Should we cry “wolf”?
Epistemic status: I am confused and trying to become less confused.
So, our shepherd community noticed the wolf. It’s a small wolf, about the size of a chihuahua, and is unlikely to cause any serious damage. Should we cry “wolf”?
Possible arguments against crying “wolf”
When someone cries “wolf”, villagers think that what is really meant is “a big wolf that is going to eat us all”. Therefore, villagers will decide that we are lying, and will not listen to us next time.
Maybe, publicity is useless, and can even harm us, we need to develop anti-wolf weapons ourselves.
In fact, we’re not sure if it’s really a wolf and not a duck with a strange combination of mutations that makes it look like a wolf, run like a wolf, and howl like a wolf.
If this miniature wolf bites someone’s fingers, villagers may finally recognize that wolves are dangerous.
Possible arguments for crying “wolf”
If we now ignore a chihuahua-sized wolf, it makes it more likely that we will ignore a corgi-sized wolf, then a sheepdog-sized wolf, and then a normal-sized wolf.
If we now cry “wolf” and no one comes to the rescue, then later we can use this precedent to argue for the creation of adequate wolf prevention systems.
If we now cry “wolf” and help comes to us, then next time villagers will already have the experience of responding to the cries “wolf” and it will be easier for them to come to the rescue again.
We have serious concerns that sufficiently advanced normal-sized wolves will have stealth technology that allows them to pretend to be the size of a chihuahua.
What is the relative weights of these arguments?
What arguments did I miss?
These are wolf cubs the villagers are raising themselves and proudly showing off at wolf cub fairs (so cuuute!), expecting them to be great help with the actual shepherding once they grow up. The popular concern is future shepherd unemployment.
(Imagine crying “wolf” in this narrative.)
Wrapping up arguments 1 and 3 against into one, if we cry wolf at a chihuahua, or cry wolf for any reason but that we think there is a wolf, we are using language in a way bumps our own discourse level up a simulacra level or two. So, we are using language not to communicate our best estimate about reality but to communicate group allegiance or suggested courses of action.
As you say, one consequence of this is that the metaphorical villagers may justifiably decide to trust us less, because they realized we are not using words to describe reality. The more important consequence to me is that our own ability to communicate, coordinate, and describe reality will become impaired.
I think just truthfully crying “let’s prepare for real wolf”, because they clearly didn’t train the miniature one well enough, has fewer downsides.
I think the parable/metaphor is hurting much more than helping. Not only is the fable pretty stretched as to actual villager behavior, the threat and context are FAR removed from any actual modern behaviors.
Propose an actual action, and analyze it’s likely effects.
While you can “cry wolf” in maybe useful ways, you can also state your detailed understanding of each specific situation as it arises and how it specifically plays into the broader AI risk context.