June 12, 2024 - 7:00am
An algorithm is standing for election in Brighton this year. AI Steve is the creation of businessman Steve Endacott, who lives in Rochdale but reportedly maintains a house in Brighton, and who promises that his AI avatar will be available 24/7 to listen to policy suggestions. Endacott himself would attend Parliament as the human proxy for his AI creation, to vote on issues as requested by his constituents.
It’s highly unlikely that “AI Steve” will succeed in being elected, leaving the cynic in me to wonder how far the whole project is a PR exercise. Endacott is director of a voice-to-voice AI enterprise, Neural Voice, which receives a logo and link on the AI Steve website, and is used to power the putative robot MP.
Neural Voice promises “Dialogue-Driven AI”, and its website suggests that such a robot might serve as customer service, medical adviser, even a lawyer. So why not also an MP? The fact that someone could seriously suggest this speaks volumes about the emptying-out of our understanding of political office — but, more troublingly yet, also the emptying-out of our understanding of human agency.
Of these, the first is far more complete. It’s not a coincidence that “comedy” Parliamentary candidates began appearing regularly in British politics from around Margaret Thatcher’s election in 1979: this was also the point at which de-industrialisation and the professionalisation of “civil society” began in earnest to hollow out 20th-century mass politics. There was a rush of joke candidates around this moment: first Auberon Waugh for the Dog Lovers’ Party in 1979, then the Monster Raving Loony Party in 1982 (and a further 40 “hilarious” years), then Billy Connolly’s wife Pamela Stephenson for Windsor and Maidenhead in 1987, representing the I Want to Drop a Blancmange Down Terry Wogan’s Y-Fronts Party.
That emptying out of political representation, to which these provided an ironic commentary, is now largely complete — which perhaps explains why Count Binface and his ilk seem so leadenly unfunny now. What they symbolise is more the rule than the exception. Meanwhile, in AI Steve, we can perhaps catch a glimpse of what might, in time, be offered in replacement.
The concept is simple enough: you talk to the robot, which collates policies with help from human sanity-checkers and Chat-GPT, then adopts the “best” or “most popular” of these as its political platform. It is a radically passive approach to political representation, which views this as the aggregation and rollout of emergent consensus, with minimal human quality control. To paraphrase another, non-AI Steve, the late Apple founder Steve Jobs, it’s the thinnest and lightest conception yet of what Members of Parliament do in practice.
And if that really is it, we might indeed replace them with robots: the most literal technocracy to date. But the obvious shortcomings in this proposal also highlight two aspects of “intelligence” whose absence in AI makes even a sophisticated instance so much less than a real human: judgement, and agency. An AI MP would be unable to judge for itself which policies are actually good; as AI Steve acknowledges, it needs human sense-checkers for that. And it can’t do anything to make them happen: in the context of Parliament, for instance, that might include in-person campaigning, meetings to bring other MPs onside, building coalitions or any of the other agentic actions that distinguish a human from a pattern-recognition robot.
To the extent that we accept an equivalence between human and AI intelligence which ignores judgement and agency, we aren’t just accepting interlocutors who are literally “not all there”, with all the frustrating customer-service interactions that will inevitably entail. We’re also, by extension, accepting a very much thinner and lighter conception of what human intelligence is. More, we’re embracing a hollowing-out with implications still graver than the fait accompli of our now largely evacuated Parliament.
A moment’s reflection reveals that this is already well under way. Its implications reach a great deal further than one PR-driven “joke” robot candidate MP.
Join the discussion
Join like minded readers that support our journalism by becoming a paid subscriber
To join the discussion in the comments, become a paid subscriber.
Join like minded readers that support our journalism, read unlimited articles and enjoy other subscriber-only benefits.
