#oneaday Day 283: We should probably be resisting generative AI more than we are

There was a good piece by 404 Media on "AI slop" today. Author Jason Koebler described the issue as AI slop being a "brute force attack on the algorithms that control reality", and goes on to explain how those taking advantage of AI are exploiting social media algorithms to such a degree that platforms are now flooded with this garbage, making it hard to find 1) anything made by a real person and 2) anything made by someone you might actually want to connect with.

There is zero value to this stuff, other than self-fulfilling engagement. Presumably the long game is to build up "the numbers" with this shit, then sell the accounts, or make bank off impressions-based ad revenue. And the platform holders don't give a shit; as Koebler points out in his piece, it seems like Mark Zuckerberg actively wants the experience on Facebook to be real humans arguing over AI-generated slop rather than anything real and meaningful.

And I don't understand why we're letting this happen. Not only on social media, but in more "traditional" industries, too. It's happening to a frightening degree in publishing, with myriad "get rich quick" schemes fundamentally being based on churning out multiple AI-generated books every week (or even day) and then profiting off, let's face it, vulnerable people who aren't able to tell the difference between garbage churned out by a robot and something written by an actual human being.

As Koebler puts it, "there is a dual problem with this: it not only floods the Internet with shit, crowding out human-created content that real people spend time making, but the very nature of AI slop means it evolves faster than human-created content can, so any time an algorithm is tweaked, the AI spammers can find the weakness in that algorithm and exploit it."

At the moment, there are a few common responses to generative AI:

"I love generative AI! The genie is out of the bottle, so if you're resisting it you're a Luddite who isn't embracing the latest technological innovations!"
"Generative AI is just a tool that people can add to their arsenal, like digital art packages. I can't really tell you how or why that's a good thing, but I heard someone else say it so I'm saying it too."
"Generative AI might be useful in certain circumstances, but I can't really tell you what they are because no-one really knows or can offer specific, concrete examples that aren't prone to hallucinations to such a degree to make them worthless."
"Generative AI sucks balls and I hate it."

I'm somewhere towards the bottom of that list, leaning towards hating it and very much wanting it to go away. At present, I am disinclined to trust the people who claim it will be "revolutionary" for things like medicine, because of the amount of times it fucks simple things up, still. I am also concerned for the field of programming, because as more and more junior coders show up who are only capable of feeding prompts into an AI, not actually doing (and checking!) the coding themselves, we're going to have a real problem on our hands with software development.

At the same time, I'm sure there are some worthwhile use cases for a means of communicating with a computer using natural language. I mean, hell, look at Star Trek; the assumption there was that you could just say "Computer" like you say "Alexa" today, then rattle off an often fairly abstract task for it to complete, and it would do it. That is, presumably, the goal.

But AI isn't there yet, not by a long shot, which is why ChatGPT costs $200 a month for a subscription and can't really tell you what it's for, let alone how to stop it making stupid mistakes, and in the meantime the companies involved in all this shit are burning through both money and the planet's natural resources in pursuit of something which might, in fact, be impossible. "Agents" are coming, apparently, but all we've seen of them so far is making things that are already pretty straightforward to do on the Web (like grocery shopping) actively more cumbersome, and OpenAI's "deep research" tool is utterly laughable at this point, pulling out citation-free forum posts and SEO-optimised slop ahead of actual, worthwhile information written (and reviewed) by humans.

You, reading this, almost certainly know all this, and perhaps you've even read or shared some articles talking about the problems with AI slop and the problems that is causing all over the Internet. But what have you done about it? Because I feel like we should be doing more about it, rather than just pointing and tutting at it, going "whoo, lad, that generative AI sure is a bit shit, isn't it? Someone should do something about it."

The trouble, of course, is that it's difficult to do anything meaningful about it, particularly when big corporate entities like Microsoft are the ones forcing it onto people through things people use every day like Windows, Office 365, and even the bloody Xbox. I mean, sure, you can find ways to disable it when it does show up, but these workarounds often end up circumvented by the corporations, meaning you need to faff around even more to get rid of the shit. And sure, you can install Linux, but that carries its own burden of needing to know how to do that. Which you and I might be comfortable doing, but what about people who use computers more casually; those who don't know how they work, but just want to be able to get on with simple tasks without intrusive AI features popping up every few seconds?

All we can do, really, is make a specific effort not to use generative AI tools when there are other alternatives available. I will never, ever use generative AI on this site, MoeGamer or my YouTube channel to produce words, scripts, images, thumbnails or videos, however tempting it might be as a "quick fix" to get something done. If that means there are things I either can't do or would have to pay a specialist to be able to do, I will either go without the thing or pay a specialist. Or perhaps even learn how to do the thing myself.

