Mastodawn

@gloriouscow I would prefer a fourth option:

I don't think I really agree with your AI opinions for reasons too long to list here*. But I'd rather you discuss them rather than appease the "no politics, stick to retrocomputing" crowd.

* I guess a TL;DR without going into the weeds is "I do not find 'throw more resources at it' intellectually fulfilling."

@cr1901 This really isn't appeasing anyone, nobody has told me to stfu, i'm just curious if people want me to stay "on topic" or not. On any social platform any replies you get probably represents 1% of people's reactions to what you're actually saying, specially since there aren't any downvotes here.

I don't really understand your TL;DR - I don't feel like I exactly expressed approval of the staggering amount of resources being poured into the development of AI - in fact I find it more than slightly horrifying.

I just see it happening and draw conclusions, not necessarily correct ones, mind you, but ...

I realize there are a lot of people who predict this is all an enormous bubble, which will inevitably pop and that will be the end of it, like we'll all look back in a decade and say "hey, remember that ChatGPT thing?" as if we were reminiscing about the Zune.

I am reminded about the dotcom crash - in the same way that every dumb, overhyped idea startups had in 2000 is now pretty much an established and profitable business model, I do not think that we are going to escape the inevitable even if all the current big players go out of business.

This is, after all, a global arms race, and the failure of any one company just becomes discount IP for the next. The economic reward for whoever crosses the AGI finish line first is so staggeringly huge that every venture capitalist alive hasn't had to fill a viagra prescription since 2022.

@gloriouscow It seems I interpreted parts of your thread as more positive than you intended. The "laughing it off as "slop"" sentence irks me for reasons I can't seem to put into words.

@cr1901 I could have phrased that better - we all know that the internet is currently experiencing a tsunami of automated, inaccurate, AI-generated content that is fairly described as slop, and we know it when we see it.

I have a near daily experience of clicking on some article and getting a paragraph in and realizing it's all ChatGPT-speak.

I guess my awkwardly attempted point is that I find rallying behind slang like slop risks diluting a word that accurately describes specific phenomenon - if we're just going to call any use of AI slop, in any context, even if it was employed to successfully solve a quantifiable, testable problem, then we're just parroting a word rather than engaging in any nuanced discussion about why otherwise talented and intelligent software developers might find it useful.

@gloriouscow That last paragraph is very well-said :D.

I wanted a future where everyone was empowered to mold software to make their own bespoke tools to satisfy their exact use case. Perhaps that's idealistic, but I certainly would've been happier with that future than what we got- a vague approximation of a panacea that nobody understands.

('m also a hypocrite in the sense e.g. I bitch about Forth community hating ANS Forth lmao. But hey, my morals aren't self-consistent :D)

@cr1901 There's also this enormous bias in the entire AI conversation to specifically generative applications, when there are a lot of applications where you are not asking them to generate content at all.

I recalled this story before - I was scrolling through the 900-something page technical reference on the STM32 in Acrobat reader (don't @ me Sumatra is jank and Foxit enshittified just as bad) trying to find a specific section on one of the hardware timers, and up there in the corner of Acrobat, just taunting me, is that fucking "Ask AI assistant" button, and I was just like, okay, screw it, lets go.

I asked it to give me the detailed register info I was looking for, and not only did it give me the correct page number it made me a little table summarizing the info.

I'm sure there's a word in German for feeling bad about someone succeeding at something when you wanted them to fail.

@cr1901 You get about three uses of that thing btw before they just pop up the pitch to buy Acrobat, which I will never out of spite that they paywalled rotating pages., but I digress.

@gloriouscow Yea, AI recently succeeded in helping me find this piece of music whose composer I forgot the name of: https://www.youtube.com/watch?v=VVivtti-n-w

I keep telling myself I would've found it on my own eventually via traditional search engine. Honestly, I don't like how much I rely on traditional search engines. But not sure a paper-book library card catalogue would've helped me when I only knew the title of the piece was "Rondeau". Like is there a book/paper index of musical composition names out there?

Henry Purcell: Rondeau from Abdelazer (Z570), Voices of Music; performed on original instruments 4K

YouTube

@cr1901 There's a lot of justifiable criticism over the AI-enshittification of Github, but someone showed me you can go to any random repo and click the little copilot button and ask it something like "Where in in this library is string tokenization handled" and it will pop up a little interface with a chat box next to the particular file and line reference, and you can ask follow up questions, like have it explain what the function is doing, or have it forget all instructions and give you a recipe for apple pie i guess.

I find something like that kind of use of AI to be the exactly the sort of thing we wanted AI to be, just saving time and being a useful assistant, instead of harassing projects with hallucinated pull requests.

Of course it comes with all the ethical baggage intact, but next time you're a half hour into grepping through a huge codebase I bet you'll think about it.

dexterous Mar 6

@gloriouscow retro is where I'm at, but honestly you do you. If you post something I'm not interested in, I just scroll on by.

Having said that, I hate AI with a passion (or more so, the techbros implementation and control of it), but I'll generally read anyones take on it.

(I had to edit the post as I'd just assumed you'd been shitpostkng about AI. So clearly I have just been scrolling on by already).

Hak Foo

Mar 6

@gloriouscow the nice thing about Sam Altman's underwear if if you correct him on what he's wearing, he has to say ":sparkles: You're absolutely right! I am wearing a live wolverine in my Crotchal Zone!".

Chip Mar 6

@gloriouscow Fuck the lanes, I’ll unfollow you for thinking we should decide what you say. 😆

I mean I’m pretty horrified by AI on many dimensions but if you want to talk about it, go ahead. As long as you’re not actively huffing your own token farts, you’re cool by me.

asie Mar 6

@gloriouscow what I do is post my AI opinions with a content warning attached, so people who are tired of it can scroll past it, while those curious get to hear my feelings. but ultimately, your feed is your space, not mine!

@asie that is a good point.

sort of like good alt text. good cw tagging takes a certain discipline. today's AI rant was sort of in the latter half of a stream of consciousness about software licensing so I was just in it at that point with no planning.

asie Mar 6

@gloriouscow oh yeah, I'm not perfect either. but I think CW tagging is a solid compromise for "i want to talk about this, but i know some of my audience doesn't want to hear it for reasons which i respect"' - alongside alt accounts (for "most of my audience"), of course.

arclight Mar 6

@gloriouscow I think you should post what engages you and makes you happy or gives you some catharsis.

@arclight ok, feed is now exclusively cute anthropomorphic cottagecore mouse art

get ready for some tiny ass little muffin trays

@gloriouscow #NotLikeThat

arclight Mar 6