Top

Talk to your family and friends

Everyone can drive social change. Social movements that succeed do so because the people who power them understand that relationships can start decisive chain reactions that tangibly shift public opinion, seemingly out of nowhere.

People becoming aware of the problem, taking it seriously, and talking about it, is what will motivate our representatives to take bold and focused political action. In turn, this will:

  1. pave the way for a pause on advanced AI, and
  2. make it easier and easier for other politicians to take the issue seriously and find their courage, which is how political momentum will snowball.

If you can guide just one person a month to understand the problem, and this person and the people they in turn mobilize all do the same, by having 12 powerful conversations over a year, you’ll have caused over 4,000 citizens to become advocates for a pause on dangerous, uncontrollable and unaccountable AI. And each of them can start their own chain reactions.

A fractal tree: a single trunk branches over and over into a canopy so dense it reads as a solid cloud. A hand-drawn arrow labeled "you are here" points at the base of the trunk.
The trunk is your first conversation. Every fork above it is somebody deciding this matters enough to tell someone else – and that is what compounding looks like once it gets going. Fractal tree illustration from Plus Magazine, University of Cambridge.

This is what real people power looks like. It’s easy to forget we have that power, and it may feel unfamiliar to use at first. The handful of labs building extremely powerful AI without accountability or safety boundaries have a lot of power, but the world is still made of people and the power they have in using their voice and making choices (as voters, community members, consumers etc) is still larger for now. It’s up to us to use our voice to speak up for a pause, or lose it to frontier AI labs.

What good is one conversation?

All of us within the PauseAI movement are doing what we can to generate the political will that will lead to political action, i.e. getting to a pause treaty.

⇒ Our efforts have a better chance of succeeding with every person who learns about the problem and its solution and realizes they can do something about it.

⇒⇒ Research shows that people are more likely to take an issue seriously and become civically engaged when they hear about it from persons in their lives – persons with whom they share context, language, trust, kinship.

⇒⇒⇒ This means that every one of us can be a powerful communicator, helping the ask for a pause and the necessity for it to reach more people, deeper into communities all over the world. To some people, you are the best possible messenger, even if they may need to hear the same message a few more times to really accept it.

There’s a useful distinction hiding in that. Information spreads easily – someone hears “PauseAI” on a podcast and now they’ve heard of it. But changing your behavior – speaking up about something unsettling, joining something, being the one who brings it up – carries a social cost, and people rarely do it after one exposure. They do it once several people they already trust have done it first. That’s why a handful of conversations inside one circle of yours can matter more than a message broadcast to thousands: you aren’t only informing people, you’re making it normal. This is the sociologist Damon Centola’s distinction between “simple” and “complex” contagion, set out in his book Change: How to Make Big Things Happen.

This is how we can grow into a political force to be reckoned with.

Tell people in your life about severe risks from advanced AI, why you support a pause, and why every person can make a difference by using their voice as a constituent and as a member of the communities they’re a part of.

This could be a 2-minute chat when your neighbor asks you how you’ve been, where you tell them AI risk is on your mind these days and ask them how they feel about it, inviting them to take a look at pauseai.info  if they’re curious to find out more, or it could be an hour-long conversation over a sunny walk with your contrarian but ultimately caring friend who likes to be skeptical of everything by default but accepts reasonable arguments if you’re patient with their temperament.

Catastrophic/extinction risk from advanced AI threatens everyone, and everyone deserves to know about it so they can take action to move the needle toward a pause on advanced AI – even if, for them, that only means signing a petition now and again.

Not all your conversations will help people understand the problem and feel the motivation to act, or the sense that they can do something about it. That’s ok. It gets easier with practice, and even people who may be shy or reluctant to change their minds now will likely come around once they see that many people in their network have changed their minds. Just be kind to your interlocutors and to yourself.

Ready to spread the word?

Here are 3 simple steps. The four pillars below go into how to do each of them well.

1. Map your social network

As a starting point, list everyone you know (you can set a 5–10-min timer). Then ask yourself: who are the people around me who are most likely to be receptive to a conversation about this? Who do I have the strongest bonds with? Who are the people in my network who are influential, and who may best help spread awareness?

