What's Your Best AI Safety "Quip"?

post by False Name (False Name, Esq.) · 2024-03-26T15:35:33.178Z · LW · GW · No comments

This is a question post.

Contents

No comments

Motivated by thinking gay rights were advanced by asking "When did you choose to be straight?" Which emphasised that what isn't a choice and doesn't harm others shouldn't be proscribed. Here, we're seeking a memetic way of framing the fact that the alignment problem is unsolved.

Author's "null quip": "Can you get a 5-year-old to do what you want? The way you want them to do it? Every time, guaranteed? What about your parents/partner/strangers, et al.? No? So there's no general way to get even a five year old to go your way - and for a computer that's able to do most of what we can (and you've seen that it's getting to be that way...?), and so much faster than we can, we're going to have much more trouble than a toddler can give us getting that to go our way, without a lot of time and thinking about that problem, no?" 

Then introduce the petition, candidate's name, funding bucket, etc. Or, could just use it to defuse an optimist's reason-free optimism in a debate (as argument by analogy from one intelligent being to another, this doesn't seem fallacious).

The Question: can you think of anything as common-sensical and dispassionate as this^[1], but more concise, more memetic? And do you know any way to "implement like hell"?

(Though, as Nvidia's "Blackwell" development attests, the Invisible Hand has determined to slash its own wrists, willing to risk it, anyway - and now it has a blade long enough for everyone else's too - and nor quips nor revolutions on history have impeded whatever Mama Money wants - Mama gets...)

1. ^{^}

   For which reasons the "doomer" tag must go: doomer is doom-sayer, a definitionally marginal figure, best ignored. An AI "Safety Advocate" can't be poo-pooed without looking like/being an idiot inclusive-or asshole who hates being safe and alive.

Answers

No comments

Comments sorted by top scores.