Post

this is the really ineffective head-in-the-sand reaction a lot of folks on Bluesky tend to have if you talk about any harm reduction-based approach to the reality that millions of people are using AI tools today. The current approach to critiquing AI is obviously not working, but they don't care.

Bluesky itself *exists* because people broadly understood that social media then, as with AI now, was having a wide variety of negative effects on society. So teaching people about less damaging approaches had a lot of value and has been effective in giving millions of people better alternatives.

Scolding people for their shitty choices is generally an effective strategy, even if you obstinately disagree your choices are shitty.

Have you ever spoken to anyone in the field of, say, addiction treatment?

Have you ever considered whether there may be chemical differences between using ChatGPT and heroin?

I agree with the intellectual substance of virtually every common critique of AI. And it's very clear that turning those critiques into a competition about who can frame them in the most scathing way online has done *zero* to slow down adoption, even if much of that is due to default bundling.

At what point are folks going to try literally any other tactic than condescending rants? Does it matter that LLM apps are at the top of virtually every app store nearly every day because individual people are choosing to download them, and the criticism hasn't been effective in slowing that?

Show me when we have harm reduction for social media and maybe I will believe that this is an argument that deserves any acknowledgment.

Could you elaborate on the tactics that would replace condescending rants? Explained for a hypothetical someone who might not know any. I think maybe part of the problem is some folks don't know other paths.

LLMs and codegen are pretty central to my job, these days. Seems central to most other semi-relevant jobs I've looked at lately, too. Responding to condescending rants doesn't pay my mortgage, so I'm just more careful and less enthusiastic about where I try to talk about what I'm doing these days.

What the hell are we supposed to do, then? You keep saying we should try some other tactic. There is no other tactic available to us! Most of us aren’t friend with power brokers in Silicon Valley. We do not have the levers that you have. What should we do?

In my job, I am the lone anti-ai voice in what we develop for our business. I demand ROIs, and call out when benchmarks aren't met for the next project that wants to add it. All for our devs using the free tools, but so far nothing that was promised when we signed a check has paid off

People who want to see *less* LLM use are not going to join you in encouraging *more* LLM use, and repeatedly using the word "condescending" as you do is very much a pot/kettle situation. Stop feeling sorry for yourself and getting all stroppy just because you got ratio'd.

This is fair enough, I agree scolding is not the answer, but I don't see how harm reduction education will get through to individual users who, as you point out, are often experiencing LLMs through bundling by established providers.

Other tactics that folks are trying include: * Lawsuits; * databases of AI harms; * pushing for regulation; * sharing info about AI harms, lack of ROI, plagiarism, etc; * sharing how-tos and resources I cover this sort of thing in my newsletter if you're interested buttondown.com/aiweek

It’s not for lack of criticism. It’s the total lack of information about what any tech actually does. People think LLMs are just more of the same; unaware that it’s explicitly not, explicitly incapable of fulfilling the label “AI”. The tech info sphere is dominated by corporations & grifters.

I don't buy the wokescold idea; if this was any other social issue, this thread would be quite gross. It's quite clear at this point you can't talk people out of doing harmful things, but there's also no other form of action that seems to work either.

Defining some development standards for A.I projects in your company is helpful. Things like archiving prompts for genai builds, preserving non a.i APIs , requiring a.i specific pentests , having a flow chart for when and when not to a.i. etc..

I was with you all the way until the “look what you made me do” meme moment at the end.

I'm genuinely not following what you're saying here. Where did I say something like that?

Someone becoming friendly with the racist/fascists who are nice to them because the left is mean to them is literally meme images.dailykos.com/images/57480...

I get so tired of the notion that we (people) should not be scolded for our choices. Life has funny way of letting you know whether or not your choice was right. Sometimes it takes the arc of time to reveal the answer. But time is a finite resource so we should be far more mindful about our choices.

Nobody is saying that people shouldn’t be scolded for their choices. What is being said is that scolding people for their choices is a piss-poor way to get them to change those choices.