SubscribeFrom the underrated Person of Interest: Season 3 Episode 22 “A House Divided” 2015. An AI without democratic oversight
https://youtu.be/sYtkDUGnztA?si=MVyCHbqGVKgKceBv
“Comedy” Parliamentary candidates *did not* begin appearing regularly in British politics from around Margaret Thatcher’s election in 1979. They’ve been a feature of British elections at least since the 50s, which is as far as my knowledge extends. This isn’t “Fatcher’s fault” or caused by deindustrialisation.
From the 1950s onwards, Bill Boaks used to campaign on an armoured bedstead bike and a zebra painted car. He once stood as the Trains and Boats and Planes candidate, later the Public Safety Democratic Monarchist White Resident candidate. He was described as “nutty”. Unlike many other frivilous candidates from the period, he is remembered because as a side gig he was also a nuisance in the courts and he kept on going until the 2000s.
Then there is Screaming Lord Sutch. Somehow the author omits to mention he first stood for election in 1963 as the National Teenage Party candidate. The Monster Raving Looney Party of the 1980s was an evolution of a comedy concept that started in the 1960s.
Many more are lost to the mists of time, recorded only in local newspapers in the absence of a plural mass media that was interested in reporting such candidates. It is this that explains why the author incorrectly thinks it is a modern phenonomen that began in the 1980s.
Everything started with Margaret Thatcher including a lot of stand up comics careers,as one of them truthfully said,”All you had to do was stand on stage and shout ‘Margaret Thatcher’ “!
In a publicity stunt for his AI product, Steve Endacott proposes that ‘you talk to the robot, which collates policies with help from human sanity-checkers and Chat-GPT, then adopts the “best” or “most popular” of these’.
Is Mr Endacott’s approach really so different from our existing party procedures? At present, a focus group talks to a policy wonk who collates these policies (with some filtering and sanitising). Then the party’s Whips carrot-and-stick the MPs into supporting them.
The end result is much the same.
Al Steve. Sounds like a mafia don
AGI driven customer service agents will abolish all customer service phone queuing oce they can identify you. This technology has almost arrived and no Luddite with academic objections can stop it.
Politicians that listen to the electorate. Wow. I think he might have hit upon something…
You jest surely – ‘listening to the electorate’ actually is more data selection meeting a ever narrowing criteria.
I’ve frequently thought about the possibility of replacing our existing political establishment with robots – mainly in moments of dire frustration with the whole shebang when they all seem intent on talking about stuff which matters not one iota to anyone outside their tiny bubble.
But AI-driven politics would have the same, fundamental defect as AI in general, which is not a lack of intelligence as such, but a lack of emotional intelligence.This is a uniquely human trait (whereby some have more and some – like Rishi Sunak – have rather less…leaving D-Day celebrations early, what was that???) which robots/LLMs will only ever be able to replicate poorly, if at all.
Just as AI-produced texts tend to be circuitous, boring, factually incorrect and emotionally unengaging, AI-driven politics would probably mean policies driven too much by cold, rational, utilitarian considerations alien to the way humans actually wish to regulate their affairs.
For example: if you ask a machine how best to deal with the fact that you have a massive overhang of elderly people in society who need looking after physically and financially when the economy is tanking, then a purely rational answer from a machine might just be to go with mass euthanasia of people over 70. It’s an answer, it would solve the problem. But never in a month of Sundays are any sane human societies going to do that.
“…But never in a month of Sundays are any sane human societies going to do that…”
Well, there’s Canada which is getting there, but as you have put the qualifier ‘sane human societies’ I guess your premise still holds.
If we can successfully replace our elected representatives with AI bots, we can replace our electors as well. Worth a try.
AI definitely generates British political discourse on the NHS and public services. Same with the fondness for the European Court/Convention on Human Rights.
Yes, but why have they made AI Steve look like Keir Starmer’s louche younger brother?
Because many people would vote for Keir Starmer’s louche younger brother.
An AI is just a search engine able to create documents based on its results by collating them based on the vocabulary used – as distinct from just providing links to the source material. Expecting insight or innovation from an AI is like climbing a tree to get closer to the moon.
