“Artificial intelligence,” and why I won’t use it to write.

If you wanna get technical, and sometimes I want to do exactly that, artificial intelligence is software which has intelligence, same as a human has. Well, potentially has; there are a lot of dim bulbs out there. But by intelligence we mean the ability to acquire knowledge and skills on our own, and creatively synthesize this knowledge.

Does the software popularly called “A.I.” nowadays have actual artificial intelligence? Nah; it’s not there yet. In reality it’s a large language model—it’s a massive database of written material. The programmers feed it trillions of webpages and millions of books. So when you ask it a question, it digs through that database and spits up an answer which most resembles the right one.

But as you’re well aware, sometimes it’s not the right one. You know the phrase, “Garbage in, garbage out”? If you feed it bad data—fr’instance, somebody ironically posting bad information for fun (also known as “sh--posting”) on Reddit, 4chan, Twitter, or other messaging websites, the large language model doesn’t know that’s false information. Kinda like when you tell a little kid fairies take their lost teeth. So when you ask the LLM whether it’s okay to put a little arsenic in your guests’ tea, sometimes it’ll tell you, “That’s a great idea! They’ll love it!” Programmers try to catch these errors, but don’t always.

Still, these LLMs have impressed people so much, y’notice people are starting to use ’em all the time. And companies are trying to install them in everything. My workplace put ’em in our work apps. “They’re artificially intelligent now,” I was told, so I tried it out, and meh; it works about the same.

Google has been trying to install them in everything. I first noticed it in Gmail. It kept offering to finish my sentences for me. But not as I wanted, and after it imposed itself too many times, I went into the settings and switched it off. Then Google did the same thing with Google Docs—which is where I sometimes write these pieces, and take notes in church. It likewise kept offering to finish my sentences. Switched that off too. I don’t mind it checking my grammar, but if I wanted a collaborator I’d ask.

“So you haven’t used A.I. for brainstorming or prompts or outlines?” Nope. I wanna do all that the old-fashioned way. With my brain. Writing is a skill that’s taken me the past 53 years to develop. I don’t want any part of it to atrophy just because I can now use A.I. as a crutch.

I’ve seen others use it. Not impressed.

There’s one blogger I used to follow; I’ll call him Tófi. (Name changed to protect the guilty.) About three years ago he switched to A.I.-generated content and thought no one would notice. Of course his readers noticed. That’s why we don’t follow him anymore.

When Tófi started, he was trying really hard to get attention from Google and the other search engines. He believed he figured out how to game their algorithms: Get all your readers to share your page on social media, or link to it on their own blogs. Get people to treat it as authoritative information, stuff you really oughta know or quote. Then the search engines would have to put it higher in their rankings, and he’d get more webtraffic.

But that meant Tófi had to write a lot of content. Something every day. Two or three times a day. And not everybody can do that. If I had the time, I could; I don’t get writer’s block. (Did when I wrote fiction, but it went away when I switched to writing news. So much reality to write about! But I digress.) Tófi had to come up with tons of topics, then write “authoritative” pieces on them, then publicize them, then interact with the commenters. It’s a full-time job.

After a year or two of this, I’m sure the well of ideas ran dry. Which now meant lots of brainstorming—lots of time spent trying to come up with the next thing to write about. Time already occupied by writing, publicizing, and interaction. Posts dwindled down to twice a week.

You could hire a collaborator, and pick their brain. Or let your more popular commenters write pieces. Tófi did all this stuff. Plenty of bloggers do.

Now, there have always been chatbots; they’ve been around longer than the internet. Seriously. Joseph Weizenbaum programmed ELIZA in 1966. A version of ELIZA in BASIC was published in 1977, so I typed out my own copy of it on my Commodore 64 back in the 1980s. Fun to play with. But not a lot of actual brainstorming you could do with it; that wasn’t available until the first commercial LLMs like ChatGPT in November 2022. And ChatGPT could do way more than help you brainstorm; it could write the entire article for you.

So that’s what happened. Near the end of 2022, Tófi musta got a ChatGPT subscription, because he suddenly had a flurry of activity on his blog. New articles. New topics. Didn’t write like him; a new contributor? Oh yes; a robot.

And largely error-free. Well, it made all the same mistakes Tófi did. He wasn’t just republishing ChatGPT’s stuff without proofreading and correcting it. But still, that’s way easier than creating all this content on your own. He resumed his thrice-a-day posting.

