#artificial stupidity

[jackalovski]

I hate how people treat AI like a search engine when it’s more like some dumb fuck off the street. Badly remembering something they heard about then making up some shit on the fly

[wordsandrobots]

A couple of years ago, I attended a (virtual) conference where one of the main topics was the impact of so-called 'AI' tools on my particular industry. I work in scholarly publishing (on the publisher side -- I know, I know; for what it's worth, I am at least at a company that's actively trying to drive reform, is anti-impact factor, tries to reinforce the value of the work over the journal name, etc) and the application of 'generative AI' to facilitate plagiarism/fake papers is an obvious risk in this sector. Such software could easily be used to overwhelm the (meagre) defences journals have against such things, especially with the pressures placed on academics to get their work into 'high impact' publications above all else. The threat of 'paper-mills' (operations paid to seek publication by fraudulent means) ramping up via the use of ChatGP was clear and present amid those heady days of the initial hype-push.

What's stuck with me from that conference is a panel participant pointing out that 'AI' hasn't created any *new* problems; it's just accelerated existing ones. That is, fraud in science and science publishing has been an issue as long as scholarly publishing has existed as an industry. You don't need a fancy tool to generate you a fake paper. It helps, no doubt, but it's not a necessary step. And yes, it makes detection harder. But the actual solution here -- the way to put a stop to fake papers, dodgy authorship claims, and all the other variations on trying to beef up an academic's publication record for career gains -- doesn't lie in some technological arms-race between plagiarism-detection and paper-fabrication. We need to change the culture. We need to put a stop to the rewards for this kind of behaviour, by assessing academics by the actual value and quality of their research, without the proxy-step provided by place of publication.

(For the uninitiated, it is a huge problem in science that certain journals -- such as the big three of Nature, Cell and Science -- are seen as *the* place where groundbreaking research is published. Not only does this expose the English-language bias within global research, it creates the idea that to 'make it', you must publish somewhere like that, rather than just, you know, doing good solid work. Journals, big name or not, also have a history of selecting for headline-making research. So on the one hand, institutions are judging their employees' careers by their citations, not their work, and on the other, you absolutely cannot trust journals not to get dollar-signs in their eyes when someone comes along claiming that e.g. a certain vaccine actually causes an unrelated health condition. To pick a deliberate, very-specific example. On top of all this, peer review is *terrible* at catching faked results because it has to be approached in good-faith. Most of the time, fraud is only caught in hindsight, once the work has had time to circulate in the community, at which point wider damage has been done.)

Now, one of the reasons I haven't blogged much about so-called 'AI' is that my hatred for it is pre-rational. What I mean is, I hate 'generative AI' with the power of a thousand burning suns. I hate it on a conceptual level. The idea of feeding real people's work, their art, into a machine and have it churn out an approximation of that same work and art is abhorrent to me. I view it as a mockery of skills I have devoted my life to. If it could produce truly breathtaking imagery and crystal-sharp prose, I would still feel the same revulsion at the thought of removing intent from an act of communication, at the idea we should be content with bathetic mirrors in place of engaging with actual human beings and what they can do.

Separate from this, I believe there is good cause to be highly doubtful about the tools that have been pushed on the public over the last few years. I haven't used them myself (see above) but everything I've seen suggests they just aren't very good. It's painfully obvious how they can be/will be/are being used to devalue people's labour, thus strengthening corporations. There's the destruction of the information ecosystem that comes from integrating software intended to reproduce tone instead of facts into major search engines. There's the impact on the actual ecosystem of pouring resources and power into this technology. There's the simple detail that a lot of the people pushing this stuff are, frankly, just the worst.

However, I am extremely, painfully aware I am the wrong person to make rational arguments against these tools because what's actually driving my objection is disgust. I'm going to assume the worst about this particular kind of automation simply on the basis that I can't stand its existence.

There may be good, productive uses for this kind of technology! I can't tell you what they might be because I'm too busy looking for the bit where my worst opinions are validated. That's where I am on this. I actively have to guard my tongue around some of my colleagues, to keep from railing at how gullible I think they're being, buying into these things.

So yeah. Not a good place for making solid arguments. But that point from two years ago -- 'AI' is not creating any new problems.

I think it's easy to lose track of that. Consider the environmental impact. In order for you to read this, some server, somewhere, needs to be powered and cooled. The device you are reading this on is likely made from relatively rare materials that have a history of being source via destructive means (both to the environment and the people involved in the extraction process). I don't say that as a guilt-trip; I'm writing this via the same means. It's simply that the current landscape of our societies is dependent on things that comes at a cost to the planet and our fellow humans. That cost is made worse by rampant capitalism, but even under ideal conditions, mitigating it will require rethinking massive amounts of infrastructure.

This is not an excuse to make things worse. I want to be very clear about that. Nor am I claiming these issues are insoluble. It's simply a good example of 'AI' being an exaggerated case of an existing problem, namely how to balance the utility of modern communication technology against the extractive activity required to build it. As with many things, the glib answer is 'don't do capitalism' and, well, err, that kind of is the answer, reorientating away from the maximisation of profit above all else and from 'endless growth' doctrine. But crucially, that answer has nothing to do with 'AI'. If the hype-train collapsed tomorrow and everyone realised they've been buying snake-oil, and somehow the tech sector didn't collectively burn to the ground about it, we'd still have a problem to solve.

Because the problem isn't new.

