They’re not really bad-faith if they’re just making the same point over and over again that you’ve failed to respond to. Like, “Anthropic is attempting regulatory capture.” That’s a good point. I can’t really get engagement with it on LW, maybe because it detracts from our push for regulation. But that promotes it to a regular faith objection, just one that’s not responded to seriously.
Or the fact that labs are constantly flooding the zone with announcements that simultaneously advertise capabilities, threaten to kill you and all your loved ones and sort of flaunt they get away with illegal behavior. Saying they’re hyping it is a regular objection to doing what they want. It’s bad faith to not engage with it.
And the bad-faith objectors might look like the more reasonable people and attract more believers. If you haven’t even admit the obvious fact that they are corporations attempting a power grab, why take you seriously on doom arguments?
And I don’t think the basic AI safety explainer audience is people who already know AI safety basics, right?
They’re not really bad-faith if they’re just making the same point over and over again that you’ve failed to respond to.
I agree if it’s a dialogue you should respond to it! [1] But imo you shouldn’t start with a presumption that this is the target audience since you are just making them think about regulatory capture (which is an incidental part of the argument) when instead you should focus on the risks.
I agree that a nontrivial fraction of the population seem to sincerely believe this, presumably due to enemy action. But they’re very much a minority and in general wasting screen time on your enemy’s arguments is a bad move,[2] when you haven’t even made the positive case for your own arguments.
To the extent defusing this minority is worth doing (I think yes), independently making the case (multiple times) against regulatory capture seems good. To be clear a bunch of regulatory capture stuff is already happening (eg the AI companies publicly saying they care about safety and then privately lobbying to dismantle safety measures).
Though again this is just my perspective and you probably have a different audience and experiences in mind than I do.
Though I think laughing in their face is an underrated strategy if you’re trying to convince an audience rather than the person you’re talking to. I’m not very confrontational, especially irl, so this is hard for me but I’m working on it.
Analogously I think Democrats shouldn’t spend campaign ads on crime or immigration, even if they think they can win the object-level point. But of course it makes sense to prep the right counterarguments to defuse any Republican talking points in a debate, and/or be able to demonstrate in actions rather than words that they’re actually good on crime.
They’re not really bad-faith if they’re just making the same point over and over again that you’ve failed to respond to. Like, “Anthropic is attempting regulatory capture.” That’s a good point. I can’t really get engagement with it on LW, maybe because it detracts from our push for regulation. But that promotes it to a regular faith objection, just one that’s not responded to seriously.
Or the fact that labs are constantly flooding the zone with announcements that simultaneously advertise capabilities, threaten to kill you and all your loved ones and sort of flaunt they get away with illegal behavior. Saying they’re hyping it is a regular objection to doing what they want. It’s bad faith to not engage with it.
And the bad-faith objectors might look like the more reasonable people and attract more believers. If you haven’t even admit the obvious fact that they are corporations attempting a power grab, why take you seriously on doom arguments?
And I don’t think the basic AI safety explainer audience is people who already know AI safety basics, right?
Again you know your audience better than I do.
I agree if it’s a dialogue you should respond to it! [1] But imo you shouldn’t start with a presumption that this is the target audience since you are just making them think about regulatory capture (which is an incidental part of the argument) when instead you should focus on the risks.
I agree that a nontrivial fraction of the population seem to sincerely believe this, presumably due to enemy action. But they’re very much a minority and in general wasting screen time on your enemy’s arguments is a bad move,[2] when you haven’t even made the positive case for your own arguments.
To the extent defusing this minority is worth doing (I think yes), independently making the case (multiple times) against regulatory capture seems good. To be clear a bunch of regulatory capture stuff is already happening (eg the AI companies publicly saying they care about safety and then privately lobbying to dismantle safety measures).
Though again this is just my perspective and you probably have a different audience and experiences in mind than I do.
Though I think laughing in their face is an underrated strategy if you’re trying to convince an audience rather than the person you’re talking to. I’m not very confrontational, especially irl, so this is hard for me but I’m working on it.
Analogously I think Democrats shouldn’t spend campaign ads on crime or immigration, even if they think they can win the object-level point. But of course it makes sense to prep the right counterarguments to defuse any Republican talking points in a debate, and/or be able to demonstrate in actions rather than words that they’re actually good on crime.