I'm just getting an infinite reCAPTCHA loop on that page.

Quite a shallow article it seems to me, almost like parody on what could be a decent analysis of the impact of AI on the politics and elections.

How long until we get AI politicians? All over the world, the human ones provably suck[1], we need something better. It's no longer a metaphor that democracy has failed, it really has.

1 https://www.weforum.org/videos/trust-in-governments-is-decli...

Not sure why you are downvoted, I think you bring up a good point about AI politicians.

Although, if you think democracy has "failed" you should see how much worse the next best solutions are.

> Although, if you think democracy has "failed" you should see how much worse the next best solutions are.

reminds me of the demotivator "Government: if you think our problems are bad just wait until you see our solutions!"

I doubt their proprietary AI replacements would be much of an improvement, who controls the AI?

The point isn't to be an improvement, but to create an entity that can get all the blame for free, to be then replaced by the same entity with a different face and voice.

a simplistic reading of that poll is that people think corporate AIs are better than governments now.

That may be true for now, because generative AI models are still at the "look at all the cool things we can do" stage of development. But like all new technology, they'll soon reach the value engineering stage - when they'll be "optimized" by removing quality, cutting corners, and papering deficiencies over with marketing, until - like all inventions before - they'll become the worst possible version of a product the vendors can still call "fit for purpose" without being laughed out off the court.

Enjoy it while it lasts, but try not to make anything critically dependent on it, as soon enough, we'll all be regretting being stuck with the inferior facsimile, victims to the usual market bait&switch.

I really wish US politicians would adopt some truth standard in campaigning. I know they won’t do that but it would be really nice if the constant blatant lying would stop. I don’t think AI will make the situation much worse. If anything, it may force the creation of some standards.

> I really wish US politicians would adopt some truth standard in campaigning.

Sounds good. I hereby volunteer to serve as the arbiter of truth for political campaigns.

Well, how do we know that your truth is truer than mine? There has to be a formal certification process. It can be called a “certified truth sayer”, and only those with such a certificate can tell us what’s true and what’s not. And to get certified you need to take a course and pass a test, which are managed and administered by accredited NGOs running on grants from the government and private entities.

I think that would be ridiculous. It’s much better that everyone drown in a sea of noise and dubiousness and disinformation than to ask that we our trust in some central entity that can be manipulated or audited or influenced or held accountable.

What is the US legal system?

Obvious questions for $400

Ding ding ding. Anyone can bring a case to court where multiple third parties can evaluate the claims. And some people lose their case 60+ times and continue to lie as if the evidence was on their side.

During the whole election fraud situation a lot of republicans wouldn't publicly say that the allegations were nonsense although they privately knew that there was no massive fraud. So with truth standard I mean an ethics standard where it's just not OK to lie blatantly. Basically have some integrity.

I’m generally of the view that voters get what they want (even if what they want is internally inconsistent, not in their best interest, or they have to contort their worldview or values to justify it).

My point being, politicians aren’t honest because they’re not penalized by voters for being dishonest.

You reminded me of a documentary I've seen of some former mining town in US where everybody was against Obamacare and liberals, and 30% the county was living on food stamps and could not afford any kind of insurance.

People literally voted against their interests just "to own the libz".

Getting people to vote against their own interests is the central goal for at least 1/2 of all politicians.

I can’t find the study but voters were surveyed regarding their spending priorities as well as their views on balance, budget and deficit spending, and the gist of the study was that most voters wanted more spending, less taxes, and a balanced budget, which is logically inconsistent (but they are getting exactly what they want (even if they complain about it)!

It will absolutely make these types of tactics (seemingly employed by only one party, you know which one) far more personalized and therefore more effective:

https://apnews.com/article/election-2020-technology-arrests-...

https://theintercept.com/2017/11/07/virginia-voters-get-myst...

Not at all. It'll just allow the politicians to blame the lies on the ai

The issue isn't that politicians lie, but that we are numb and don't hold them accountable for it.

Politics are only a reflection of society.

I don't think they are just a reflection. They also shape society.

