The Great UnGoogling Side Story: The Optimal Death of Prime

https://www.peppercarrot.com/en/

I've been invested in the Amazon ecosystem for many years. You might even say I was an early adopter, especially when they expanded their offerings from books to DVDs. It was simply the best place to get movies if you had particularly obscure tastes like I do. I mean, if Suncoast didn't have it, it might as well have never existed.

Hopefully, you don't need me to tell you about everything wrong with Amazon. If you do, let's just say their convenience comes at a steep price, and I don't just mean the flow of money. 

Speaking of money, I had a Prime membership for years. It seemed like a no-brainer; 2-day shipping, Prime Video, Music... You were paying for the streaming service but with a boatload of perks. 

As time has gone on, however, the perks have not increased in value. In fact, many features have ducked behind paywalls and has left my membership feeling rather anemic. 

First, there was Amazon Music. Instead of giving you unlimited plays of certain songs, everything went to a Spotify-like shuffle mode. The entire catalog was available, but you'd have to wait for it to randomly show up. The biggest problem with this was that it affected any singles or albums that you purchased. This was resolved eventually, but that it happened in the first place showed how much of a rushed solution this was to a cashflow problem. 

After that, they started putting ads on Prime Video. Here I was paying about 140 dollars a year, and they were still going to have me sit through ads while watching movies and shows. Rather than make a new, ad-supported tier to draw in new users (like HBOMax and Netflix), they changed the tier I was already in to the ad-supported one. Simply put, this is not what I signed up for, and I wasn't alone in feeling this way. Ironically, I wouldn't have minded if they simply upped the price of Prime as they've done in the past. I pay it annually because that is much more convenient than juggling monthly charges. However, getting rid of the ads on Prime Video requires a monthly charge of 3 dollars. There's no option for me to pay more annually. Bear in mind, I also pay for Kindle Unlimited and Audible, so this was turning into a mess. I can understand making those a separate charge, but now they're going to nickel and dime me on a service I already paid for in advance. 

Another service I'd paid for annually was Disney+. However, after that business with Jimmy Kimmel, an increase in their annual price, and now their current deal with OpenAI to flood Plus with slop, not only did I not renew my Disney+ subscription, but I'm not giving them any money for the next year at least, including going to their movies. I'm not paying for slop. If they want in bed with OpenAI, they can have each other.

Speaking of AI, Amazon made their own AI deal that served as the straw that broke the camel's back for me. I have an array of smart speakers around the house. They're very handy. I use them for kitchen timers while cooking, controlling lights, making grocery lists, playing music (if I don't mind the shuffle mode), and sometimes as an intercom with my roommate, among other things. Again, very handy. Alexa Plus, meanwhile, is only handy in the sense I want to put my hands around its neck and squeeze its damned life out of it. At first, it was offered as a beta program and an update to the Alexa app. In addition to the hassle of dealing with how everything was rearranged in the app, using the new Alexa Plus was equally tedious and even counterproductive. Without getting into the weeds of how virtual assistants work, they replaced Alexa's hard-coded routines with a chatbot that turns nearly everything you ask it to do into a big production. I almost wouldn't mind it except everything took slightly longer than it normally did. I swear the first time I tried adding something to the grocery list in the new app with Alexa Plus, it took about five minutes. Full disclosure, I've used AI chatbots before, and I hate them. I hate everything about them. It gets under my skin because chatbots are essentially what customer service jobs want you to sound like in a call center scenario. It's that mode of speech which tries to sound casual and friendly but is so forced as to be legitimately insulting. There's an uncanny valley aspect to it as well; it's trying to sound so human that it backfires. 

I tried to back out of the Beta program, and was successful, but only until the app update forced Alexa Plus onto me. If I kept my Prime membership, I was getting Alexa Plus whether I wanted it or not. 

This led me to look into my Prime membership and see just what exactly I was (and wasn't) getting for my money. One of my favorite aspects of Amazon is something called Subscribe and Save. At the risk of sounding like a sales pitch, you lump so many items together in a single shipment and set them to be dropped off at regular intervals, and this not only comes with free shipping, but discounts on the goods in question compared to one-time purchases. This is how I get my cat food, kitty litter, paper towels, bath tissue, and even my protein bars. Having boxes of paper towels and toilet paper dropped off on my front step instead of dealing with it during a Target trip is so convenient. It's pretty much like having a bulk store membership only you get everything delivered. I'm even using it for my roommate's colostomy supplies. I was convinced that all this was part of the Prime membership; it seemed like the next logical step from the 2-day free shipping that Prime offers. Looking into it, though, it turns out this is not tied to Prime. It's a completely separate service that anyone can sign up for. 

