This is fundamentally dishonest. If you're writing this, you should think "am I the bad guy?"
Yeah, maybe you'll get some supporters by saying you just want to stop those polygamists and tech bros and getting people to hate the Big Other when you're really trying to prevent AI from taking over the world. What's the worst that can happen if you're wrong? Well... stopping polygamists and tech bros and getting people to hate the Big Other, for one. History is full of people saying "we're going to tear down norms for benefit X", and benefit X turns out not to be real, but the norms that they just tore down are. It would sure suck if there wasn't any AI threat, but people did go to jail for generating that picture of Mickey Mouse, because they're just collateral damage.
Libertarians know that "regulations" is another way of saying "we are going to shoot you if you do this". Sometimes other people forget this.
I wrote my beliefs right on the tin of the post: I don't believe in existential risk threats. I simply would like to see the sorts of people who do believe in them be a little bit less unsuccessful. It breaks my heart to see people sincerely advocating for their beliefs in a way that renders them completely ineffective; every "Sorry, we're Canadian," that I find myself muttering to clipboard holders while walking with others makes me feel just a bit sad. If they were more entertaining, I'd find myself signing petitions more often, and then we'd both come out ahead. I like signing petitions. But they're not entertaining! They aren't good at making compelling pitches! This is entirely fixable with the correct approach; my aim is to equip them with the knowledge of the correct approach: Figuring out what people actually want, and then offering them a way to achieve their goals while aiding your own. The best persuasion is a positive-sum game.
I see parallels in presentation styles between the EAs I have spoken with and the humans outside of the Dolores & Market Whole Foods who have tried to get me to sign petitions: Fantastic folx, both groups, wonderful people, but the way I felt when some recovered felons tried to explain to me the value of improving the legal situation of attorneys who represent such-and-such cases, or something (I rapidly lost enthusiasm and quickly muttered "Ah, I forgot, I'm not registered to vote. Sorry."), is the same way most people feel whenever someone starts talking about existential risk. It's simply not parseable for most people; not at a glance, not with the level of investment most people have in a streetside conversation. People want to be helpful about things they care about! They want to sign petitions! Killing their enthusiasm by not knowing how to target them makes both of your lives worse, and if you genuinely believe in a compelling existential risk scenario, you're making everyone's lives worse by not doing your part to lower the chances of it happening.
Under the understanding of "Everyone will die if we don't," "We are going to shoot you if you do this," makes perfect sense. I don't hold that understanding, but I'd rather the people who do at least get an honest attempt at trying to solve their perceived problems. Them getting what they want faster, and as such, getting out of view once they've gotten what they were trying to get, would allow for people to fill the void who are trying to achieve things that I find fun, which is valuable to me. Further, the things you cite aren't norms outside of the Bay. That's why I noted they would only be persuasive elsewhere. There's nothing to destroy in them; they're common opinions, that normal people already hold.
What we foresee in the near future is the emergence of AIs who are far more capable than any humans. If the ASIs aren't aligned to a human-set target, they commit genocide. If the AIs are aligned, then we face the threats like a CEO taking over the world and the Intelligence Curse letting oligarchs lock in the entirety of power.
As far as I understand laypeople, they do NOT foresee AI taking over the world because they fail to understand the true extent of the AIs' capabilities (see, e.g., Freddie de Boer's attempt to drive home the point that the AIs are unlikely to acquire such capabilities, Casey Simpson's video full of factual errors or a video mentioned in my quick take). Therefore, laypeople are to be explained that the AIs WILL outsmart even the most capable humans or a majority of humans, thus rendering mankind or the laypeople unnecessary. Even the latter scenario would require laypeople to either stop the tech bros or be replaced.
I understand where you're coming from, but I think it's worth pointing out that every time you lie, your position becomes less coherent, in a way that the people who most matter can probably tell. Maybe the average person on the street can't, but Bernie Sanders (https://nitter.poast.org/SenSanders/status/2053867144564081151#m) probably has a rock-solid bullshit detector and his support is worth far more.
What do I mean by my position being less coherent? I want an international agreement to avoid building superintelligence, because I think it might kill us all. Very few of the more "mundane" concerns justify this - my belief in what actions we need to take follows directly from the severity of the consequences if we don't.
"AI might kill us all, therefore we should internationally coordinate to make it illegal to build superintelligent AI" at least makes sense if you accept the first premise. Try replacing "AI might kill us all" with "AI uses a lot of water" or "AI produces slop content" or "AI steals from artists" and my next question would be "Well, hold on, isn't the cure worse than the disease here?" It doesn't actually make sense, to say that we should ban AI development because images were stolen. If I'm worried about AI stealing from artists, we should compensate artists or enforce copyright. If I'm worried about AI water use, we should tax or cap water use in data centers. If I'm worried about AI job loss, we should find a way to tax AI revenue and support those who were automated away. None of this prevents human extinction. Why would it? You've told someone to work on a totally different problem, you can't be surprised if they go and do that.
