?
The Centre’s research program works across all the key different dimensions of automated decision making as we’ve conceptualised it, and as we understand its social consequences. So we have four research programs, each of which are involved with understanding what we see as a critical component of automation in this area. The first research program is about data. Where does the information that our machines use come from, how is it structured, how is it organised, how is it circulated, who produces it and whose data is it in the sense of who is it about, and where does it come from? These are all intensely political and terribly important, technical questions. The second of our research programs is about machines. The systems which actually do the automation, and these range from things like facial recognition systems which we’re starting to see in places like airports and elsewhere, to recommend the systems of the kind that we encounter on the web all the time. We’re on TV when we watch shows, we watch platforms like Netflix. They’re very diverse, they’re ubiquitous, they really need to be understood.
The third of our research programs is about the people. The people who design and build automated systems and the people whose lives are affected by them. We’re interested in the social distribution of automated decision-making, we’re interested in how people imagine the future of these things, and how that creativity feeds into the way these systems work. And the last of our key components of automated decision making and the last of our research programs is about institutions the the administrative governmental, legal, and business systems which influence how and where automated decision making happens. And we see in some cases that what is going on with automation is that some of those machines are starting to take over the role of institutions, just as in an earlier wave of automation machines took over the role of human labor. So they’re the four key constituent elements of automated decision-making systems. They’re all at play as automation rolls on in countries like Australia, and as our work in the Centre progresses, we’re very interested to see how all of these different elements interact with each other.
But it’s very important with our work that we’re not focused on purely academic debates about the dynamics and distribution of automation, we’re really interested also in where automation is playing out and how the debates we’re interested in are affecting developments on the ground. So our research program also comprises what we call focus areas; specific domains where we see automation already being well advanced, and a lot of the debates we’re interested in already going on. So these include areas such as news and media, where we’re familiar already with the effects of automated systems on shaping news feeds for example, on social media. They include social services, where we’ve seen the consequences in Australia of poorly designed automated systems on ordinary people’s lives already. And the reason why we have to work much harder on this than we have so far. We’re interested in social services generally, we’re also interested in health where automation is also moving quickly. And our last focus area is about mobility and transport, which of course is also changing very quickly with the impending arrival and evolution of automated vehicles, autonomous systems, and all of the logistics that sit behind that. So they’re the four domain areas, the four focus areas where we want to bring all of our researchers together to think about the on the ground problems, the real world problems that we’re encountering when we look at automated decision making, as it is playing out now, in this country.
The centre is going to be working with a large number of partner organisations they range from community organisations, non-profits, corporations and international research partners around the world. And what we’re doing when we’re working with organisations like this is extending our expertise and developing strategies for the translation of our research into real world impact.
So how do I get to be thinking about automated decision making as a researcher who’s really come out of the area of digital media studies and communications? My original training actually is in the area of history and I’ve been interested for a long time in the history of new communication and information technologies. So what are their social impacts, how does the arrival of large-scale communication or information systems such as the postal system the broadcasting system or the internet change the way we live? That’s been a question I’ve been interested in for a very long time. I’ve been interested in the social consequences of the uneven distribution of communications and information. I’ve been interested in how we govern these systems. Whether through the law or through code or through other systems. I’ve been interested in the relationship between those unregulated parts of the media and communication economy, and those bits that are more regulated. So I’ve been always interested in this problem of uneven social distribution and uneven regulation and governance when it comes to emerging technologies. And of course in the area of automation this is what we see. And what interests me I think most about it is that larger view of what sort of transformation our country is going through and other countries as well, as we encounter these kinds of new automated systems. With my background in history I learned a lot about what went on and the kinds of debates that occurred in the 18th century, in the 19th century around the earlier phase of automation when machines were taking over the work of humans in what we would now recognise as the industrial revolution. That transformation sparked a revolution in not just in industrial affairs and in economies and in societies, but also of course in how we understand and think about these things and I think we can see the same kind of of change beginning to happen here. So I’m very interested in how this new next wave of automation, driven by computers and taking over the work of institutions in making decisions- how is that going to change the way we understand society and the world around us?