This is a special post for quick takes by catpillow. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Does anyone know what the theory of change of AI Safety comms is? The main people working on this, Frame Fellowship and plzdontkillus, seem to have different strategies for comms, which seem to indicate people have different theories of change for what AI safety comms is even supposed to do.
From my perspective, the main difference seems to be:
From that, I'm guessing that the theory of change that frame fellowship has is something more like "appeal to people in positions of power/with some expertise -> convince these people to join the AI safety movement/make decisions sympathetic to AI safety", whereas plzdontkillus seems to be more like "appeal to the general public -> make these people more sympathetic to voting for legislators who promise AI safety bills". (I'm oversimplying the proposed ToC here)
But from that framing, it seems like frame fellowship should be tunneling in way harder on LinkedIn rather than Instagram/TikTok, and plzdontkillus should be tunneling in way harder on boomers, who turn out in local elections far more than younger audiences. Am I missing something here? Is there some obvious reason why these two comms fellowships aren't doing that?
My personal theory of change for AI safety comms: If the salience of AI safety as an urgent issue increases among the public and among policymakers, it is much more likely there will be an AI pause and/or meaningful international treaties soon, which I think of as the #1 priority for AI safety right now.
I think what you are missing is that you will have a much easier time convincing the average 25 year old to really value AI safety than the average 75 year old, and that linkedin has very limited potential for reach. I don't know enough to really judge either fellowship project specifically, but I don't have a better idea for increasing AI safety's salience with the public than "get more people to throw AI safety content at the wall and see what sticks".
Your personal ToC seems pretty similar to what I've seen the two comms fellowship say so far, but I think it misses a lot of steps between "people care more about AI safety -> meaningful actions are taken". I think this matters because a lot of movements, for example like the Pro-Palestine movement, have roughly the same ToC, with major success in getting the majority of people around the world to be passionate about their cause, but limited impact in actually achieving their ends (e.g. stopping the war). I think the subpar outcomes came from major confusion about what their actual goals and bottlenecks are (e.g. convince Israelis to vote for more anti-war candidates, make Hamas more amenable to negotiations), and targeted actions in achieving their specific goals. Right now, I'm seeing AIS communicators make the same mistake.
And as for your points about LinkedIn & Boomers
I'm very confused by this take. Comms is a broad field - within that field you'll find that different organizations focus on different activities. Of course their ToC will be different as well.
You might as well point at BlueDot and 80,000 hours then wonder what the theory of change of AI Safety fieldbuilding is, if each org does something different.