Forget chatGPT, meet chatMP

Adam Bell

Our Director of Policy and self confessed AI nerd has taken a look at the possibilities and pitfalls possible with greater use of the evolving technology in the political sphere.

“Such a thing (a) cannot work, (b) is immoral, (c) should be declared illegal.” These were the partially prescient words of an adviser to the Kennedy campaign after being pitched the world’s first computerised model of voter preferences, the Simulmatics ‘People Machine’ [I). This was an effort to pull together all the bits of data available on voters, who they were, what they wanted, and how they felt, and then try to predict how they would behave following a particular intervention.

Unfortunately for that adviser, the track record of applying data analytics to political campaigns is a successful one, albeit one fraught with controversy. The Cambridge Analytica scandal, in which millions of people unknowingly had their data harvested for the purposes of political targeting, brought the fact that political parties globally have continued to employ techniques like those developed by Simulmatics into the spotlight.

On the face of it this is easy to understand. The kind of population and message segmentation this approach implies appears to deny individual dignity. Arguing with a politician on your doorstep carries an authenticity which being peppered with tailored statements by a machine does not.

But the practicalities of political communication have compelled this change. In order to talk to every single one of their 100,000 constituents for five minutes every year an MP would need to never eat, sleep, or indeed vote on legislation. These are statistics that oblige the death of authenticity in politics, and your MP instead finding ways to connect with their voters that do not rely on individual conversations.

So: letters, leaflets, emails, and social media. But if voters only have a limited amount of time, any communication must be carefully tailored to ensure that it means the most to that voter. This tailoring is where worries about manipulation come in, but many years of analysis have indicated that you cannot get people to act against their fundamental values. This is why communications instead seek ways of emphasising that MPs and their policies share their values. Rishi Sunak’s recent statement that he is “on the side of motorists,” reflects the importance of control for many people across the country, control they feel they are losing in the cost-of-living crisis, and the fact that they find a sense of control and liberty behind the wheel of their vehicle. You may feel people are wrong to value their cars so highly, but if so, the task is not to condemn people for having the wrong values but instead figuring out what non-car alternatives will give them the same sense of control.

The real-world opportunity presented by AI MPs is that the enhanced formula of personalisation, relevance, and proximity could lead to much higher levels of engagement especially for a disappointed, disheartened disconnected younger audience.

AI allows this kind of value segmentation to go even further, to provide content beyond just text messages precisely tailored to not just the segment you happen to inhabit, but your precise mix of values, concerns and worries. Perhaps a little further in the future – although not too much further – politicians may choose to experiment with converting their written and recorded material into training data for an AI capable of producing videos and text. It is very easily foreseeable that rather than deepfakes [II] being something politicians are desperate to regulate they may represent a political opportunity: interactable AI-generated videos of politicians to be used for campaigning. Everyone can finally have a conversation with their MP, of any length they choose.

Of course, politicians will be tempted to tweak the AI version of themselves to render them more agreeable than their real-world counterpart, and it is only a small step from there to envisage a politician entirely generated by AI to represent a party’s political platform. A Prime Minister able to be present in every living room, providing an explanation of what they’re doing in language tailored to that person’s values, would be a formidable opponent indeed.