One could even say... they have a dialectical relationship

Seeing them qualify their statements with "I will renounce my position in office and consider myself a disgrace to my party if I do not fulfill ____". They could still reneg, but it's putting up some collateral. Remember how much bad PR Bush I got for "read my lips" ?

We’re going to see a click of pols try to replace speechwriting or social media management with ChatGPT and it’s going to have hilariously bad results. The Waluigi effect is going to make for some interesting policy arguments.

We have had politicians using radio.

We have had politicians using television.

We have had politicians using Twitter.

Politicians using large language and diffusion models was only a matter of time.

I agree that it was a matter of time, but I think there's an interesting conversation to be had about how each of those three technologies changed the way elections were run.

Radio allowed for real-time, direct address to the people for the first time.

Television allowed voters to see the candidates' faces as they gave speeches or responded to questions for the first time.

Twitter allowed candidates to broadcast their thoughts and real time, and interact with voters' questions in real time.

I don't think the question is whether language and diffusion models will be used politically; of course they will be. But, how will they affect and change the discourse as they are used? That's a conversation worth having.

You can just have LLMs rewrite all communications to follow the party line as part of an image and reputation management thing. Saves time. Campaigns can run leaner.

You probably want to use that the other direction:

Rewrite the party line into the vernacular and emphasizing the particular concerns of target voters.

consulting firms: we're so happy we can hardly count!

Given LLM are only trained on an english dataset, it will give some amusing results if politicians of non-english speaking countries try to do the same!

I don't think you can prepare, other than disbelieve most of what you see online. Almost no one has time to check, especially when modified content feeds into levied tropes against an unfavored candidate (however you define it).

A good chunk of it is pernicious. The legislation will need to catch up around AI-usage in elections, but the consequences of this are already stark.

I think by "prepare" they mean "start worrying about", which is not productive but does generate clicks and fear!

Worthless article. The conclusion.. "politicians might use AI to create more spam." The article even blandly suggests without any follow up that simply issuing more spam into the world is all that it takes to get you elected and that we're going to need a "congressional blue ribbon" commission to create white papers about it.

"AI" is just another information manipulation tool, like all the other information manipulation tools before it, it just looks scarier than others as it is branded as being closer to our own intelligence -- because survival is a game of intelligence against other animals trying to survive, something of greater intelligence (in all forms, or even just in the current environment you are in) is always an enemy of sorts.

It also scales in ways that no other information manipulation tool has ever scaled before. I'd say that AI is actually scary, even though I don't believe that it's truly intelligent.

And right now, just about the only things LLMs in particular scale really well at doing is “grassroots” propaganda and misinformation. Well, that and commercial spam, especially the really scammy sort.

The stuff’s a 0.1-2x multiplier for most things one might apply it to, but it’s a 100x multiplier or more if you don’t care about reputation (it’s all sock puppet accounts anyway) or accuracy. IOW it’s currently most-beneficial for evil use cases, by a long shot.

I am still pretty firmly of the belief that "difficulty/cost of generation" has not been the bottleneck on "disinformation" for quite some time, and also, quality of disinformation has not really been that important. Yes, AI might make "better" disinformation cheaper and easier to produce. But I don't think that those are the qualities that are holding the disinformation problem back from getting bigger, so I don't really expect AI to be able to make the problem much worse.

I'd be happy to see politicians use any form of intelligence in their campaigns.

I'd say they're highly intelligent... at manipulating us with wedge issues like immigration, abortion, and guns. A little effort at specializing in wading through feces saves a lot of hard work later by actually solving healthcare, debt, or myriad other things that affect 100x more people.

I’m not sure I’d be able to tell the difference. The emails are madlibs anyway.

It will go badly for at least one campaign; the news media will then trumpet that example of why AI will never work for political campaigns; meanwhile everyone else will keep on using it.

Unrelated but I was shocked to hear Joe Biden reciting the lyrics to the song What Would You Do by City High in a link that I was sent by a friend. I will be changing my vote accordingly.

Prepare for [everyone] to use AI in [everything].