Mary’s point about the “hollowing out” of our society applies equally to many other aspects of our lives. AI references only the received knowledge that it has been given. It’s not unlike a human “genius” whose claim to that title rests solely on the fact that he’s memorized the Encyclopedia Britannica. (The fact that the internet contains vastly more info doesn’t change my point. Much of it is nonsense.) Information, no matter how much, is not knowledge, nor is it intelligence. You can’t reach understanding by reading up on any particular topic. It’s just not that simple.
So while everyone worries about AI monsters out to destroy the human race, the reality is that we’re already allowing it to make us into idiots. No need to grind us all up into fertilizer. Just our brains.
“An AI MP would be unable to judge for itself which policies are actually good; as AI Steve acknowledges, it needs human sense-checkers for that.”
Judging by the some of the policies attempted in Scotland I would say AI MPs have already been hard at work.
But presumably you have to PAY humans MONEY,so skip that bit.
A good way to bring back public executions then.
Parliament has ignored technological progress for more than a century. The country was divided into constituencies which each elected a representative, because it was impossible to collect people’s views on anything but a local level. We don’t now need constituencies, because most people vote for national parties. most people are incapable of naming their local MP because s/he is just a figurehead. In fact, the British system is verging on the Presidential as the country effectively chooses between the Labour and Conservative leaders. Alternatively, major issues could be determined by referenda.
The reason why we don’t fundamentally reform British politics is because a removal of constituencies would lead to PR and neither Labour nor the Conservatives want that. Likewise referenda are hated because the views of the British electorate could no longer be obfuscated by the media and the political class.
I’m not sure this molehill is worth building into a mountain. This is a pretty standard looking publicity stunt with the obvious purpose being to hawk a product. It’s an old trick. Take a new product and set up some contrived display of its abilities, using as much sensationalism as possible, then alert the media and get a bunch of free advertising and publicity. This stunt isn’t far removed from the traveling snake oil salesmen who set up a stage and waited for a crowd to gather, then hoodwinked the audience by using some magician’s trickery and/or pre-planted ‘volunteers’ to greatly exaggerate the product’s efficacy. That’s all that’s going on here. This is not some grand plan to replace our elected leaders with AI robots or redefine humanity’s relation to machine. If it were, I expect it would gain more traction, as the current governments of the UK and USA are about as unpopular as any government ever, and any serious alternative, however bizarre, would probably enjoy a certain amount of default popularity on that basis, particularly on Unherd.
Why do you need stand alone crackpots of the past when almost all political parties are now monster loony. How prescient some of those political pioneers may have been now that nonsense reigns. Perhaps one other thing we could with from the past; humour.
I thought politician’s already were robots, albeit robots with principles. And if it turns out that those principles become inconvenient for some reason, they can quickly be replaced by others.
One of my concerns about AI ‘candidates’ in politics etc. is what might happen if someone could claim to have revived a (mostly revered) figure from the past, based upon all they said and did, and could then have this avatar pronounce what policies they would follow, a sort of digital ‘What would Jesus do?’ -type of thing.
Imagine if we could “know” for certain what Margaret Thatcher (I know I’m on Unherd here…), would do about immigration. Or Nelson Mandela, or *whoever*.
I felt a wave of grief and sadness rise within me upon reading your article Mary.
“you talk to the robot, which collates policies with help from human sanity-checkers and Chat-GPT, then adopts the “best” or “most popular” of these as its political platform.”
Is this not yet another linked-in development of the near complete erasure of the human. Another link in to the phenomenon already highlighted here – the Global Disinformation Index, sheparding and erasing any dissent to correct ideological ‘human thought’. The so-called panacea is becoming the dis-ease.
We have lost our central tenant to our ‘collective good’ – sorry to state the bleeding obvious again. Further down the tube though – where for art thou children……………