Bit late to the party, but this should prolly be tagged “Paper” on pivot.

I want to put together a little pitch for the data-brained that AI is Not Good Actually®, and this is the most smoking gun I can think of.

Fresh of the press, only in german unfortunately: tagesschau

President of bavaria (and head of the extra fashy bavarian splinter party of the christian conservatives) wants to cut 10.000 public service jobs using AI.

Does anyone have a good definition or classic examples for the term mall ninja at the ready?

I first heard that term on this channel, and I feel like I should understand that phenomenon better.

I have some stale notes on the AI Hacking chapters of Schneier’s “A Hacker’s Mind”, and in one of those chapters he theorizes how AI might hack human emotions. He brings humanoid robots with cute faces as an example.

You know, I think I have an idea how a social countermeasure to that might look like.

Maybe I’ll add, my first though was this must be some kind of LLM. Some kind of compressed model or something. But that it’s just a 20 line string combiner with a handful of templates is somehow even cooler.

Seems topical to Pennycook et al. You don’t need a lot of complexity to make it seem like a lot of complexity.

Nice! Here’s a GitHub link for my fellow sauce enjoyers.

Where did you get the generator & the word supply? Original? Getting the conjugations right must have taken a lot of fiddling.

Seems easy and fun to modify too. Maybe consider slapping on an OSI license?

What’s this “RL” thing?

Thx for the sauce. Im already hooked.

archive link

whatever minor body of irrelevant five-follower internet loons might bother trying to argue the literal uselessness of a predictive text generator

/c/techtakes mention!?

de-paywalled link

I’m interpreting your phrasing as you believing that the non-profit “taking over” is somehow good, because profit motive bad presumably. But regardless of incentives, everyone involved is trying to flood the world with slop by incinerating cash and processors on industrial scales.

But so far the cash incinerator has been running on speculative financial products issued by a club of esoteric computer scientists trying to awaken the robot god. Investors are slightly uncomfortable with this, so Sammy boy is trying to offer them a more traditional vehicle to incinerate their cash (while indulging in his personal profit motive a bit).

You’re certainly hitting some nails on their heads here. The normalisation of AI is absolutely happening, mostly because the buttons start showing up on Google/Microsoft/etc. products with massive market share.

Also the manlyman blogs bitching about beaver hair brushes. I was looking up safety razors in german (“Rasierhobel” btw, totally unaware of that until now), and Wikipedia was referencing one such archived blog, bitching about pig bristle brushes being “drug store” garbage. I might still get one. (The razor that is, not the brush. Spray on foam will do for me.)

Not so sure about the scythe/mower thing. My battery powered mower & motor scythe slap. (Stihl btw)

While browsing some german news media outside my usual territorry (DW and tagesschau), and was fooled by this chameleon of an ad on the front page of WELT (trying for classy, but obvious conservative bias).

The heading means “Bitcoin could protect from inflation”. If you want to check out some retail investor shilling in the wild, here you go!

archive: https://archive.is/4RT80

I can see tante’s point. Besides AI datacenters being used for surveillance tech, I can also see LLM tech itself used nefariously post-bubble. Maybe maintaining an up-to-date LLM as a product is not viable, but a custom-trained model to snipe public online discourse around a crucial election could remain affordable for a wealthy fascist.

On the bright side, I am hoping for a brief period of powerful yet affordable gaming PCs thanks to retrofitted, slightly singed Blackwells.

First I thought “Oh jeez, what a wall of text” but now you gave me my own thoughts that I want to share.

I don’t think callling genAI output “not art” is a very defendable statement. I believe art is ultimately a type of activity, and one that is very hard to draw a strict line around. If I find a cool piece of driftwood and frame it, did I do art? That’s kind of what that artist did when he picked his album cover.

But I also share your sentiment about “AI artists” pretending to work in a medium of which they understand 0% of the nuance. I think it makes more sense to call those people hacks instead of “not artists”, because that’s what you call people who use shallow, formulaic methods to dabble in a medium of which they are wholly incompetent.

And finally, AI as toolset does of course uniquely pander to hacks.

I always thought you could do interesting stuff with genAI, especiall when it goes into mangled, uncanny-valley territory. Though I can only think of examples for visual generators, like this album cover or the AI Pizza commercial.

The only text-based example that comes to mind is I forced a Bot to write this Book and that’s just a guy imitating LLM writing style. (Hillarious though!)

For those who just can’t shake their Wordle habit:

Try https://duotrigordle.com/

32 times the Wordle and none of the NYT enshittification

While browsing the references of the paper, I found such a perfect evisceration of GenAI.

We have confused what we can write down with what we usefully know and compounded the error by supposing that because computers can help us write down more they can obviously help us know more.

The marks are on the knowledge worker - Kidd, Alison

That’s from 1994 folks, they were talking about the wonder of relational databases.