That's a crucial one, I think. Over the years, I've learned how to do a lot of things on computers simply by running into an issue I don't know how to solve, researching it myself and learning how to deal with it. Some of that knowledge I've retained, some of it fell out of my brain the moment I finished using it, but on the whole I've had a net gain on knowledge simply through running into problems and taking the initiative to learn how to fix them myself. I suspect many people who grew up with computers throughout the '80s and '90s are the same.

I'm not going to tell you what to do. But I am going to tell you what I'm doing:

I will not use ChatGPT to research anything, when perfectly good information is available through well-established, reliable, trustworthy and peer-reviewed sources both online and offline.
I will not use AI image or video generation for anything, period. If I need an image or video of something, I will produce it myself, search for a usable (and suitably licensed) stock or otherwise publicly available image or video, providing credit where appropriate, or just not use that image or video.
I will not use AI voice generation to make a "famous" voice say something it never said. Even if it's really funny. I will freely admit to having done this in the past (only among friends), but that was before we really knew or understood the numerous negative impacts that generative AI has on both the environment and on culture.
I will not use AI to create content for the sake of content. I write here because I like writing. I write on MoeGamer because I like writing about games. I make videos because I like making videos. I am not entitled to a "share" of the Internet based on the volume of stuff I churn out, nor am I entitled to be able to make a living from it. I will not pollute the Internet with meaningless slop.

Someday, there may be a valid use case for generative AI. I am open to that. Right now, I do not believe that is there, and I believe the continued proliferation of generative AI online is actively harmful to the Internet specifically, and human culture more broadly.

It needs to stop. But I'm concerned the "genie in a bottle" people are right, and that now we've started this process of enshittifying the entire Internet, we can't stop it again.

But we can make our own little corners of the Internet a safe haven away from the deluge of sewage. And that's what I'll continue to do.

Want to read my thoughts on various video games, visual novels and other popular culture things? Stop by MoeGamer.net, my site for all things fun where I am generally a lot more cheerful. And if you fancy watching some vids on classic games, drop by my YouTube channel.

If you want this nonsense in your inbox every day, please feel free to subscribe via email. Your email address won't be used for anything else.

#oneaday Day 257: When it's bollocks, say it's bollocks

I am an avid reader of Ed Zitron's blog (sorry, newsletter, because apparently that's just what we call blogs now) Where's Your Ed At? If you're at all interested in the tech space, I highly recommend you subscribe or at least check in on it regularly, because Zitron is one of the only people in the space who has the balls to say it like it is: that an awful lot of what is coming out of the mouths of tech companies right now is complete and utter bollocks.

Today, a story went round about a research project at Microsoft where they were using generative AI for "game ideation", and also noted that they thought they could use their generative AI models for "preservation". This was reported on by Tom Warren, senior editor at The Verge, thus (screenshotted rather than embedded 'cause the coward deleted it after everyone dunked on it):

Now, if you know anything about video game preservation, you know that feeding an old game into a generative AI model and then hoping it will hallucinate at least a rough approximation of the original game experience is not "preservation". It's bastardisation at best, a completely useless endeavour at worst, and a massive waste of energy and money regardless of the result that comes out of the other end.

Game preservation is a problem that, for the most part, we have solved. We have excellent software emulation solutions, built over the course of decades of development. Hardware emulation via FPGA at an affordable cost for the general public has advanced hugely in just a few short years. Software libraries for pretty much any system you can think of are archived in their entirety at numerous places across the Internet, and strong strides have been made in providing commercial, legally relicensed versions of classic games for a modern audience, both on existing modern systems and on bespoke emulation-centric devices.

So why, then, why the fuck would we want a generative AI model to make a best guess at what a video game that already exists and has been preserved perfectly well might look like if you play it for longer than 10 frames?

That paragraph above is what tech journalists should be asking. And the reason I bring up Ed Zitron at the start of this post is because he's one of the only people to actually ask questions like this: to take a look at the utter garbage being spewed by today's tech companies and to say "this is complete horseshit, what the actual fuck are you on?"

And Zitron, being an outspoken type, is not afraid to call out today's tech journalism space for not doing this. And he's absolutely right to do so. It is the tech journalism sector's job to look at what it going on, to realise that it is complete horseshit and then have the confidence to say that it is complete horseshit.

But they won't do that, for a variety of reasons. Advertising deals. Exclusive access. PR partnerships. An inexplicable desire not to rock the boat, despite the fact the boat has a huge hole in it and has been steadily sinking for 15-20 years at this point.

