The Myth of ‘Perfect Knowledge’
The practice of voting in free and fair elections is a prerequisite for a thriving democracy.
As the pioneer of preferential voting, this is not news for Australia.[1]The Economist’s 2024 Democracy Index ranked Australia 11th globally.[2]In a world backsliding into populist leadership, our country is one of several in Australasia that recorded an improvement in the quality of our democracy.
However, theory does not always translate into reality. An ideal democratic model assumes that voters have ‘perfect knowledge’ of the representatives that they vote for. Even prior to Chat GPT’s launch in 2022, the belief that every constituent has informed, fully fleshed political beliefs is easily debunked; A 2019 Guardian poll found that only 15per cent of respondents follow domestic politics closely.[3]
With the proliferation of models such as GPT-4 and Gemini, generative AI threatens democracies by poisoning our information ecosystems, distorting truth, and eroding the foundation of a democratic society: trust.
Before we begin, what is Generative AI?
Generative AI (GenAI) is a computer-based model that can generate or modify existing content according to user prompts. It does this through machine learning, which recognises and draws upon complex patterns in large datasets.[4]
AI misinformation has inundated media channels, even spilling into political communication.[5] It ranges from comical ‘AI slop’ videos that rack up millions of views on TikTok,to more insidious deepfakes that hide in plain sight.
The Threat to Truth
Generative AI’s threat to truth is multi directional. For constituents, GenAI makes it difficult to monitor what elected representatives do, eroding democratic accountability. For public officials, it undermines efforts to gauge constituent sentiment, diminishing the quality of democratic representation.
One example of this was a 2020field experiment conducted in the United States where researchers sent35,000 AI and human written letters to state legislators.[6]The response rates to both were virtually indistinguishable, demonstrating AI’s ability to distort communications between representatives and their constituents.
The reach of misleading GenAI content is another cause for concern. According to IPIE, 80per cent of election-holding countries in 2024 experienced GenAI use during the campaigning period.[7]Of these incidents, almost half had no identifiable creator attributed. Over two-thirds of all incidents sought to degrade election integrity.
A growing number of Australians source their news from onlineplatforms where GenAI content proliferates. The Digital News Report Australia 2025 found that social media platforms have over taken online news as a main source of news (26 per cent), second only to mainstream television.[8]This is especially true for young people; Instagram is the most widely used platform for news among 18- to 24-year-olds.[9]
In addition to sowing misinformation, GenAI also weakens our democratic checks and balances. Search engines such as Google and Bing have incorporated LanguageLearning Models (LLMs) that siphon traffic from small or middling media organisations.[10]Forgoing trusted media organisations for AI outputs risks diminishing the media’s ability to act as a check against government power.
The Threat to Trust
So, what happens to our democracy when AI erodes truth?
Where truth is attacked, trust suffers. Trust is the foundation of a democratic system – it fuels political participation, public engagement and fosters confidence in democratic values. As put by German philosopher and historian HannahArendt, the ideal subject of totalitarian rule is one for whom the distinction between fact and fiction no longer exist.[11]
A healthy scepticism is often encouraged when consuming media. But in a world where inauthentic content can be created from a one-sentence prompt, healthy scepticism can easily tip into cynicism.
Giving up on discerning fact from fiction has flow-on effects for trust in public institutions. According to the 2025 Edelman Report,Australia’s trust index has dropped from neutral territory (51 per cent), to distrustful (49 per cent) in the last year.[12]
Our national rise in distrust goes hand in hand with an increase in apprehension towards misinformation. Australians’ concerns regarding misinformation is the highest globally at 74per cent.[13]
Generative AI may also reinforce existing divides. Edelman reports that four in ten of those surveyed see hostile activism as a viable means to drive change. This is higher in young adults, at 53 per cent.[14]AI’s promotion of partisanship on social media may be partly responsible for this generational gap. While social media increases political participation, it also increases polarisation and decreases trust in institutions.[15]
Defending our Democracy
How we can successfully combat AI ‘s effect on our democratic institutions is a question shrouded in uncertainty; AI’s impact on democracy is difficult to diagnose and hard to measure.
A preliminary step would be promoting a healthy information eco system through government collaboration with journalism. For example, in2016 Estonia’s state electoral office created an inter agency taskforce that worked with journalists to correct false messaging on its democratic processes.[16]
Education is another powerful inoculator against misinformation. Estonia’s additional incorporation of mandatory media literacy classes in school has paid off - in 2023, the Baltic nation ranked fourth in Europe’s Media Literacy Index.[17]
Promoting Australian involvement in international academic networks would also help inform policy solutions to combat AI-generated misinformation. Organisations such as the International Panel on the Information Environment (IPIE) provide guidance on how to safeguard against AI.For example, more than three quarters of experts in their 2025 Expert Survey overwhelmingly recommended labelling AI content to counter misinformation.[18]
Reaching people beyond their screens is essential. As mentioned in a report published by the Department of Home Affairs, there is a proven connection between engaging with the community and confidence in democracy.[19] Those who participate in cultural activities are more likely to vote, volunteer and uphold civic values.
