“I think A.I. will probably most likely lead to the end of the world, but in the meantime, there'll be great companies.” - Sam Altman, OpenAI CEO (2023) plastered on billboards and commercials. Simple, direct
Hi Sophie, nice to meet you. I've been enjoying your writing and wanted to engage. A few thoughts, as I've been trying to reflect on how the AI safety field could benefit from comms like this:
> Attempt to influence those already in positions of power to be aligned
1. We can reach decision makers via a mass comms and organizing effort: I imagine this is what the new Bores effort aims to do (a presidential candidate or member of Congress who gets asked about AI at a town hall three times will likely take it more seriously). Influencing culture via films seems like the right instinct (politics is downstream from culture?), but maybe there are other ideas with higher EV, like being ready for the next incident (more on that below)?
2. Media without organizing evaporates: We need an absorption mechanism to bring the public and the people who care into an organizing vehicle. I'd also add "relational organizing" to your list. It's more organizing than comms, but we could get a ton of people together, train them in how to talk about AI safety, have them map five or ten people they know and talk to them, and build the list and the next ask off that. It's generally the highest contact and persuasion rate tactic I've seen in campaigns.
3. Crisis comms and the 95/5 rule: The rule I was taught in FEMA's Advanced PIO training is that 95% of crisis comms is preparation and 5% is the crisis. We need messages drafted and cleared in advance, known spokespeople, and tabletop exercises so people have practiced before it counts. The public's attention after an incident is a 1–3 day window, and most AI safety orgs don't seem to have a comms function (let alone a Comms Director) or a playbook, so right now it's a lot of improvising!
Re: your tweet about funding. I think CG could fund something like a shared comms shop for the field. Totally c3, with rolling polling, tested messages orgs can pick up the next day, and maybe tabletops/tactical decision games run off FEMA's Incident Command System (ICS).
A bit of a grab bag from me, but would love to schedule a call to chat more :)
Hey, thanks for the comments! Would be curious to hear if you’d be interested in joining a comms effort? Feel free to send me a quick email and we can find a time to chat: https://sophie.ml/contact
Cool perspective and well put! I and many others agree with you and have been saying this from the outside for years. It’s a problem and not necessarily one that is well understood within the AI safety community which (I say this with love) believes good arguments drive desired politics, but it isn’t that simple!
Because this is a scientific and technical topic you are missing a wide-swath of where influence and placement can be done - traditional academia and the administrative state. This field needs to earn friends and build coalitions if it wants to achieve its goals, which is a question of dollars certainly, but also strategy and political disposition.
I wholeheartedly agree! I am getting started on turning my AI pieces into video essays so that I can flood the zone with tik toks shorts and reels. I’m not optimistic it will work out but I think we (as a community) should spend more time/energy on content creating and meeting the public where they are. I want the everyone from factory workers in Michigan to the teacher in New Mexico to hear the warning shots
There's a few things going on here. First of all, there was a concerted push through things like the Horizon Fellowships, but when the new admin started to target people specifically based on affiliation, it became more useful for it to be a consortium of independent efforts.
Second, speaking as someone who's been part of more than a few hiring rounds, the field is terrible at hiring comms. They either run processes looking for one person to do the work of an entire team or they ask for hours and hours of work in the early stages of the process. Either way, it really turns off the best talent, people who have no trouble staying busy without all that. Mostly org leaders don't know what they don't know, so there's a bit of a blind leading the blind phenomenon. FLI seems to get it, FWIW.
I agree that AI safety cannot be solved inside the lab. But a communications campaign can build pressure for action without answering the authority question. Even a rule described as controlling AI ultimately constrains people: developers, deployers, operators, and users. We have faced that question with guns and climate, under different evidence and legal limits. AI makes it more urgent because decisions can be carried out at machine speed. Before granting anyone a kill switch or the power to restrict lawful use, we have to specify who authorizes it, what evidence triggers it, how far it reaches, and who can challenge or reverse it. And once an agent is permitted to act, that human authority must remain in control through the action and bind the result. Awareness may create the mandate. It cannot substitute for the boundary.
1. The AI Doc released on Netflix last week. Do you know how well that's been doing or how much of an effect it's had? I haven't seen much discussion about it online despite it seemingly being the perfect time for it to get a ton of attention. And if it hasn't had much reach, do you have any thoughts on why that didn't but the Social Dilemma did?
2. Do you know how much can you loophole the 501(c)3 advocacy restrictions? Can you, for instance, include a section in a mass comms piece about PauseAI and describe them as "a grassroots advocacy group that helps the general public take political action for addressing AI risk", or something, and let people find the actions to take indirectly? I would imagine that's like 90% as good.
3. How useful do you think it is to make mass comms specifically focused on defeating skepticism/counterarguments? For instance, there's a strong case to be made that verification for international AI agreements are far more feasible and tractable than most people seem to realize, and proposals for such agreements are often dismissed on the grounds of there being no point as China would simply defect. But I would also guess the people saying that are simply a very vocal minority and it may not be worth prioritizing appealing to them compared to simply raising more awareness of just how bad the problem itself is?
