The relationship between AI and democracy is the topic of many increasingly urgent conversations taking place around the world. The head of Australia’s election regulator has warned the organization “does not possess the legislative tools or internal technical capability to deter, detect, or adequately deal with false AI-generated content concerning the election process.”
This remark, made to a senate committee on adopting artificial intelligence (AI), is not an isolated comment.
More than 60 countries will head to the polls in 2024, in what has been dubbed “the biggest election year in history.” Australia is expecting to hold elections in the Northern Territory and Queensland this year, with a federal election due by May 2025.
At the same time, the explosion of generative AI tools for text, images, audio, and video is dramatically shifting the way Australians create and engage with information. How can we maintain the integrity and trust of elections in the age of generative AI?
Deepfakes and disinformation
The most obvious risk AI poses to democracy is via synthetic content (or “deepfakes”), which could be used to misinform voters. A World Economic Forum survey conducted last year found experts ranked “misinformation and disinformation” and “societal polarization” as the first and third most severe global risks over the next two years.
These risks are already manifesting. In the United States, a political consultant, who used the synthetic voice of U.S. President Joe Biden in robocalls, faces fines of several million dollars. In India, AI-generated videos have become increasingly common in this year’s election campaign.
But deepfakes and misinformation are far from the only risks. AI also presents new opportunities. In the evidence I gave to the senate committee, and in a submission from me and my colleagues at the Tech Policy Design Centre, we argued that a fuller national conversation on this topic is essential.
A broad view of a healthy democracy
A comprehensive policy will not focus purely on deepfakes swaying votes, but the health of democracy more broadly. Free and fair elections are one characteristic of democracy (albeit an incredibly important one), but there are many others.
Informed civic engagement, tolerance, and political pluralism are other important ingredients of a thriving democratic system. The system also needs to identify and respond to the needs of the electorate, and the government must be transparent and accountable.
So when we consider the relationship between AI and democracy, we need to consider perennial concerns such as political representation, public interest journalism, media literacy, and social cohesion.
Risks — but also opportunities
A balanced policy should recognize AI technologies present opportunities for democracy, as well as risks.
For example, it is absolutely reasonable to worry that informed civic engagement may suffer due to the tendency of generative AI models to “hallucinate” and produce misinformation.
However, the very same technology can also engage more voters in civic discourse: It can convert complex policy concepts into relatable content, or create automatic translations into many languages.
Learning from the past
Some elements of the challenges we face are not as new or certain as they seem.
AI, and the generative AI boom in particular, certainly injects some unprecedented elements into the democratic ecosystem. But there are still lessons we can learn from the past.
Anxieties about technologies that make novel forms of communication widely accessible are not new. Nor are efforts to regulate and control who can influence public flows of information.
In the 15th century the invention of Gutenberg’s printing press stoked fears over what we might now call “fake news”. There are plenty more recent examples, including what we have learned from the rise of social media (which is in turn shaping the story of AI and democracy).
We need to be discerning about the elements of generative AI that are fundamentally new. At the same time, we can look for applicable policy tools and lessons from previous information technology revolutions.
What now for Australia?
Australia is standing at an interesting crossroads. Up to a year out from our next federal election, several interlocking branches of policy are in development.
The department of industry is working on a response to last year’s Safe and Responsible AI consultation. This will include considering rules for compulsory watermarks in AI-generated content.
The department of communications is also reworking proposals for new powers for the Australian Communications and Media Authority. These will help to combat misinformation and disinformation, and may include AI-specific measures.
At the same time, the Online Safety Act 2021 is being reviewed. This may result in powers to address online abuse of public figures, which again may involve AI.
I propose four key actions the Australian government should take.
First, it should develop a coordinated national approach to the relationship between AI and democracy. My colleagues at the Tech Policy Design Centre have offered more detailed recommendations on how to coordinate the development of national tech policies.
Second, the government should pay close attention to the dozens of national elections around the world this year. We can monitor the success or failure of different policies in different contexts to learn from the experiences of others.
Third, we can learn from South Korea by requiring politicians to disclose and watermark any deepfakes or other AI-produced content used in election materials. South Korea has barred politicians from using AI-generated materials in their campaigns completely. However, the lower bar of requiring Australian politicians to be transparent may be less controversial and easier to implement.
Fourth, the government needs to ensure the Australian Electoral Commission and the Australian Communication and Media Authority have the staff and resources they need. Their task of tackling emerging challenges, including those posed by AI, and equipping Australians to engage with a complex information landscape in the year ahead will not be easy.
These steps should be just the beginning of a comprehensive, balanced, and informed national conversation about how we can support Australia’s democracy to flourish in the age of AI.
Zoe Jay Hawkins, Head of Policy Design at the Tech Policy Design Centre, Australian National University
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Follow us on X, Facebook, or Pinterest