Rather than destroy democracy, maybe AI can rebuild it.Sequestering empathetic and trust-building values into ‘pro-social algorithms’ can reduce misinformation and eliminate echo-chambers,rebuilding social cohesion.[20] When designed ethically and administered transparently, perhaps artificial intelligence can become part of the solution to strengthening our democracy.
Written by Annika Ramasamy
I am an intern at Small Giants Academy and a current Arts/Law student at Monash University. During my time at Small Giants, I developed a deep interest regarding the impact of artificial intelligence on democracy while working on the upcoming 2026 Wisdom and Action Forum: Trust in the Age of AI.
References
1. “2025 Edelman Trust Barometer: Trust and the Crisis of Grievance”, Edelman Trust Institute, 2025, https://www.edelman.com/trust/2025/trust-barometer.
2. “Democracy Index 2024”, The Economist, 2024,https://www.eiu.com/n/campaigns/democracy-index-2024/.
3. “Expert Survey on the Global Information Environment 2025: Safeguarding Epistemic Security”, International Panel on the Information Security, 2025, https://cdn.prod.website-files.com/643ecb10be528d2c1da863cb/68f8f39bd88d918638f1d67e_Expert%20Survey%202025%20-%20SFP2025.4.pdf.
4. “Generative AI in Electoral Campaigns: Mapping Global Patterns”, International Panel on the Information Environment, 2025,https://www.ipie.info/research/sfp2025-1.
5. “Media Literacy Index report: Measuring vulnerability of societies to disinformation”, Open Society Institute &European Policies Initiative, 2023, https://osis.bg/wp-content/uploads/2023/06/MLI-report-in-English-22.06.pdf.
6. “Strengthening Australian Democracy: A practical agenda for democratic resilience”, Department of Home Affairs, 2024, https://www.homeaffairs.gov.au/about-us-subsite/files/strengthening-australian-democracy.pdf
7. Allan Smith, “Trump posts AI video dumping ‘No Kings’ on protesters”, NBC News, 20 October, 2025, https://www.nbcnews.com/politics/donald-trump/trump-posts-ai-video-dumping-no-kings-protesters-rcna238521.
8. Amy Yee, “The country inoculating against disinformation,” BBC Future, 31 January, 2022, https://www.bbc.com/future/article/20220128-the-country-inoculating-against-disinformation.
9. Annabel Crabb, “Australia’s democratic system is unlike any other on Earth,” ABC News, 9 November, 2025, https://www.abc.net.au/news/2025-11-09/civic-duty-compulsory-preferential-voting-rules-aec-secret/105969502.
10. Hannah Arendt, “The Origins of Totalitarianism”, Schocken Books, 1 January,1951.
11. Katharine Miller, “Building a Social Media Algorithm That Actually Promotes Societal Values”, Stanford HAI (Human-Centered Artificial Intelligence), 8 April,2024, https://hai.stanford.edu/news/building-social-media-algorithm-actually-promotes-societal-values.
12. Katharine Murphy, “The big tune-out: few Australians follow politics closely, Guardian Essential poll shows,” The Guardian, 4 September, 2019, https://www.theguardian.com/australia-news/2019/sep/04/the-big-tune-out-few-australians-follow-politics-closely-guardian-essential-poll-shows.
13. Parket al. “Digital News Report: Australia 2025”, News and Media Research Centre, 17 June, 2025, https://www.canberra.edu.au/research/centres/nmrc/digital-news-report-australia
14. Sarah Kreps and Doug Kriner, “How AI Threatens Democracy,” Journal of Democracy34, no. 4 (2023): 122–31, https://www.journalofdemocracy.org/articles/how-ai-threatens-democracy/.
15. Suzanne Bearne. “AI Challenges the dominance of Google search”, British Broadcasting Corporation (BBC),16 September 2025, https://www.bbc.com/news/articles/c1dx9qy1eeno.
16. Victorian Department of Education, “Generative Artificial Intelligence: Introducing generative AI tools,” Policy and Advisory Library, 2024, https://www2.education.vic.gov.au/pal/generative-artificial-intelligence/guidance/introducing-generative-ai-tools.