These questions can have a lot of different answers, because there are many ways to be receptive, many ways to be connected and to nurture a relationship, and many ways to have influence. Prioritize having conversations with people who come up in response to two or more of these questions, especially people whom you expect to be receptive and people with whom you have a strong bond, because that brings a ground of trust and openness to the conversation.

Look out for bridges, too – people who connect two different circles of yours, like a colleague who’s also in your running club, or a cousin on the parish council. Reaching one of them can carry the conversation into a whole group you aren’t otherwise part of.

Keep this list to yourself, on your own paper or your own device. There’s no need to upload it anywhere.

2. Start your first conversation

Introduce the topic of AI and share a headline mention of your concerns.

If you’re not sure how to open, being plain about why you’re raising it tends to work better than easing in sideways:

“Hey [name], how are you? Listen, something’s been on my mind a lot lately: where AI is heading, and how fast. I’ve been reading about it and honestly it worries me. Have you been following any of it? I’d love to hear what you make of it, if you’ve got a bit of time.”

That works because it’s honest about why you’re raising it, it leads with your own worry rather than a pitch, and it asks for permission and hands them the floor instead of opening with a monologue. PauseAI can come up later, once you know where they stand.

From there, Pillar 2 below covers listening to your interlocutor and checking you’ve understood where they stand, and Pillar 3 covers bridging their concerns to what’s at stake with advanced AI.

3. Suggest an action

Suggest actions your interlocutor can take if they seem responsive – see the ladder in Pillar 4 below. Don’t ask them to commit to doing great things; show them a couple of simple things they can do that will allow them to step into their power as an informed citizen and community member. If you feel your interlocutor is still taking it all in, don’t press them to do things; let them process things. Let them ask follow-up questions, let their mind and emotions adjust. You can point them to pauseai.info  or another resource of your choice if they want to find out more.

If the person you’ve been talking with seems unresponsive or hesitant, of course, that’s fine. This is their journey, and many people take time to adopt transformative new ideas and ways of being (such as seeing themselves as a citizen who has plenty of power to take peaceful and effective action to protect what they care about alongside a like-minded community).

The four pillars of a great conversation

1. The relationship is the foundation

Your shared context and mutual trust is what allows the other person to situate themselves in relation to the information you’re sharing. It takes the support of relationships and trusted conversation partners for people to deeply change their mind or confidently, wholeheartedly adopt a different behavior.

Sometimes it can also be good to connect by simply acknowledging that this isn’t an easy conversation for you to have, and why that is. That can instantly help them relate to you, and to how important this problem is in your eyes, and make them more open to having the conversation in a way that feels good for both of you.

2. Listen before you speak

If you don’t make space to appreciate your interlocutor’s way of making sense (cognition, beliefs) and of making meaning (emotions, identity) of the world, your efforts to communicate will be disconnected. Just like a plane pilot, you want to understand the “terrain” of the other person’s perspective to have a chance for your message to land well for all involved!

Remember: people both have intuitive and rational beliefs and assumptions and complicated and delicate emotions (and you’re people, too, so don’t forget yourself in that picture!). You need to stay present with both aspects, cognitive and emotional, in order to be able to grapple with this topic well together.

By understanding where your respective models of the world differ, you can give the other person pieces of the puzzle that are useful to them specifically. That’s a gift for them and for your relationship.

In practice, aim for roughly 80% listening, 20% talking. Open questions do most of the work – “what’s worrying you most at the moment?” – and so does reflecting back what you heard: “so if I’ve understood you, the part that bothers you is…”, which gives them room to correct you. Leave silences, too; people often say the most important thing just after one.

Watch how it’s landing. If the topic is clearly weighing on them, or they’re going through something hard, let it go – you’re opening a conversation, not adding a burden. And you aren’t trying to win an argument: understanding each other is the win.

Appreciate their openness if they can be curious – it’s hard for humans to change their minds, especially around scary topics.

