In my last stint in a newsroom, which coincided with Donald Trump’s 2016 presidential win, I observed a bending of reality. The result was stunning and hard to believe – Trump wasn’t a politician and actually lost the popular vote. Then in the aftermath something strange began to occur. ‘Alternative facts’ were invented, as was the more Orwellian sound ‘post-truth’.
Fake News
Media outlets sought to explain Trump’s victory through pointing towards fake news, which appeared to be running riot on Facebook. It was if there must be some ulterior and insidious explanation. Collins Dictionary named ‘fake news’ their ‘Word of the Year’ in 2017, after Google searches for the term shot up.
Trump inverted the term and used it to attack ‘the fake news media’. It wasn’t so much that fake news on social media was his friend, but that the established media was his enemy. I dug into the fake news rabbit hole, appearing on think tank panels and writing articles, but could only conclude that it gave Trump little advantage in the 2016 election. Social media is simply not where the political conversation starts or ends in America – cable TV is. In the aftermath, Trump’s use of it against his very attackers was a masterclass of post-truth deflection. Google searches for ‘fake news’ in the US peaked in early 2017, as Trump was inaugurated.
Whatever Trump’s war with the established media, fake news was a PR disaster for big tech. For a long time there had been accusations of filter bubbles and privacy concerns, but after the shock victory, news media also went after Facebook and Google. Mark Zuckerberg was summoned before a Senate Committee in April 2018 to explain what his company was doing to quell fake news and the harvesting of data by Cambridge Analytica.
In the event, not much came out of it. It didn’t seem like Facebook was any more likely to be regulated. After all, it was the Trump administration in the White House. But then Trump lost the 2020 election with a backdrop of misinformation about coronavirus. Under the Biden administration, Big Tech would have to clean up their act.
Generative AI
By the time of Biden vs Trump 2024, the game had changed again. From 2022 onwards, generative AI had made significant onward leaps. Midjourney arrived over the summer, and within a year showed it could produce photorealistic AI generated images. ChatGPT burst onto the scene in November – a kind of ‘holy shit tech’ moment where the very nature of productivity could be transformed. Both of these developments – in image and text creation – raised questions of what could be interpreted as real.
2022 could be seen as the start of what Microsoft AI CEO Mustafa Suleyman has defined as The Coming Wave. He has written the essential macro AI bible, addressing (to use the subtitle) Technology, Power and the 21st Century’s Greatest Dilemma. Within it, Suleyman makes it clear that, at least in his view, AI represents a profound historical shift. Google CEO Sundar Pichai said that it could be more profound than fire or electricity.Mustafa Suleyman, cofounder of Deepmind and CEO of Microsoft AI.
But underneath the potential for revolution, there is a fear. Suleyman is an AI optimist, but on first read The Coming Wave often felt like a tome of dread. His insight into how autonomous AI agents could run riot over commerce, politics and defence had me rather concerned at times. It is quite clear that the ageing democratic states of the West are already creaking.
Had everything been fine and dandy, you could have hardly seen Brexit ever happening, or indeed Trump appearing much at all. Instead the two have dominated Anglo-American politics for nearly a decade. Despite my view that fake news did not have much electoral sway, post-truth is a key element of populist malaise and a rising sense of cynicism. At many rallies Trump sways into rants about the fake news media and tall tales of immigration.
There were great fears that AI would change the face of this coming election. In 2023, concerns grew of the potential of AI avatar representations of the candidates spouting nonsense and humiliating themselves. With the ageing and at times stumbling Joe Biden, this appeared a distinct possibility, but once he dropped out of the race it had nowhere near the same potential. Indeed, in the campaigning after the main summer party conventions, Artificial Intelligence has barely been mentioned in the political arena at all. Trump did mention it in an interview on Fox, saying, ‘it is so scary… Something has to be done about this and it has to be done fast.’ Despite his vague protestations, it hasn’t made much impact at all.
Why AI hasn’t had a significant political impact
Firstly, and most fundamentally, the technology to make extensive believable deepfakes isn’t readily available. To create a ‘deepfake’ you need a substantial amount of footage, verification and an understanding of nascent technology. This isn’t the type of thing you can pull off quickly with an avatar app like HeyGen or Synthesia. It takes a substantial amount of skill to deep fake a political candidate with something that will actually be convincing. Secondly the major VC backed platforms have skin in the reality game, and being used to this end would be a disaster – so they’ve policed it effectively. You cannot generate an image of any presidential or vice-presidential candidate on Midjourney, as an example. With fake news there was a much lower barrier to entry.
Connected to this point is that people are also wary. Having felt a sense of duplicitousness around politics and media for many years, and trust in established media collapsing, they are more likely to call it out. You see an AI, then the first comment, with the most likes, is likely to be something like ‘this is AI’. Indeed, there’s a wider backlash against AI usage in general as creatives have increasing concerns about its potential to replace human made art and livelihoods.
Third is that after the giant fake news PR disaster of 2016, and the increasing threat of regulation or breakup, big tech has shifted to more self regulation vs its more libertarian past. From what I can tell via its demotion of links in its algorithm, Facebook has largely abandoned hard news. I have made three election posts on LinkedIn (owned by Microsoft) and seen their engagement relatively crushed vs my normal output. Similar to Facebook, the news link on LinkedIn is close to irrelevant. YouTube (Google) and Instagram (Meta) have required users to clearly label AI generated content – sometimes doing it automatically. An AI to counteract AI, if you will.
Lastly, with the absence of Biden, the two candidates are difficult to fake. Harris is a confusing and awkward communicator, but she is not likely to have freezes and long silences. Trump, meanwhile, has substantial difficulty with the truth already. It’s difficult to imagine what a fake version of him could really achieve.
However, the race is extremely tight and consequential. The candidates are neck and neck. But should Trump win, don’t be surprised if the misinformation flag is waved by the established media again – this time with added AI. Yet should this happen, don’t expect Trump to come after it. AI, ironically, is likely to be more regulated in the event of a Harris win. Should it not be, the technology will be much more ready to run wild in 2028.