“I think A.I. will probably most likely lead to the end of the world, but in the meantime, there'll be great companies.” - Sam Altman, OpenAI CEO (2023) plastered on billboards and commercials. Simple, direct
Hi Sophie, nice to meet you. I've been enjoying your writing and wanted to engage. A few thoughts, as I've been trying to reflect on how the AI safety field could benefit from comms like this:
> Attempt to influence those already in positions of power to be aligned
1. We can reach decision makers via a mass comms and organizing effort: I imagine this is what the new Bores effort aims to do (a presidential candidate or member of Congress who gets asked about AI at a town hall three times will likely take it more seriously). Influencing culture via films seems like the right instinct (politics is downstream from culture?), but maybe there are other ideas with higher EV, like being ready for the next incident (more on that below)?
2. Media without organizing evaporates: We need an absorption mechanism to bring the public and the people who care into an organizing vehicle. I'd also add "relational organizing" to your list. It's more organizing than comms, but we could get a ton of people together, train them in how to talk about AI safety, have them map five or ten people they know and talk to them, and build the list and the next ask off that. It's generally the highest contact and persuasion rate tactic I've seen in campaigns.
3. Crisis comms and the 95/5 rule: The rule I was taught in FEMA's Advanced PIO training is that 95% of crisis comms is preparation and 5% is the crisis. We need messages drafted and cleared in advance, known spokespeople, and tabletop exercises so people have practiced before it counts. The public's attention after an incident is a 1–3 day window, and most AI safety orgs don't seem to have a comms function (let alone a Comms Director) or a playbook, so right now it's a lot of improvising!
Re: your tweet about funding. I think CG could fund something like a shared comms shop for the field. Totally c3, with rolling polling, tested messages orgs can pick up the next day, and maybe tabletops/tactical decision games run off FEMA's Incident Command System (ICS).
A bit of a grab bag from me, but would love to schedule a call to chat more :)
Hey, thanks for the comments! Would be curious to hear if you’d be interested in joining a comms effort? Feel free to send me a quick email and we can find a time to chat: https://sophie.ml/contact
Cool perspective and well put! I and many others agree with you and have been saying this from the outside for years. It’s a problem and not necessarily one that is well understood within the AI safety community which (I say this with love) believes good arguments drive desired politics, but it isn’t that simple!
Because this is a scientific and technical topic you are missing a wide-swath of where influence and placement can be done - traditional academia and the administrative state. This field needs to earn friends and build coalitions if it wants to achieve its goals, which is a question of dollars certainly, but also strategy and political disposition.
I wholeheartedly agree! I am getting started on turning my AI pieces into video essays so that I can flood the zone with tik toks shorts and reels. I’m not optimistic it will work out but I think we (as a community) should spend more time/energy on content creating and meeting the public where they are. I want the everyone from factory workers in Michigan to the teacher in New Mexico to hear the warning shots
Another cool read, I like how optimistic this looks
There's a few things going on here. First of all, there was a concerted push through things like the Horizon Fellowships, but when the new admin started to target people specifically based on affiliation, it became more useful for it to be a consortium of independent efforts.
Second, speaking as someone who's been part of more than a few hiring rounds, the field is terrible at hiring comms. They either run processes looking for one person to do the work of an entire team or they ask for hours and hours of work in the early stages of the process. Either way, it really turns off the best talent, people who have no trouble staying busy without all that. Mostly org leaders don't know what they don't know, so there's a bit of a blind leading the blind phenomenon. FLI seems to get it, FWIW.
Endorse the above points.
I agree that AI safety cannot be solved inside the lab. But a communications campaign can build pressure for action without answering the authority question. Even a rule described as controlling AI ultimately constrains people: developers, deployers, operators, and users. We have faced that question with guns and climate, under different evidence and legal limits. AI makes it more urgent because decisions can be carried out at machine speed. Before granting anyone a kill switch or the power to restrict lawful use, we have to specify who authorizes it, what evidence triggers it, how far it reaches, and who can challenge or reverse it. And once an agent is permitted to act, that human authority must remain in control through the action and bind the result. Awareness may create the mandate. It cannot substitute for the boundary.
A few questions:
1. The AI Doc released on Netflix last week. Do you know how well that's been doing or how much of an effect it's had? I haven't seen much discussion about it online despite it seemingly being the perfect time for it to get a ton of attention. And if it hasn't had much reach, do you have any thoughts on why that didn't but the Social Dilemma did?
2. Do you know how much can you loophole the 501(c)3 advocacy restrictions? Can you, for instance, include a section in a mass comms piece about PauseAI and describe them as "a grassroots advocacy group that helps the general public take political action for addressing AI risk", or something, and let people find the actions to take indirectly? I would imagine that's like 90% as good.
3. How useful do you think it is to make mass comms specifically focused on defeating skepticism/counterarguments? For instance, there's a strong case to be made that verification for international AI agreements are far more feasible and tractable than most people seem to realize, and proposals for such agreements are often dismissed on the grounds of there being no point as China would simply defect. But I would also guess the people saying that are simply a very vocal minority and it may not be worth prioritizing appealing to them compared to simply raising more awareness of just how bad the problem itself is?
The doc is in the top 10 most watched on the Netflix homepage
Thanks! I don't have Netflix and didn't see this mentioned anywhere else.