That 'summarisation' tool Google or Adobe have swung on you, that shortens text with no regard for the actual information contained within what it's reducing is not some novel horror; it's just an acceleration of the same approach to design that sees 'engagement' as the primary driver, detached from what is actually materially happening to cause everyone to flock to a single place. MidJourney or what-have-you, allowing X or Y group to churn out endless cloying representations of their ideal reality, is just bad Photoshop composites with less effort required on the part of the person pushing the button. People will airbrush reality whether they have to do it with a prompt or an actual airbrush. We know this! Thomas Kinkade made a whole flipping career off it! It's the heart of mass-media advertising, to cheaply reproduce visions of simpler worlds for the sake of selling you something.

The truth is, grifters are going to grift, with whatever tools they have at their disposal. As long as there is a market for snake-oil, an incentive to cheat, a reason for people to be dissatisfied with their lot, there is going to be space for someone to sell an everything-app. A quick solution. An easy fix. We don't address that by playing whack-a-mole with every single dumb vapourware 'solution' that results; we address it by collapsing the space that permits those things to find their marks.

I think it is an objectively bad thing if paper-mills can work faster and easier and flood journal submissions with more junk than ever before. But it is also objectively bad for academia to be held hostage by a for-profit system that silos and constrains their work while being treated as the bar for judging how well they are doing their jobs. And the latter is the problem that actually *needs* to be solved, if we're going to have a hope of addressing the former.

Anyway, thank you for coming to this edition of 'Words sorts through his disgust to work out if there's a sensible position obscured beneath, for the sake of not being a raging arsehole to people who like shiny toys and haven't been in a love-hate relationship with their ability to draw for thirty years'.

[404daydreams]

"It's all an illusion, a shadow that they've created for me to chase, and attain nothingness. The person I look for everywhere I go, on the road, while driving, while static is not even here. It's all an illusion, a drama for insanity. They're not here, they're not here. But a fact stings, they know what's happening but never saved me from the little insanity I befriended. They could have saved me, but they didn't because this is what they want. Controlling someone's mind and soul but I ask, "Do you even have control over yourself?” "

[raindingrandom]

Team Chaotix Wins By Doing Absolutely Nothing

[neovid]

Wow, satire really is dead.

Turns out LLMs have been learning from Onion articles.

I'm liking this trend. Let's get more straight-faced satire out there for AI developers to scrape and sabotage themselves!

[tlaquetzqui]

No Google Docs I don’t think I will be changing “I was hated by my charges” to “I hated my charges”, and I think the entire grammar-check team should be sent into the mines to be worked to death, for the suggestion.

[rabbided]

Sketches inmight finish Because YES! Inlove these two i love proffesr barrajco 3s servant rabbid in “artificial stupidity” hes so me and . Inlove pb2 hes so awesome i love him so much

what do u evej do in this app besides sharign art ❤️

[thisisevenharderthannamingablog]

One of the things I loathe about the creeping prevalence of AI is that I am now sometime required to compose full sentences to search for things, as if I was asking a person, when all I want is a keyword. That's stupid.

[the-most-adorkable-smile]

Yes yes, we've all seen Google's new ai tool return exceptionally stupid results.

Now can you PLEASE stop using Google ai? Remember, this is a tool that's stolen the words of almost everyone who's ever used the internet to train its data; the environmental cost of ai is also prohibitive, using around half a litre of water per search just to cool the servers (which is not counting the energy cost).

There have been numerous documented methods of turning it off, blocking it, or just not using it (switching to a different web browser / search engine being the easiest). Try one of those. Here's a link to an article with some more concrete blocking methods:

[fae-sodapop]

OH MY GODS

WHY DOES EVERYTHING NEED AN AI? WHAT POSSIBLE BENEFITS DOES IT HAVE TO SOCIETY RIGHT NOW????

I'm not using your search engine for your SHITTY AI to tell me to give me a reddit post as a top result.

I fuckin hate it to "We've added an AI assistant for your convenience"

Said convenience is actually ruining the interface and rendering it unusable

[more-wrong]

“skibidi toilet” is actually a documentary from the future, okay. The skibidus were created by Ringo’s Basilisk to destroy humanity, hence why the only people who can fight back are humanoid robots , all the real humans are already dead. This is a grim warning for the people of today

[religion-is-a-mental-illness]

"Ours may become the first civilization destroyed, not by the power of our enemies, but by the ignorance of our teachers and the dangerous nonsense they are teaching our children. In an age of artificial intelligence, they are creating artificial stupidity." -- Thomas Sowell

Don't think this is inadvertent or incidental. Marx was frustrated that the proletariat didn't seem to want a revolution. Critical Theory manufactured and delivered such a justification.

[corvidconventicle]

At some point my phone added AI recognition for photos so it can identify and group photos based on the subject.

So now it has autogenerated an album for my dog and an album for my cat… except that I don’t have a cat, it’s just my dog again.

[toku-fangirl-2015]

Hey Meta AI, what’s going on with the ears/helmet?? Even the “art” they’re using to advertise the feature has glaring errors!

[phantomjeanne85]

Am...am I interrupting something? Should I come back later? O_o

[sakura2arashi]

I've been using Microsoft Word for over twenty years and somehow it gets worse each year...

[Image Description: First image is a Microsoft Word screen shot where it suggested that the user change "mauled" to "mailed", changing the sentence to "The body had been mailed to the point that it was unidentifiable."]