By Melissa Donnelly, CPSU ³Ô¹ÏÍøÕ¾ Secretary
Artificial Intelligence (AI) has the potential to be beneficial for public services, workers, and workplaces. It could offer greater efficiencies and free workers up to undertake more complex work by assisting with their mundane or repetitive tasks.
But it’s not all smooth sailing. And you don’t have to look much further than at the AI generated image of ‘salmon in a river’ to know that AI’s best biggest strength is the human who double checks it.
Whether the AI hype will live up to expectation remains to be seen, but it is clear the uptake of this technology is happening now.
The current use of Artificial Intelligence and automation in the Australian Public Service (APS) varies from agency to agency, but we know these capabilities are being used. And this is despite the absence of training and safeguards. The APS must catch up.
A recent survey by the Community and Public Sector Union (CPSU) found that over 40% of survey respondents are aware of the use of AI in their current workplace and 12% use it in their day-to-day work. It also found that 68% of people are finding it saves them time in doing their job, and 72% agree it helps them do their job more effectively.
So, we know people are using this technology, and we know there are potential benefits. The challenge now is how we manage and mitigate the risks.
That is, how do we harness the potential of AI in a way that fosters accountability, transparency, and serves the public interest and workers?
We do that by putting people at the centre of decision making. Consistently consulting workers and their representatives, providing training and supports, giving workers a say on the way they use AI in their work, and reviewing these systems hand in glove with the people who are using them.
If we are using AI with the goal of improving services for the Australian public and improving the work that APS employees do and how they do it, we’re staying on solid ground.
But if we use AI with the goal of cutting corners and cutting costs, the path ahead is riddled with risk.
We should be looking to use AI and automation technologies in a way where the intention is for them to amplify human intelligence and productivity, not replace it.
Any savings made by workers increasing their productivity with the assistance of AI should be re-invested back into delivering quality public services for every Australian.
The CPSU AI survey heard from almost 2,000 people working in the APS, across almost 90 agencies and departments. It provides a comprehensive snapshot of AI in the public service, including insights into the prevalence of the technology, its influence, and worker concerns about consultation, training and governance.
97% of respondents agree that workers should be consulted on the implementation of emerging technologies, but only 1 in 5 said they were consulted prior to its introduction in their workplace. Concerningly, 92% of survey respondents have not received any training on the use of AI at work, and almost half rated their current level of understanding of AI as poor.
There is a clear need for increased training about both the ethics and the use of AI tools. There is work to be done to better equip APS workers so that they are not only able to use the technology but can identify issues such as errors and biases and mitigate risks.
Concerningly, the technology is already being used in APS recruitment processes.
There are significant concerns regarding the ability of this technology to uphold merit principles, about built-in selection biases, and about the accessibility challenges it poses. Put simply, this technology is compromising APS recruitment practices, and its use in this area must be reconsidered.
That said, as we grapple with ethics and frameworks for the safe and responsible adoption of AI, it is important that the government and the APS are brave enough to identify when its use isn’t appropriate, and to rule it out. At least for the time being. If we do not have the necessary frameworks in place to manage and mitigate risk we must proceed with caution.
The word ‘Robodebt’ springs to mind.
Robodebt started back in 2015 and while it was belatedly wrapped up almost 5 years ago, it’s devastating impact is still being felt today.
This automated debt recovery system, driven by the goal to cut costs and speed up processes, bulldozed through without any human oversight. The result was a catastrophic failure of governance that caused immense harm to millions of Australians and eroded public trust in government and government services.
It is not a coincidence then, that the CPSU AI survey found that almost 80% of respondents were concerned that the use of AI in the public service could erode public trust.
Government decisions on social security, migration or the NDIS can and do have a significant impact on people’s lives and we must ensure accountability and human involvement in these processes.
An example of this is in Services Australia, where staff use their experience, expertise, and humanity, to assess the risk to an individual and how they can be supported, prior to making a decision that will affect them. This is something that AI or automated decision-making processes cannot do.
But lip service about retaining human oversight isn’t enough. Wherever these technologies are used, the public should have confidence that the final decision will always sit with an APS employee. We want to see the government mandate this protection in the public sector.
Australia’s public sector can lead in the ethical and responsible adoption of AI, serving both the public and public sector workers.
To do that, our approach must prioritise people – not the technology. And the government must be led on this by workers, now and into the future, who they have closely consulted, upskilled, retrained and supported to have the powerful tool that is AI, in their pocket to better serve Australians.
28th October 2024
