The comment by Yuval Noah Harari seemed insightful to me. If you argue against a bot about a political matter, not knowing it’s a bot - you always lose on a long enough timeline. ie you can never pursuade the bot but it can wear you down / eventually find an argument that works.
The only winning move is not to play. So I could see this having a chilling effect on all discourse
Aside from the whole skynet thing the above is what spooks me the most
The chilling effect may actually be a good thing, given that discourse these days is overheated.
There's a weird magic trick that social media companies have played on people to convince them that the text and images consumed on their websites are socially/culturally/politically relevant. Once it becomes clear how easy it is to fake that text people will come to understand how cheap and irrelevant "opinion" has become and this magic trick will become weakened.
We want an an appropriate degree of emotional engagement with discourse, which is 1-1 with how much discourse is happening. People being too angry is caused by people discoursing too much and vice-versa. There are opposite problems associated w/ too little discourse, but we don't suffer from those.
Things are hyper-polarized right now and there is no magic political synthesis that is right over the horizon if only we could just keep discoursing a little bit more. This is like a heroin addict thinking they'll cease being addicted after that last fix. The solution is to cool things down.
Personally, that's not been my experience at all. I often find that when I have two friends with highly disparate and deeply held beliefs, the intense emotions they associate to these ideas are due to them not actually engaging each other but, instead, taking their emotional cues from their respective ideological silos (where no real discourse is occurring), and then proceeding to talk past each other.
Learning how to actually talk to one another in good faith with humility and charity is a skill that comes with practice. Deciding to engage each other less can worsen the situation by allowing one camp's preconceived notions about another camp to go unchallenged by reality. This allows each camp to tell an increasingly vilifying story about the other, which increases, rather than decreases, the emotional charge between the two.
engaging w/ someone is different from discoursing with them. "engagement" is what social media companies say they provide - but really they just offer "discourse".
I would predict that the chilling effect will be lesser for "unreasonable" voices such as trolls and extremists, and will be greater for the moderate voices.
This is not a good thing, as the past 10-15 years of social media has shown.
The entire thesis of TFA is that human psychological traits on societal scales are not prepared to handle an arena of discourse where that's true, that people will either be duped or completely check out of discourse per se, not limited to online social media.
No, it's a feature of the human mind that truth is considered objective independent of the lens used to acquire it. "Post-truth" sources in some arenas will be conflated with "real-truth" sources in others, leading to a blanket demotion of the perceived value and quality of truth. The whole argument from TFA is about human psychology, not about specific offerings of "marketplaces" where truth may be more or less maligned.
dunno what "TFA" refers to here, but it seems like we're heading into a argument regarding epistemology, which is not a discussion that HN handles well.
IMO "truth" is distraction here because what is at stake here are people's values, not their understanding of math and physics. When people worry about "post-truth", they're worried about liberal values no longer being the unquestioned default. It is absolutely a marketplace, and if people switching marketplaces en-masse makes it harder to launch rockets and develop vaccines, then it probably means those activities are making people net unhappy. People are a lot smarter than we give them credit for, even the dumb ones.
TFA = the f'ing article, something quite commonly understood on HN for time immemorial. Its snark is borne of the community's distaste of the kind of people who dive into comment sections without engaging with the very subject of and reason the comment thread exists in the first place. The fact you don't recognize this acronym calls into question your authority on what HN can or can't handle. But anyway, that's beside the point.
We are (well, I am, and TFA is) not talking about epistemology so much as the public's inability to engage with epistemological problems on systemic scales. Instead the limits of human psychology control how we as a society respond to these issues. Your argument is a distraction that remains uncontextualized within the conversation it finds itself in.
People are not dumb animals but you won't be able to engage anyone toward a solution on the basis of an argument about how they just need to understand more about epistemology. That's the kind of thing that people can only internalize via empirical means.
At this point it feels like you're being deliberately obtuse. I've been quite clear about the primacy of human psychological limits as the main aspect of the argument and you simply refuse to engage with this point. You haven't been very good about adding to the conversation, only diverting it.
> the public's inability to engage with epistemological problems on systemic scales
The public's inability? What about everyone's inability. No one deals well with epistemological problems on a systematic scale, not even the technologists who delude themselves into thinking that they're driving anything.
I am exactly talking about psychological limits. The difference is that I don't think the psychological limits of the creators are any different from those of the users. If anything, I think the creators are more psychologically limited than the users. This is because the creators need to explain to themselves why they are creating the thing - everyone else just puts up with it. When you say ppl will either be duped or completely check out of discourse, don't forget about yourself.
right - like how the stock market behaves nowadays w/ quant traders - which then motivated people to construct darkpools of liquidity where the real trading happens.
There are plenty of pushy humans who are prepared to talk you into submission. Check out your local Scientology or Lyndon LaRouche movement office. Would you talk interminably with a pushy human? How is the AI any different?
The differences that jump out at me are in cost, scale and accountability.
It's work recruiting people to your cause. It's expensive, and takes time and attention and resources. People have consciences that might flare up, they have loose lips, they need training and guidance and oversight.
Letting a bot loose for your cause costs pennies. It can be updated with a few clicks. You can reverse course or fine tune with a few sentences; you can even tailor it to the vulnerabilities of your victims with just a few data points.