3. Start from the other person's reality

What are your interlocutor’s concerns? Perhaps they’re concerned about climate change – in which case, they already understand that humans haven’t got it all figured out and that we’re vulnerable to things more powerful than ourselves. Perhaps they’re concerned about the erosion of democracy, or about powerful tech being used in warfare (whether AI-based or not), or about geopolitical instability, or about cybersecurity. Most of the things people are concerned about have some property that is similar to AI harms and risks, or are likely to be worsened by uncontrollably powerful AI.

Besides, there are many ways that AI risks and harms may already show up in their lives, and they are all part of the bigger picture: AI is being introduced everywhere, without accountability or democratic consent, and we don’t know how to keep it safe and serving human interests. It’s introducing massive changes to most aspects of society, and this is only the beginning of what we should expect as these systems become more and more powerful and less and less controllable. If your colleague, friend or aunt are concerned about job losses or AI psychosis, that’s not mutually exclusive from the catastrophe/extinction risks we’re beginning to face, and those are genuine human problems that deserve consideration and compassion. They’re facets of the same picture, and they all warrant a global pause while we work out how to protect human interests and what AIs we might be able to safely build.

Two worked examples

If you’re talking to your property developer/builder uncle, explaining that AIs don’t have to “want” to destroy humanity to do so by using the metaphor of “when humans build houses, they kill ants in the ground below not because they want to, but because their priority is building houses; if AI becomes able to do more than humans, that’s the situation we end up in by default” could land quite intuitively for them.

If you’re talking to a friend who cares about the mental health consequences of using chatbots, you can help them notice that this is an early form of misalignment, whose full form will become catastrophic as these systems become more powerful and more embedded in society’s systems; generally, showing people bridges between current harms they already see and emerging catastrophic risks can be really useful.

4. Start with small asks, then encourage bigger steps

It can be unfamiliar or uncomfortable to take one’s first steps in any kind of advocacy. Most people don’t think they have it in them, and don’t realize their power as citizens, community members, professionals etc. unless an issue becomes so important to them that they’re forced to look for ways to protect what they care about. Like most things, it gets easier, so small steps are a great start!

Whatever you suggest, three quick checks make it far likelier to actually happen: is it one clear action (one verb, one link), can they do it in under two minutes, and can they do it right now, while you’re still together? An ask that fails those three tends to dissolve into good intentions. It also helps to have a lighter fallback ready, in case the first suggestion lands as too much.

Start here

You could encourage them to:

If that felt good

Once they’ve taken a step, if that felt good to them and they feel motivated to do more, you could:

  • invite them to a relevant event, online or in person
  • recommend a deeper dive into some more detailed content (again, see our resources page for suggestions)
  • encourage them to join a group (either a local group or a remote volunteer group) and get involved in tasks or projects

Shifting gears

If they want to get involved more significantly, they could then, for instance:

  • run a local group or become a volunteer team lead
  • organize awareness-raising events
  • join our next PauseCon  organizer training conference

Preparing for the conversation

Tips before you start
  • Help your interlocutor find things they can do about the problem; invite them to your PauseAI online community channel of choice, ask them to share the video you just showed them with someone else, show them guided actions they can take on pauseai.info/join. Be sure to end on an empowering, hopeful note.
  • Tune into why you care about this, and why working toward a pause is a convincing approach to you. Let the other person feel and understand what drives you.
  • If you can, prepare a bit. You can role-play with someone in our community, try to predict where the conversation might get stuck or feel scary for you or them, brush up on common objections at pauseai.info/counterarguments, or ask yourself what kind of explanation would be easiest for this person to understand. (For two worked examples of that last one, see Pillar 3.)
  • Consider whether you have time to have a full conversation, rather than briefly showing e.g. a video from our YouTube channel  , which is also good!

You can connect with fellow PauseAI community members over our community calls, our Discord server, and our chapters’ own online and in-person spaces. Practicing having these conversations together, talking about the successful and difficult conversations we’ve had, and exchanging models of catastrophic AI risk and of why a pause is needed are a great way to gain confidence! Head over to our events calendar  to find events to participate in, or join our community platforms.

If you find such conversations daunting, you can focus on pointing your interlocutor at a few key facts about the problem that you think will make sense to them, or recommend resources if you don’t feel you can explain things into too much detail yourself, e.g. pauseai.info/learn.