I'm not one of those people who thinks that journalists are taking bribes for positive reviews in literally all circumstances — I have experience in the industry, remember, and the most I had to worry about in that regard was a mild admonishment from my editor for criticising a Mortal Kombat game's DLC plan when Mortal Kombat was the cover game for that issue of GamePro.

But come on now. Tech journos should be looking at this utter garbage that keeps getting flung our way, and instead of declaring it "interesting" and doing the stupid looky-eyes emoji that makes their post immediately look like a 14 year old girl wrote it, they should be going "hang on a minute, what does that actually mean?" then exploring it further, asking some probing questions (which inevitably won't get a response, but that in itself says something) and then confidently declaring the latest generative AI "innovation" to be what it is: complete and utter horseshit doused in the finest snake oil.

And people wonder why the entire journalism sector is floundering. Could it perhaps be because very little actual journalism seems to be getting done?

Shout-out at this point not only to Ed Zitron's aforementioned blog, but also the excellent coverage of the Elon Musk nonsense in the States by Wired's politics department, 404 Media being a rare example of tech journalism that actually asks those hard-hitting questions, and Aftermath for doing something similar with games journalism. There are still people doing good work out there. But the people on the big, well-known mastheads, like Warren above, need to step their game up, stop being so incredulous and start acting like actual journalists.

If you want this nonsense in your inbox every day, please feel free to subscribe via email. Your email address won't be used for anything else.

#oneaday Day 179: Your occasional reminder that AI can fuck off

I saw a TV ad for "Apple Intelligence" yesterday. The concept of the ad is that someone is angry someone at their workplace keeps stealing their pudding — hahaha, so hilarious and cosy and relatable — and writes them a furious email. They then click the "Friendly" button on Apple Intelligence and the email is rewritten to be the most milquetoast, handwringy, insincere thing you've ever seen. And this is supposed to be a selling point.

Elsewhere, a YouTuber I know had someone in their comments getting pissy about how they pronounced "ZX81", and, presumably in an attempt to further their argument, the commenter in question then copy-pasted a ChatGPT conversation — without editing out the "ChatGPT says:" bits — that didn't even particularly help their cause.

I keep seeing YouTube thumbnails made with AI art-stealing machines. Coca-Cola made a Christmas ad with AI. The memorial lunch for beloved broadcaster Steve Wright had an invitation that was made with AI. Entire websites are made of AI slop. And even here in fucking WordPress, I can't escape the sodding "Generate with AI" button.

I fucking hate it. I want it to go away. I want people who say "but it's good for summarising things" to drown in the sea. I want people who say "but it's better than doctors at diagnosing problems!" to be the victims of the worst malpractice the medical industry has ever seen. I wish eternal loneliness and desolation on those who use it to write emails. And I want it out of the pieces of software I use on a daily basis.

We're even starting to get accounts on BlueSky that pretend to be real people, but simply respond with ChatGPT answers that are tuned to be deliberately argumentative. What is the fucking point of all this shit? How is it benefiting humanity and productivity in any way whatsoever?

It isn't. All it's doing is continuing to make tech worse, year on year, while keeping oblivious shareholders — who aren't interested in anything but seeing "growth" — happy that companies are providing supposed "new innovations" that actually don't provide any sort of useful functionality whatsoever.

I'm aware I'm ranting incoherently, but honestly right now it feels like it's pointless to even try and come up with a cogent argument. This shit is infesting everything, and it's becoming impossible to escape from. And I legitimately do not understand how anyone can possibly think this shit is in any way better than what we had before.

I guess the one upside is that with how much AI is being used pointlessly to provide "summaries" of Google Searches, YouTube videos and other such shite, the planet will burn down all the sooner, so eventually we won't have to worry about it at all. Then the Great Thinkers of the day — assuming anyone survives — can stroke their chins for two hundred years about "where it all went wrong".

Here. Here is where it all went wrong.

If you want this nonsense in your inbox every day, please feel free to subscribe via email. Your email address won't be used for anything else.

#oneaday Day 154: If you can't create without AI, you shouldn't be creating

YouTube recommended me a video earlier. I'm not going to share it because it's shit. But I will tell you about it. It was a video supposedly "reimagining" the video game series Streets of Rage as a "movie". If you've spent any time on social media in the last year or so, you already know what's coming: yes, it's entirely AI-generated drivel, and the complete content of the "movie" consists of nothing but people staring moodily at the camera while there's a slight pan around them, and people walking towards the camera in slow motion.

