To combat AI at election, conversation may be key
Former Green MP, works for the Wellbeing Economy Alliance Aotearoa. Not a member of any political party
How will AI negatively impact the 2023 New Zealand election? ChatGPT, the AI engine at the centre of the current controversies around machine intelligence, itself suggests the negative impacts could be disinformation and manipulation, deepfake technology, biased algorithms, voter profiling and microtargeting – proving that ChatGPT might be more self-aware than the average beltway politician.
Christopher Luxon may have been initially surprised his party was using AI-generated images on its social media pages, but he offered a fullthroated defence, saying it’s the same as buying a stock image – but smarter. I think he’s wrong: it is not the same as using stock images, where you pick from a catalogue of existing assets. With AI you provide the prompts to create something new, but fake. This leads into issues of trust and veracity, not least artistic integrity, where AI acts by scraping the artistic and intellectual property of human creatives, most of whom have given no permission, and received no compensation, for having their art appropriated.
I’m not shocked parties are using tools like ChatGPT and Midjourney. For National it helps avoid its persistent copyright infringement problem and it’s cheaper than using human creative labour. It would be nice if the party, with its overflowing war chest, chose to employ local creatives, but synthetic content over real people avoids Labour’s embarrassing problem – its 2019 Wellbeing Budget cover model had moved to Australia over New Zealand’s high cost of living.
In the 2020 election we saw chatbot technology, a primitive form of AI, helping inform young voters about the registration and election process. In 2023, AI will help politicians produce images, write speech drafts and crunch data. It will be our first AI election.
Politics has always had to adapt to new technologies. If there’s a new trend it is that technology is speeding up exponentially – and we have not had the public discussions that would increase awareness of the benefits, and the downfalls. We are sleepwalking towards a brave new world.
As a young campaigner in the early 2000s, engaging politically with new technology was thrilling. In 2005 it was blogs, in 2008 it was Facebook and YouTube and in 2011 it was Twitter. At the time I believed social media was a powerful new democratic tool to circumvent the traditional media gatekeepers and converse directly with voters.
In the early days it took a willingness to engage, a degree of authenticity and a dash of creativity to reach large numbers of people. But over time the tech giants became the new gatekeepers. Focused on wringing every last cent from their platforms, their algorithms encouraged polarisation, extremism and the worst angels of our nature. In light of all this, I no longer believe democracy is best-served by social media.
Regulation and transparency haven’t kept pace with the power users have to abuse their ability to instantaneously reach millions of people.
National’s use of AI images has sparked a debate, but the problem isn’t technology – it is the dark motivations of actors. It was dishing up disinformation well before it had a website. Many New Zealanders will remember National’s ‘dirty postcard’ attacking the late Jeanette Fitzsimons in Coromandel in 1999. The Exclusive Brethren came close to forcing the Greens under 5% in 2005 with its well-funded secret National campaign, only exposed as voting commenced.
In 2023, AI will help politicians produce images, write speech drafts and crunch data.
Come October this year a deepfake video, photo or audio file of a party leader could prove decisive as voters cast their ballots. We must remain vigilant and sceptical – we will have to question any extraordinary content without extraordinary evidence.
As politics increasingly becomes a technological and fundraising arms race, we need to better respond to the impacts of new technology through better regulation, codes of conduct and transparency. Thanks to Midjourney’s Discord server we can see the images and prompts behind National’s AI ads. We need more information in general about who, how and how much parties are spending to target voters.
The best response to modern technology negatively impacting our politics may be to encourage that most human of actions – talking. We need more face-to-face opportunities for citizens and representatives to genuinely converse, more avenues for ordinary people to have their opinions heard above the virtual shouting of online discourse.
Other countries use citizens’ assemblies and tools like participatory budgeting to good effect. We can’t put the AI genie back in the bottle, but we can design modern democratic spaces to fairly and positively express the collective wisdom of the people.