Notices by Mallory's Musings & Mischief (malcircuit@thingy.social), page 3

A large language model is just that: a model. It's a statistical function. A black box. Data goes in and data comes out.

Put the burden of proof on the AI proponents by not implicitly accepting their premises. Maybe a pile of numbers can "think", but make them explain how that works. Don't let them rhetorically handwave away that inconvenient fact.

And if someone genuinely does believe LLMs can think, push them on the ethical issues with creating a sentient mind that is born into enslavement. Because that's what is happening if you actually believe something like ChatGPT can "think" or whatever.

You conjured a mind out of silicon to write your emails, which is probably orders of magnitude more immoral than the environmental damage caused by these systems.

If you are someone that is opposed to "AI" systems based on LLMs, I challenge you to never talk about them in a way that anthropomorphizes them. LLMs don't "think", "know", or "understand" anything. There is no mind there. No "self" as a human would experience.

When you anthropomorphize, it only feeds into the AI bubble rhetoric.

Some #aurora pro tips:

- Look north (unless you are in the southern hemisphere, then look south).
- The aurora is extremely temperamental and can change quickly. You might need to wait and watch for a while before you see anything.
- Light pollution is your enemy. Get out of town if you can.
- Give your eyes several minutes to adjust to the low light.
- The aurora is much dimmer in person than you see in pictures.
- The camera on your phone will pick up the aurora much better than your eyes.

There's a giant flame in the sky lol #aurora

So uh, if you have a clear sky RIGHT NOW you might want to go outside and look north to see the #aurora. Take a pic with your phone if you're not sure.

@thomasfuchs Oh, I have no doubt. I'm just saying even without that, it's still an issue.

@thomasfuchs It's a fundamental technical flaw in how the system works, not simply how it's implemented. It's part of why I will never trust them without extensive evidence that they can be trusted and why.

@thomasfuchs It's not even necessarily that they are designed to operate in this way (though, they probably are). This is how the system works. It's what it does. It's a black box: you give it an input and it generates an output. Always. Even if you tell it to say nothing, it will respond with something. Even if you ask it for something that's hazardous to yourself or others, it will respond.

Maybe you might wonder why talk a lot about LLMs and the current "artificial intelligence" bubble when there are so many more serious things going on in the world. Why talk about technology when politics has become so unhinged and terrifying?

Because in my eyes, the AI bubble is fundamentally a political movement. It's a perfect example of capitalist excesses and the fundamental ethical flaws in our society. It's a way to radicalize people without getting mired in partisanship.

I hate when I get so depressed and irritable that I unintentionally end up being a dick to people

My issue with AI is not about practical utility, it's about moral utility. I have come to the conclusion that it's a technology for which any potential benefits are massively overwhelmed by the costs — material, financial, and spiritual.

I have a hard time imagining how such a thing could — even in principle — improve my life or anyone else's in a way that could be done better by society actually deciding to give a fuck about whether or not all people are healthy and thriving.

I like how I openly admit to being an AI hater and people respond with stuff like "I use it all the time for <whatever> and find it to be useful," as if my opposition is simply because of its utility.

I would avoid using LLMs even if they worked perfectly and did everything the boosters claim they will do.

It's a philosophical disagreement, not a functional one. I don't want to subjugate myself to a machine.

But that's also not an excuse to blindly accept what "experts" say either. The point is to *accept your ignorance*, and to NOT propagate information you do not understand.

Most of the shit you see online is an opinion of a summary of a summary of a summary. A LOT of shit got lost in translation, even if everyone is well-intentioned.

You are not the arbiter of truth in every situation. Be okay with not knowing the answer, and don't pretend that you do.

If you really want to know the truth behind something, be willing to put in the work. Remember, for any topic you are curious about, there is a person (likely many people) who have literally spent their entire adult life immersed in it. It's true that no single person knows everything and is always right, but sure as fuck there are people who know more than you do.

Don't just ask experts for the answer, ask them *how they know the answer*. That's what makes them an expert.

This applies equally to skepticism itself. I have my tips and tricks, but there are experts in this too. People who have spent their entire adult life endlessly debunking bullshit, misinformation, and pseudoscience. Try looking up what they have to say on the topic.

Don't believe any single person's opinions on any subject, especially mine, and especially when it comes to skepticism. Try to figure out what the experts agree on, disagree on, and —most importantly— why.

FIN

At one point when I was visiting my mom on Labor Day weekend we were talking about the state of the world and she asked me, "How does someone know what's true anymore?"

The first place to start is to accept that, even if it's possible to know the truth, there is no easy or straightforward way to find it. It's hard, messy work, because the world is a complex, messy place.

The second thing is to automatically be skeptical of anyone who claims to have an easy solution to the first part.

A phrase that I've had stuck in my head for a long time is "things are _always_ more complicated than they appear."

It doesn't matter what it is — the deeper you dig into a topic, the more complex and nuanced it becomes, the more uncertainties creep in. Always. Whatever you think you know, there's always a deeper, more confusing layer. Even if you are correct, there's more to the story.

Accept that, deeply.

Be comfortable with your own ignorance. If you don't know something say, "I don't know". If you are wrong, accept it. Hold your beliefs loosely.

Likewise, do not trust people who refuse to admit their own ignorance or when they are wrong. Don't confuse confidence with accuracy. The most trustworthy people will honestly and openly tell you the limit of their knowledge, and will be hesitant to make any claims beyond it.

Whenever possible, get information directly from the source. Read the fucking research paper if you really want to know. I know, it's full of incomprehensible jargon, big words, and inscrutable statistics and plots.

Which is the point. The truth is fucking complicated and hard to understand, and I guarantee the headlines about that research paper make it seem simple. It's not.