The video has a quarter of a million views, and a comments section full of people gushing effusive praise over the "creator". This is not the first video that this channel has put out like this; to date, over the last year they have spaffed out 215 videos that are all like this. And not only that, their About section on their channel is very transparently written by ChatGPT.

I hope you don't think it controversial of me to say that if you can't create without using AI, you shouldn't be creating at all. I will, under extreme duress, concede that there are certain uses of AI tools which might be useful as part of the creative process. AI music tools, for example, can be quite effective if you feed them some lyrics you've written yourself. (Getting AI to write lyrics results in extreme garbage.)

But if all you do is put shit into an AI video generator and then spaff it out on YouTube — and you can't even be arsed to write your own bio — you are not a creative person.

The standard argument that insufferable AI types like to trot out at this point is that generative AI "democratises" creativity by allowing anyone to "create". Except it doesn't, because the output is not your own work; you haven't created anything. The result is a pile of plagiarised crap that is immediately identifiable as the product of generative AI.

And anyway, creativity is already democratised. There is tons of free software out there that anyone can download and use, for computers, phones and tablets. YouTube is full of video tutorials on how to get started, improve your techniques and take on some advanced challenges. There are several decades' worth of text tutorials archived across various websites from over the years.

There is no excuse for taking the lazy option and just feeding a prompt into the lake-boiling plagiarism machine. If you want to be creative, be creative. But know that it takes work. And that work is worthwhile, because people will be able to tell when you have put that work in. Those who follow you will see your work improve, adapt and change over time. You'll find your own unique "voice". And there really is nothing quite like being able to express yourself in your own distinctive way; being able to sit back and look at something you made and think, with a slight smile on your lips, "I made that".

With generative AI, there is none of that. You get derivative, immediately identifiable slop that all looks the same and all contains the same basic errors. You never have the experience of refining and improving your own work, because you haven't done any work. You never find your own unique voice, because you are beholden to the algorithms and training data that the LLM you're using has stolen from all over the Internet. And you never, really, truly, express yourself.

In the meantime, until you come to the realisation that all you're doing is spaffing out hollow, soulless pieces of crap that look like creative works but are actually anything but, you are poisoning the entire Internet with your garbage… and, judging by the comments section on the video in question, poisoning the minds of people who aren't familiar with just how fucking easy it is to tell the magic art-stealing machine to steal some art to your specifications.

The planet is doomed. And no-one gives a shit. Sleep well!

If you want this nonsense in your inbox every day, please feel free to subscribe via email. Your email address won't be used for anything else.

#oneaday Day 144: Are we fucked?

With each passing day, I feel like it's getting harder and harder to ignore the feeling that we, the humans, are completely and utterly fucked.

And I mean this in a variety of ways. For one, it's impossible to ignore the wilful ignorance "big tech" is demonstrating in the current "AI" gold rush. After years of getting things like carbon emissions and sustainable energy production into a good place — likely far too late, but still, the effort was happening — it seems all the big players in the tech space have just gone "haha! Fuck that, we've got a new toy, and bollocks to the frankly unnecessary amount of energy it consumes to power it".

And this is about the third or fourth energy-inefficient tech gold rush at this point, after cryptocurrency and NFTs. The one thing those things and "AI" have in common is that they're hawked by people who are completely unable to explain how their supposed benefits outweigh the absurd cost in resources required to use them. At worst, they're used as outright scams.

AI is even worse at this point. In the last few weeks, I've seen several adverts for "AI". Not an AI product, not a revolutionary and groundbreaking use case for AI… just adverts for "we are a tech company and we're doing… something… with AI because every other fucker is". It is absolutely telling that not one of these adverts has been able to depict AI doing anything remotely useful, and that none of them even attempt to explain why AI is a good thing. Because, as most people who have been paying attention will know at this point, there is no product.

Seriously. There is no compelling use case for AI that isn't already covered in a more energy-efficient format by existing tech.

Want to find information? Search engines exist, both for the whole Web and within a single site.

Want to write code? Well, you'd better learn, because ChatGPT is going to spit out bullshit that is full of errors that you won't know how to spot without knowing how to code.

Need an image? There are billions of images online, many of which are royalty free or suitable for use via fair use provisions. There are free art packages available. And there are lots of artists who will draw whatever the hell you want — yes, even that — if you give them some money.

Can't write an email? Bullshit you can't. If you can write a ChatGPT prompt you can write a fucking email. Stop being a lazy cunt.

Need AI to "summarise" something for you? Just read the fucking thing, it's not hard, and as a species we've spent several thousand years mastering that basic skill.