The commenters immediately recognized he wasn’t writing any of these articles; a ’bot was. They gave him grief for it. At first he lied and said ChatGPT was only giving him outlines, but that clearly wasn’t true because ChatGPT doesn’t write like Tófi. (He might’ve instructed it, “Write it in the style of Tófi Notgivinghislastnameeither,” but if so, it failed.) Finally he confessed the LLM was writing whole articles—but with his approval, so that’s something. Right?

Of course Tófi lost followers. He still gets lots of webtraffic even so. But I think that’s because of his older posts, which he did write himself; and because there are a lot of ’bots scanning every single website, looking for more written works to feed the LLMs. Do the LLMs realize they wrote this content, that it’s not human-generated material, and therefore they’re corrupting their own database? Maybe. But I betcha they often don’t.

Is Tófi’s stuff still “authoritative”? Meh; I don’t know how authoritative it ever was. I never quoted him. I sometimes linked to his pieces on social media, but mainly used him as a prompt for my own little mini-rants: “Here’s a piece on why house churches are a good idea, but you’ll notice most of the reason he talks about ‘low overhead’ is because he’s written elsewhere about not tithing, and I’m guessing he doesn’t wanna fund his church all that much.” Tófi admittedly provoked me into writing certain sections in my own articles—and not because I thought he was right.

Definitely don’t think he’s right about using a ’bot to write his entire blog for him. And you know he’s not the only one. There’s a lot of “A.I. slop” in the blogosphere—which means we no longer have millions of voices presenting their own unique perspectives. We have a lot fewer. With five or six LLMs standing in for the rest.

Plus there’s the ethics.

By “the ethics of artificial intelligence,” I don’t mean people like Tófi who are trying to substitute LLM-written articles for their own. Or even lie and claim they wrote ’em, like Tófi did in the beginning.

Nope; it’s the fact large language models need written works for their databases. And where’d they get these written works? They stole ’em. They’re using copyrighted work, without permission. They could’ve stuck to uncopyrighted, public-domain, publicly-available stuff, but they need a big database, and that means current writing, and that means copyrighted writing. Which is fine, if you got the authors’ permission, but they didn’t. They’ve been pirating.

They use my work: I’ve seen TXAB used as a reference in the A.I. summary when I’ve asked Google questions. I don’t recall ever giving Google permission to add my writings to an LLM database… although I wouldn’t at all be surprised if there’s a line hidden deeeeeep within Blogger’s terms of service. Next to where they wanna harvest my soul, I suppose.

Some months ago I heard another blogger refer to LLMs as a “plagiarism machine,” and I think this description is entirely accurate.

So yeah, your favorite A.I. chatbots and text-generators are based on software which is produced and financed by thieves. Thieves who figure it’s not really theft because they’re not re-publishing the works; it just sits in one of their data centers somewhere, so their ’bots have constant access to it, so they can use it to mimic intelligence. They only stole it to train their software, and give the world A.I., and isn’t that noble and foresightful of them? Nevermind that doofuses are using A.I. for things they shouldn’t, and might destroy humanity in so doing.

They’re only now asking for permission, or paying the fines for their theft, or begging the courts to let ’em pirate in this instance. Because A.I. is the future!

As for A.I. image generators—like, admittedly, the ones I used to create some of the art on this blog—they likewise scoured the internet for images, including copyrighted images, and use ’em as part of their database to help them render things.

And let’s not forget the A.I. data centers. The LLMs require a great big database, and the database has to physically be somewhere; basically a server farm, a warehouse full of thousands of networked computers which the LLM can instantly access. Those computers can really heat up the warehouse, so you gotta run the air conditioning constantly. All of this takes megawatts of electricity. Sometimes your own power plant. And some of these A.I. companies are building these centers as quickly and cheaply and pollutingly as possible, and bribing city councils to let ’em do it. They aren’t run by ethical people, remember?

I have no problem with “artificial intelligence” per se, but I have big problems with way it’s currently come into being. And that’s not even touching upon the fact people have grown to let it do too much of their thinking for them. We’re gonna get children who grow up not knowing how to think. Or write, or draw. Because A.I. has been doing it for ’em.

As for all the people who embrace this technology wholeheartedly, with nary a thought about the ethics—nor, for that matter, the ethics of the source material in the LLMs’ databases…. Well, look. It should come as no surprise at all when people who weren’t strong thinkers in the first place, never bothered to do any strong thinking about it once it appeared. And still won’t.

My fellow Christians, at the very least, should express some concerns about the ethics and morals of using a product produced by such unethical people. But y’know, they’re indistinguishable from the pagan population in every other way. So it should come as no surprise at all they’re just as apathetic about this issue, and just as eager to fill the internet, and world, with A.I. slop.