The brainwashing inflicted on people by advertising and the like has had a tremendous cost to society, and the planet; absolutely incalculable. It would be prudent to be alert to the danger of all that being exacerbated 100- or 1,000-fold.
it's not just cheaper, its eminently more reactive.
your human recruiter is going to have their own biases, could be tired, may not know how to deal with personality X or Y, they may not learn.
the bots can track how they do by upvote, react quickly, won't get tired or annoyed, and can use existing data to filter and target those who are extremely vulnerable.
hell, with real-time analysis by data mining companies (cough cough FB, reddit) they could even figure out if/when a normally stable-ish person is in crisis or has a bad week, and hit them at their most susceptible. 99% of the year they might be able to see through the BS, but this week things are hard and what they're seeing in their feed just feels right, for once...
> If you argue against a bot about a political matter, not knowing it’s a bot - you always lose on a long enough timeline. i.e. you can never persuade the bot, but it can wear you down / eventually find an argument that works.
That's a useful insight. The one on one wearing-down process came from theology. It's sometimes called Jesuitry. Members of the Society of Jesus were trained to do that sort of thing, and they got so good at it that the whole organization was suppressed in the 1700s.[1] There are still Jesuit-run schools in the US, and they do tend in that direction. "Never argue with a Jesuit", said Richard Nixon.
Since mass media, large scale use of one on one convincing has gone out of fashion. It's not cost-effective. With large language models, it's back.
You may not ever be able to persuade a bot, but you can identify if it is capable of repudiating your points or not. You can ascertain if it is arguing in good faith and chose to end an argument if it isn’t.
When you say “find an argument that works” isn’t this just saying that they’ve said something you find persuasive?
What’s wrong with this? How is this different to any other argument in good faith?
Because we have to assume not everyone can ascertain what are bad faith or dishonest arguments, or even good faith but illogical arguments. People are fallable and can be persuaded by bad arguments. So the point is, given enough time, the person will make a mistake and fall for a bad argument.
As someone struggling to destroy some of my own beliefs, due to their potential to waste a lot of my time. I have failed to do so.
The perceptions of other disagreeing people ended up just being one more way to validate them because you end up seeing the same pattern of mistakes over and over again.
It is a straw man to pretend that political discussion looks like an Intelligence Squared debate when almost all the time it looks like a food fight about nonsense.
The idea the world will be worst off with less flame wars is simply wrong IMO.
I would already rate the discourse I have had with chatGPT4 as the best of my life.
I win if I learn something. It strikes me as highly perverse to view that the only way to win is to not play the game because you can't brow beat the bot into submission.
THAT is though what we mean by "political discussion". A bullshit pie throwing contest until one side quits. A twitter flame war. Yes, hopefully AI completely destroys that.
I've noticed some humans are willing to argue without giving up too. Sometimes online arguments follow a pattern where each response is longer than the last until one person decides it's not worth their time to continue the argument. In my head I've called this "argument by attrition".
From a Bayesian perspective, the only way to play and not lose is to not update your priors, at all, forever. (I suppose that "not playing" winds up much the same - you don't have any input on which to falsely update your priors.)
But people who absolutely will not update their priors is not likely to work out well in the long run...
Bots or no bots, that'w why arguing on the Internet is pointless.
I could argue with you, "Havoc", for a while, and perhaps convince you that there are 2/3/7 genders, or whatever, and then as sure as the sun rises, along comes another user. Or infinite set of.
Most social media is "man yells at cloud", quite literally.
A lot of times you don't argue with someone to prove to them you are right, but to prove to bystanders that there are different ways of looking at this. Refusing to engage will not make the world a better place if the only ones talking are idiots.
I talk about political matters for ideas, education, and stimulation (mine and the other person's), rather than to beat them into taking on my position. I'd certainly walk away from a conversation long before I'm worn down into taking on their views.
I think this interchange of thought and sharing of ideas is the true essence of discourse. That other thing is more of a verbal tribal battle. I can see issues with bots swinging the political landscape in ways that cause harm, but I think there is good reason to think discourse might flourish. Fundamentally, if the competitive arguers are discouraged, whilst the curious and conversational are encouraged, or at least not hurt, then discourse itself wins.
I don't think political swings are likely to happen from improving curious thoughtful discourse, these things are dwarfed by blunt evocative approaches that reach broad segments of voting society
We might find discourse flourishes corners of discussion on the internet but the wider internet is much more at risk to these approaches
I think you're mistaken at least with current AI. It's very easy to condition them into a new frame or mind if you speak to them in a certain way. They're also very prone to "trope" - where when your sprinkle a few hints of something that's common they'll fall right into line with behavior associated to it.
We forget that AI becomes more human as it gets more intelligent. These are not computers programmed with a hard limitation anymore, they're actually more prone to being manipulated than other humans are.
This is dumb idea from someone who does not use online forums. When someone keeps making stupid points against me online, I get only more dug in on my views. If anything, you would want to make a bot to say "Not X" so the pro-X side could feel better about how dumb the opposition to X is.
I've yet to see an AI bot that is both realistic and intelligent enough to keep me engaged (GPT4), but also can't be convinced that 1 + 1 = 3 (also GPT4)
The only winning move is not to play. So I could see this having a chilling effect on all discourse
Aside from the whole skynet thing the above is what spooks me the most