A Generative AI Election - The New Frontier

“Personally I think the idea that fake news on Facebook, which is a very small amount of the content, influenced the election in any way — I think is a pretty crazy idea.” – Mark Zuckerberg, 2016.

The echoes of Zuckerberg’s statement back in 2016 resonate loudly today.

What may sound less crazy now is this: The 2024 US election cycle could possibly be the first authentic AI election. Donald Trump and Gov. Ron DeSantis have already used deep fake images of each other and we are still 16 months out from the vote.

This is an election where the capabilities of Generative AI aren't merely speculation but an influential reality. We understand its potential, we anticipate its use, yet governments appear largely indifferent to its probable role in election campaigns. The potential repercussions on geopolitics and the global economy are staggering.

It's almost a certainty that the forthcoming election will transcend the era of Facebook ads. We are venturing into a realm where an underlying uncertainty will accompany everything we witness. Our world has advanced past rudimentary deep fakes into a phase that we term in AI as ‘No Noticeable Difference’.

Democratization of Disinformation

Unless we personally attended an event, we may forever question its authenticity. Considering this evolving reality, along with the apparent legislative apathy, it's crucial that we understand the redefined landscape of politics in the AI epoch.

Historically, election interference has been a costly and challenging endeavour, typically orchestrated by states such as Russia, China, and North Korea. These entities have effectively sowed doubt in rival state elections, at the very least diminishing confidence in the electoral processes of democratic nations.

Now, anyone with a laptop and personal election agenda can fabricate anything they choose.

Near-Perfect Duplicates

Generative AI thrives on the data set it's trained on. In this context, politics is exceptionally vulnerable. The public life led by candidates offers a treasure trove of data, with endless recordings across media platforms providing rich training material. Combine this with the readily available Generative AI tools capable of generating near-perfect duplicates of voice, video and images, and the populace at large possesses the means to create what could be accepted as 'real' by an unsuspecting voter.

The true democratization of a technology is signified when it becomes integral to the political process. We've witnessed this evolution with print, radio, TV, and social media. The next stage in this progression is the advent of Generative AI.

Essentially, the subterfuge we've seen in electoral manipulation is set to intensify exponentially. Digital falsifications will be of higher quality. Microtargeting will be significantly more potent, possibly reaching the granularity of individuals through AI agent scripts.

Political advertising can graduate from being hastily cobbled together talk pieces to cinematic-quality productions potentially painting dystopian visions of the opposition in power, thus inducing irrational fear. Those who once lacked resources can now participate on par with well-funded political entities.

However, it's the subtler uses and their societal implications that warrant our attention.

Subtle Social Implications

For conscientious voters, it's usually feasible to expose a fabricated picture, video, or statement, particularly if it purportedly originates from a political event or speech that never transpired. Mainstream media generally does a commendable job at fact-checking in this regard.

Yet, if present-day AI is used to redub a speech, subtly modifying a few words or a sentence to distort a candidate's message, with impeccably mimicked facial movements, the game changes entirely. Consider the scenario where this happens during a presidential debate, with the manipulated footage making its way into the multitude of YouTube highlights reels. Discerning the truth from the fiction could become difficult indeed.

Furthermore, once this influx of fake information permeates the political landscape, we risk descending into an era where cynicism usurps belief in everything we see. This could be as detrimental as its opposite. In such an environment, a candidate could easily deny any factual information by claiming it is an AI-generated fiction. This would especially be the case with controversial or scandalous footage never intended for public consumption.

Legislative Solutions

Just this weekend, Communications Minister Michelle Rowland released draft legislation to give the Australian Communications and Media Authority (ACMA) powers to hold digital platforms to account for spreading harmful fake news. While it seems like we are finding our teeth on false information with potential fines being as high as 5 per cent of global revenue – which would be $8 billion for Meta platforms – the law will be retroactive. The proposed laws focus on a platform’s failure to combat misinformation and disinformation.

Even after a take down, the internet never forgets and digital copies can spread freely. Add to this a business which relies on unfettered access for anyone to publish anything, and it is hard to see anything changing. It is a post action control.

A Blockchain Moment?

It often turns out that an unexpected use case emerges with a technology searching for a home. This could well be a “Blockchain Moment” – if we have the wisdom to embrace it.

Blockchain technology's key feature is actually immutability. Hence, it can be used to combat fake information during elections by creating a decentralized, secure ledger of data. This would allow any digital content, like news articles, photos, videos or social media posts, to be 'hashed' and then have that hash stored on the blockchain at the time of creation. This hash serves as a unique, unalterable fingerprint for that piece of content.

Later, if someone encounters this content, they can hash it again and check the blockchain for a matching hash. If a match is found, they can be sure the content is unchanged from when it was first registered.

In the case of an election, this could be used to verify the authenticity of candidate speeches, ads, debates or other related content, ensuring voters are making decisions based on accurate, unaltered information.

However, this would require a common standard and widespread adoption to be effective. Also, while blockchain can help verify content hasn't been altered, it doesn't guarantee the truthfulness of the original content put on that blockchain – which would again lead us to need some kind of centralised, trusted institution to post the original content.

Still, blockchain may finally find a widely accepted use case beyond crypto coin speculation.