‘Artificial intelligence’ (AI) is a hot topic in policy making.
Data usage, privacy and algorithmic bias are the focus of discussion in numerous policy settings. The regulation of emerging technologies, such as facial recognition and autonomous vehicles, is also receiving significant attention. In addition, there is some talk about ethics – in the 2018 May budget, the Australian Government dedicated $2.3 million to the development of an AI ethics framework.
It’s great that Australia is starting to talk about these things.
However, these conversations are still focused on the technologies and concerns of the here and now, and the not-too-distant future. We need policy makers who can look past these immediate concerns and actively engage with a future that is rapidly approaching. Importantly, we also need to equip them with the right skills, knowledge and resources. These may not yet exist, but we are working hard to fill this gap with a new applied science.
We are looking for collaborators to help build it.
The future is here
“We are now talking about systems of computational objects that aren’t waiting for us to program them, but where human and machine-generated data drives decision-making and action.”
Around two years ago, the World Economic Forum published an interesting characterisation of the future we are beginning to see unfold around us. They refer to it as the fourth industrial revolution.
The first wave of industrialisation in this particular framing involved the steam engine and mechanisation. Machinery did more than our bodies could do – it could move faster, push harder, lift more. Machinery allowed us to scale human labour in new ways.
The second industrial revolution is pegged to electricity and mass production. Think of a factory line. It wasn’t just about the human body pushing machinery, but that machinery was now using electricity to drive production timelines faster and faster, day and night.
The third wave of industrialisation, post-World War II, came with electronic computers. The first electronic computers were just calculators; simple computers which worked with numbers at great speed, through a combination of automation and electricity.
While modern computers have more complicated programs, they continue to be underpinned by command and control architecture. Programs tell computers what to do and they do it.
Now, according to the World Economic Forum, we are entering the fourth wave of industrialisation. That fourth wave builds on the others, and is characterised by cyber-physical systems.
With cyber-physical systems, we move from talking about command and control to talking about AI, first by talking about algorithms and then new forms of learning. Computers are now small enough and cheap enough to embed into many everyday objects, connected as the internet of things. Many of them sense and react to data in real time, learning from their environment rather than following a rigid program.
We are now talking about systems of computational objects that aren’t waiting for us to program them, but where human and machine-generated data drives decision-making and action.
We need something new
“Despite the hype, AI is no more and no less than the steam engine to this next industrial revolution.”
Presented like this, these revolutions seem straightforward, logical. But the reality is that the real world applications are not so tidy and simple.
In each of the industrial revolutions already past, there were massive social upheavals and dramatic political transformation – what got made by whom and in how it was circulated, consumed or rejected. We created new academic disciplines to contend with these things, new forms of public policy and new forms of regulation. Wars were fought over and using these technologies.
Despite the hype, AI is no more and no less than the steam engine to this next industrial revolution.
It’s not the end game, just a tool or method that will make things possible. It isn’t even a single technology but rather, a constellation of learning techniques, algorithms, sensing technologies and data sets. And it certainly isn’t magic – it is mathematics.
Like the steam engine, it only gets interesting when you know its potential – what it can do or what systems it might power.
For a long time, steam engines sat next to mines and pulled water out of holes. When the same technology was transported into trains, the conversation suddenly expanded beyond the engineers behind the engine. With this new application, we needed to include people who thought about train construction and train tracks and camber and grade and safety.
The popularisation of railways then led to lengthy discussions about the creation of Standard Time, ticketing systems, and international trade. It opened new possibilities for tourism and labour force mobility.
If AI is just the metaphoric engine for next generation cyber-physical systems, where does that leave us?
From talk to action
Each of these industrial revolutions would have benefited from earlier and more proactive policy making.
We want to avoid a wave of technological change for which we are unprepared. From experience we know that this leads to people getting hurt. We over-correct in a legislative or regulatory sense. We end up curtailing the technology before we get it right. We need to think about this from first principles all the way through.
The Australian National University (ANU) is initiating this effort. The newly created Autonomy, Agency and Assurance Innovation Institute (3Ai) aims to apply science to the process, thereby promoting safety in the management of cyber-physical systems as they go to scale.
Work has already started in scoping the intellectual framework for the new applied science, and exploring and defining the key questions and our research agenda.
Our ambition was to launch our first educational program in 2022; however, we’ve realised the world needs the new applied science now.
We have prototyped a curriculum and are now accepting applications for an intensive 12-month Master’s-level program, commencing February 2019. We are designing by doing – the first year of teaching will be experimental as we apply start-up design methodologies to education. We will iterate and build the curriculum in response to diverse feedback and the changing technological environment.
We are currently recruiting students for the year-long program. We will cover tuition fees for students of the trial course and provide scholarships to offset some of the costs of taking the year out to help build Australia’s future. We need public servants among them – the role of government in shaping the future cannot be underestimated.
To find out more about the postgraduate program with the 3A Institute at the Australian National University, visit https://3ainstitute.cecs.anu.edu.au/#apply