It would be delightful to have an LLM fight back. It could stare at my incoming political texts and respond as though I am the most interested voter in the world while simultaneously hiding them from my eyeballs.

Good, I say. Right now political success is closely associated with how much money you raise. And how much money you raise is closely associated with what kind of special interest groups you promise handouts to.

AI will close this gap and allow smaller campaigns keep up with larger campaigns.

At first, it was an ear-piece with AI assist that some politicians used to make themselves more informed. Shortly after most politicians had to use this AI assist to keep up with the competition. And then nobody could tell who runs the state, as everyone blindly followed commands from the AI.

Good ol’ Manna

I think we're right on the precipice of a catastrophic shift in news/video/audio thanks to AI and deepfakes

Given how good deepfakes are now, how long before we have a video released to the major news outlets of the world of a world leader saying something very dangerous and scary. i.e. Putin could announce an all-out invasion of the US, Biden could announce aliens have landed and are here to exterminate us, etc. etc.

It could be something more benign like the NASDAQ has gone into a catastrophic dive never before seen, or that Tesla have been found to be breaking too many laws and are being shutdown effective immediately. (That might be a good way for the shorts to profit...)

I think it's literally a matter of months until it happens... and from that day forward we will never again be able to trust any video or audio recording. Ever.

> a world leader saying something very dangerous and scary. i.e. Putin could announce an all-out invasion of the US, Biden could announce aliens have landed and are here to exterminate us, etc. etc.

Who needs deep fakes when you have hot mics?

https://en.wikipedia.org/wiki/We_begin_bombing_in_five_minut...

I really hope all the Modern Major Generals at DoD, the ones who relentlessly plan their "strategies" for how WWIII will play out, incorporate this into their threat models. It's not as glamorous as the logistics, the horsepower, the death count, but it's probably one of the most likely scenarios that could lead us into war (and nearly has a few times due to false alarms). It should be a pact with all nuclear powers : "unless we see ICBMS flying towards us (with our own classified eyes in the sky)" we will not press the button.

I'm currently building a company around using LLMs to communicate and understand voters. (https://colloqium.com)

My view here is a bit more optimistic. LLM will allow campaigns where each individual essentially has a personal outreach plan. It's like a candidate can have a personal conversation with every voter. This will make things less divisive because you can talk to each voter about things that matter to them. There's no need to say the most inflaming thing just to get engagement.

One other bonus is that smaller campaigns can get their message out in a way that's authentic to them. If you're a parent running for school board because you care about your child's education you don't have an army of volunteers to do the outreach for you.

What you’ve described sounds like a one-way conversation, if not an insult: I get to be (essentially) propagandised to and misled into thinking I’m talking to a real person who will pass my concerns up the chain, when in-reality I’m not going to be heard.

If any campaign does to me what you describe then they will have lost my vote.

Politics is people.

We're straight forward with voters that it's an AI

And we actually do have alerts that go to the campaign team if a voter asks something that the agent can't answer.

I don't see how it's anymore propaganda than having them hire a bunch of people to text you.

My concern is the enshittification of news media. Outlets must often optimizes for sensationalism over substance to survive in a market which demands outrage, fear, drama, and brevity.

There is no story here. Yes, we've all benefited from AI advancements for years, including occupations like politicians. Yes, multipurpose tools can do multiple good things and multiple bad things. Yes, the world is full of spam & convincing misinformation. Yes, we must continue authenticating people online and continue using trustworthiness (not volume) of sources to determine fact.

None of this is new(sworthy). Pieces like this are lazy, devoid of value, and are actively harmful due to warping people's perspective with hyper over-emphasis of a tool's negative use cases. For shame, Bloomberg. Do better.

Who hasn't been dabbling at least? chatgpt is useful when it works lol.

You now understand why they have been applying a political bias to their AI.

i love content like this on youtube, i show it to my kids and they just can't believe it. They're still young enough to not question intuition when it comes to math haha. On a road trip once, my oldest ( at the time he was 12 ) tried to convince me for over an hour that he knew how to divide by zero...