Your advice around figuring out your audience and what they care about is good. I think that the correct lesson to learn from the arts of persuasion is how to present your case in a way that resonates with your audience. But every lie you tell makes your story worse. That's a cost too.
Notice that you said "if you accept the first premise." Normal people don't accept the "AI might kill us all" framing device (I'm one of them!) because it goes against all of society's priors, and Sanders isn't meaningfully in power. He's the opposition party's internal opposition party head. Coming off in a Stallmanescent fashion in order to sway the opinion of a politician without meaningful power whose views have never taken off in a country is a flawed strategy toward achieving goals. To achieve your goals, you should take people who will never see things your way and find a way to align their views with yours without changing either party's.
Making AI development less profitable and regulating it will slow it down. "Regulations slow progress" isn't particularly controversial. Slowing it down gives everyone more time to solve the hypothetical problems at hand. Stoking the public in such a way that they feel compelled to demand regulation of AI companies will work, given how reactionary politics in American history have historically worked out. People's views tend to follow society's, and pushing society further in the direction of disliking LLMs makes the situation more likely that the average person eventually agrees that further development should be banned.
Starting at the state level makes perfect sense; the federal level is pretty captured at the moment, and chasing LLM companies out of California would likely cause significant employee atrophy at said companies, which would aid in slowing down timelines.
I consider myself more plugged into the issue than most people, but even I'm not being swayed by x-risk arguments. Ikigai risk is something that far more people care about, far more intuitively, and it's worth taking advantage of that, if actually trying to reduce the chance of existential risks.
or, "A brief introduction to actually getting people to listen to your uninteresting-to-them beliefs, sign petitions, and generally not laugh you off the sidewalk outside of the Trader Joe's you so thoughtlessly set up camp next to."
or, "I thought you said you were consequentialists?"
Every day, I walk past petitioners. If you live in certain parts of San Francisco, California, it's likely a daily occurrence for you as well. I have enough samples at this point to know what works and what doesn't; what sorts of pitches tend to cause more head-shaking than others, and what actually gets people to sign their name and address on a stranger's form.
To put my cards on the table: I don't, at the present time, care very much about x-risk at all. I'm only aware of a few examples of mass extinction events, and I am unaware of the historical existence of ASI; I simply don't have the samples to justify a serious time investment into caring about x-risk, I'm not incentivized to work on it, and none of the people who believe in that sort of thing are good enough at writing fiction to sway me, even if they tried. I was inoculated by sheer hopefuel from significantly better authors at a young age.
However, it pains me to see street preachers get ignored, whether in real life or on social media, any more than they strictly have to be. An idea is entitled to the lighting it deserves.[1] Unfortunately, most idea-men would make terrible cinematographers. So my goal with this is to bridge that gap.
Stop trying to get people to like you.
Your goal is to decrease the odds of catastrophe. It is not important that you get your targets to like you, personally. It is not important that you share with them your own views around the odds of human extinction. It is not important at all, actually, that they even consider the idea of human extinction. Every human who spends enough time thinking to be useful for our purposes has developed a built-in aversion to thinking about doomsday predictions, and the odds of you turning that around for a stranger on the street or in a random twitter thread is extremely low. Not every format is the right format for conveying exactly what you're thinking; in a format like street preaching, you have to work backward from your goal to find what will aid in achieving it; you can't start with where you are and try to pull people back to you. Your goal, ultimately, is not the perfect alignment of your target. It is a "good enough" alignment. Once you accept that, we get to...
Figure out who your targets are for the day
Create a theory of mind for what your targets actually care about and can plausibly believe will happen, and then lean into it. This may be difficult, but it is worth doing. Nobody cares about the end of the world; the Mayan calendar in 2012 and y2k ruined that for all of us. There are only so many end-of-your-life parties you can go to before it gets boring; before the end is transparent before the beginning.
What do people care about? Well, right now, people really hate generative AI. Not for world-ending reasons, just for "this makes the world worse" reasons. Fears of economic impact, hatred for having their work stolen and productized by companies that choose to assist in genocide; depending on your audience, an infuriation with how these companies keep crashing their web servers. Capitalize on this.
Them developing an understanding of x-risk implies accepting more of your narrative than is practical for the format you have available to you (quick, street-side conversations with strangers), or that they'd ever be open to doing. The cognitive burden you'd be placing on them is simply too high, and the luck you'd need to do it in the span of a brief conversation is simply astronomical. After all, evil companies already sound a lot like you; the people working for them wear your faces like masks. The winning strategy is simple: Don't seem like you. Pick what they're afraid of and press where it hurts.