So that's a concern, both for the environmental impact and for how it will affect the job market. I'm also rather concerned about how medicine seems to be super-keen to use AI rather than, you know, human doctors. I'm sure that's going to be a fun few lawsuits in the near future.

And outside of all this nonsense, we have the current state of world politics, particularly the US. I have friends in the States who are — quite rightly, I feel — legitimately terrified about what the upcoming election will result in. And while I thought people were overreacting somewhat back in 2016 when this situation last presented itself, having seen how utterly deranged Trump and his most obsequious sycophants are behaving in the run-up to this election… yeah. I get it.

So the inevitable conclusion to all this is to find myself sitting alone in a hotel room asking myself "are we fucked?" Cause I think we might be, y'know. I think we might be.

If you want this nonsense in your inbox every day, please feel free to subscribe via email. Your email address won't be used for anything else.

#oneaday Day 19: The AI Rot

Look at this bastard little icon. You probably see it every day right now. Hell, I see it every time I pop open the WordPress toolbar, because Automattic, makers of WordPress and Jetpack (back-end technology that helps WordPress sites do what they do) are cramming it in absolutely fucking everywhere, just like every other tech company is right now. No-one asked for this, no-one wants it, no-one is happy with the results it produces.

And yet, look at that bastard little icon. Such promise it carries in its little sparkly starbursts! The suggestion that magic is about to happen! The implication that, were you just to click that bastard little icon, creativity will be magically produced from nothing, allowing you to truly express yourself without any of that pesky "thinking"! You will truly be once and for all free!

As a creative type, naturally I object to generative AI being jammed in everywhere that it doesn't belong. I'll admit to having found some uses of it potentially interesting — music generation is intriguing, feeling like a step onwards from a program we used to have on the Atari ST called "Band In A Box" — but whatever use case I come across, it's hard to shake the feeling that its only real use is to enable laziness, and to prevent having to pay a real person for doing the creative work that is their specialism. (The actual computing and environmental cost of such tech doesn't matter to AI zealots, of course.)

That's not to say there's no money in AI, mind; no, by golly, the big tech companies are falling over themselves to hoover up investor cash right now, and every big generative AI site features some sort of predatory monetisation system, usually involving "credits" that obfuscate how much you're actually paying, and/or "monthly" subscriptions that are actually charged annually, because apparently that's just a thing you can lie about now and no-one calls you on it.

I think one of the clearest signals I've felt that AI bullshit has gone too far is its encroachment into pornography. It's now easier than ever to produce "deepfake" pornography featuring people who have not consented to appear in pornographic material. Of course, AI-generated slop has plenty of telltale signs, still, but the fact this stuff exists at all was already cause for concern even before it was easy to produce it.

On top of that, sites that were once about posting collections of erotic art and animations from artists, movies, anime series and video games are now overflowing with AI-generated swill; a cursory glance at e-hentai's front page earlier revealed a multitude of galleries tagged with "[AI Generated]", making them virtually worthless. Of course, e-hentai and sites like it already skirt the borders of morality by often including artwork artists intend to be kept behind Patreon, skeb or Fantia paywalls — but many of these galleries seem to suggest that there are a significant number of individuals out there attempting to position themselves as "artists" when all they are, in fact, doing is plugging prompts into an AI model that doesn't chastise them in a patronising way when requesting erotic material.

I'm sick of it. I'm sick of Jetpack emailing me to join an AI "webinar", I'm sick of ClickUp, the productivity tool we use at work, constantly spamming me about some AI feature I don't care about, I'm sick of the breathless zealotry from the cryptobros who have found the next big thing to latch onto before it all inevitably comes tumbling down in burning wreckage… and I'm sick of the uneasiness that I'm sure anyone in a vaguely creative field is feeling right now.

And I'm not sure it's going to go away for a while. Big Tech seems determined to make "AI" a thing. And while I'm not averse to actual, helpful uses of it — which I'm yet to see a convincingly working example of that can't be better fulfilled by other, existing methods — I think we all know that with the people we have in charge, those actual, helpful uses are inevitably going to take a back seat to ways of screwing poor old Joe Public and his friend Struggling Artist out of their hard-earned money more than anything else.

(Aside: I tried running this article through Jetpack's stupid "AI Assistant" to "get suggestions on how to enhance my post to better engage my audience", and the thing just crashed. Good show!)

So fuck that bastard little icon. Take your magic sparkles and jam them right up your robotic arse. The only things allowed to sparkle like that are fairies and ponies, and AI is neither of those things. So into the trash it goes, so far as I'm concerned.

If you want this nonsense in your inbox every day, please feel free to subscribe via email. Your email address won't be used for anything else.