Figure out who you are for the day
We've established that you simply can't be you today. This presents a problem: Who are you? You don't want to lie too much, but as we've established, it's far more effective to push for regulation on grounds that make intuitive sense to the working stiff or American college student than existential risk scenarios that would only make sense with a background of reading sci-fi novels. The difference between an unhoused person and a petitioner is often not immediately obvious on first glance, and you'd rather not come off as the wrong sort, regardless of how charming the Scientologists handing out fliers in the park come across.[2]
Figure out where you are for the day
If you're targeting students, try outside a uni, or a bar that's popular among students (altered states are useful for altering subjects further; opinions become very malleable two beers in). If you're targeting software engineers, go to FiDi. The unemployed? Maybe try a beach, or a park during the daytime. The unemployed are underrated at achieving goals; they have all the time in the world. Under no circumstances should you set up camp on 4th Street, because lately there's been a bunch of people campaigning to save BART on 4th and Market, and that's much more important than existential risk, and brings me a greater amount of joy than your presence there would. Unless you're wanting to make them more effective, in which case: Go and tell them the advice I've given you! And then sign their petitions! And see if you can go campaign for them; knowing what you know now, you could get the required signatures in FiDi in a few hours.
Press into the wound
This is the fun part. You've identified what your ideal target's pain points are. Now you've got to build a sound way of making them seem worse than your target previously perceived them. Catastrophic labor effects; increased centralization of wealth into the hands of billionaires or "those damned tech workers;" intellectual property theft; the disgusting artifacts generated by these people's technologies that cause a visceral disgust reaction in so many charming people. The fact that these things are more or less entirely made up of negative externalities in terms of impact. Outside of San Francisco and Berkeley, you can even throw some sneakier slights in, with as bad of faith as you want: "And all to line the pockets of a bunch of polygamists!" Outside of San Francisco, that concern goes crazy. Press your fingers into their pain points until your forearm is covered in red. Spell out exactly what they're afraid of, and give them no way to bargain with themselves. It's not just coming for an abstract everyone: It's coming for you.
Apply a tourniquet
Okay, okay. You've had your fun. Your face is thoroughly splashed with blood and your shirt is a new reddish-brownish color. Now offer the person a life-raft. There's a solution to what ails you: Regulation! Regulation can tank the value of these companies; regulation can make them less useful at replacing you; regulation can punish these evil people; regulation can stop them from putting up data-centers; regulation can stop them from killing the cute, fluffy, innocent, harmless polar bears. If anyone builds it (and they've already built it), everyone's life gets mildly worse. The only solution the commonfolk have, now, is to ask politicians to do their jobs: To pressure them into actually wielding the regulatory hammers that they have at their disposal. Non-consensually using data isn't a right; the ability to do so can be taken away with law. Generating pictures of Mickey Mouse is, again, not a right. It can be taken away with law. The standard VC-funded, subsidized, "ask for forgiveness, not permission" at-a-loss strategy to eventually create a monopoly and profit isn't a natural law. With regulation, it can be taken away. Here's a petition that you can sign that could potentially solve one of those problems. Here's a flier to help you contact your state representative, and here's a paragraph to read that suggests to that representative a course of action. Here's a way to put out the fire.
Practice, Practice, Repeat
All getting good at speaking to strangers is is getting really familiar with the character you're playing. This will take a bit of struggling when you're not playing you. If you have severe social anxiety, maybe try in a social simulator, like VRChat, first. In present times, a headset will run you back around $200, and you can take advantage of the GPU you're probably already using for mischief anyway. A small investment for becoming a more effective agent of lowering existential risk. The expected value is almost infinite!
Or, of course, you could just go outside and start trying. This is probably better for your anxiety in the long-run, too. Rejection is something that you'll have to get comfortable with, though, and if your anxiety is bad enough, you can always try pulling these steps on-line while developing the skill. The on-line is where everything happens nowadays, anyway, even if it's harder to get people to take direct action in the on-line. You're probably already doom-looping over reddit.com or twitter.com or catatonically swiping on dating apps every day; put the time to use! The steps aren't that different from these steps; figure out who you're trying to get to, figure out who they want to hear from, make them feel worse than they already do, and then offer them a way out. Easy.
Under no circumstances should you talk to chatbots to practice this art. People can smell it on your breath when you talk to chatbots too much. Even people who love chatbots complain about hearing a chatbot's speech and patterns in the voices of others. It simply is not helpful in becoming competent at socializing. You need a situation that can challenge you. Chatbots cannot challenge you; not right now; not in this way. Go outside; touch some pavement; touch some people;[3] get good. It might just save the world.
If you understand what this is referencing from memory, good job! If you used an artificial conversationalist to guess, apply a weight of -1000 to the neural network in your head.
The author recently was handed a Scientology flier on Market Street. The author had to inform the wonderful people trying to offer their advice that they had already considered joining Scientology, but that they're simply much too neurotic to participate in the auditing process, as appealing a way out of the author's current situation as getting to work on a boat under an intense and strict organizational structure would be, and despite how strangely romantic the feelings were that the author felt after reading about Scientology's means of behavioral correction.
Not literally. Just their hearts and minds. There are better strategies for convincing someone you're touching to positively contribute toward lowering x-risk.