That was my a-ha moment. The only thing that my Prime membership was offering besides 2-day shipping on some items with no minimum total in the cart was Prime Video, and that was really not worth having the ads on it for what I was paying. If I really want to see something, it's easy enough to just wait for the whole series to drop, pay for a month (maybe even shell out the extra 2 or 3 dollars to watch without ads), and then binge the whole thing before the time goes up. My roommate's a little disappointed as she's a big Critical Role fan. Even though this membership won't lapse until June, that means future seasons of both Vox Machina and Mighty Nein will have to be consumed via the binge and run model.

I wonder how long before Amazon enacts an "x-months minimum" stipulation to their Prime Video memberships. 

The Correct Resolution of Paper

I’m not one for New Year’s Resolutions, typically; I prefer CGPGrey’s idea of having themes. However, in 2025, almost as an afterthought, I vowed to go paperless for the year. This was partly spurred by having gotten some Mobiscribe e-ink tablets, one to use at work instead of notepads or composition books and one at home for reading (and a color one because it was a good deal and I was curious about the tech). I was planning on using them for Inktober, but decided against it as getting notes and drawings off of the devices is more than a little tedious for something I’m supposed to do daily 31 times in a row. In the end, with only about a half-dozen Post-It notes at work as the exception, I stuck to the resolution well beyond the 365 days. 

Reflecting on this, I had another thought of something I could do as a resolution for the year 2026. It occurred to me while playing a typing game called Glyphica. In the game, you occupy a central turret and fire at invaders coming at you from all sides. You select your target by typing the word that appears above them. The game is a lot of fun; reminds me of the quality time I spent with Typing of the Dead on Dreamcast, an  edutainment title that doubled my typing speed in the course of a few weeks. Though the games are very different in terms of presentation, the biggest difference is one of quality of life. In the decades between the two titles, we’ve gone from spellcheck being a luxury to the software going the extra step of not only correcting your errors without you asking, but even predicting what it is you’re trying to say in the first place. If I start typing a word like constitution, I only need to get about as far as “const” before the autocomplete shows me the rest of the word, at which point I need only press the tab key to finish what I started and move on to the next word. A hundred keystrokes can drop down to as low as sixty-five in this way. So, what, you may ask, is the problem? 

Between Glphyica and an online typing speed test called Monkeytype, it's occurred to me that I've amassed a sizable number of bad typing habits because autocorrect swoops in and fixes them, sometimes before I even notice. Don’t get me wrong, it’s convenient when it’s not trying to forcefully make me say an incorrect word, but I feel like this is offloading vital aspects of a skill I happen to be very proud of. Between that and autocomplete, the software is doing too much heavy lifting for my liking. 

My worst habit is something I like to call chording. It happens with especially short words such as "the", "to", and... well, "and." The problem is I hit all the keys at once and I end up with "teh", "ot", and "adn." I can't help but wonder if stenographers have this problem outside of the courtroom. Although chording has technically been with me since I first learned to type, autocorrect and autocomplete have made it substantially worse. Rather than typewriter keys jamming or the timing of my keystrokes being measured in imperceptible milliseconds, the software is able to work out that I'm writing "the" and not "teh" as "teh" isn't a word as far as I know, at least not one in English. 

If you've ever seen that brain teaser where the words in a paragraph all have their respective letters out of order (yet it's still readable because your brain unscrambles it from context clues), that's more or less what my typing feels like to me at this point. It's like I know words as clusters of letters rather than sequences of them. My muscle memory has contracted to the point where it's folded in on itself. My speed has been relatively consistent, but my accuracy has taken a massive hit, and that's no good to anybody ecepxt vrey wreid poelpe who lkie tshoe arfoneemtoiend barin tseaers. 

This has led me to the decision to disable autocorrect and autocomplete on my Mac. Spellcheck gets to stay, inadvertent brain teaser construction be damned. A red line appearing under a word I just typed doesn't bother me. That's instant feedback on a mistake I made and it's on me to fix it. It's gently saving me time, not doing the work for me. It's what The Oatmeal would call an administrative task and not a creative one. When I give what I've written a once-over, it's highlighted the areas that need my attention first. It's working ex post facto rather than trying to get ahead of me like far too many "smart features." Those assume what you're going to do next. Sometimes they're right, but other times I want to write construction or constriction or consternation or even constituent rather than constitution. 

As for my iPhone, predictive text (which is essentially autocomplete) remains enabled on my keyboard of choice, Microslop's SwiftKey, the only Microslop product I use willingly and with any consistency. The reason for keeping this feature enabled is simple. It's a small and cramped keyboard and I need all the help I can get. It's like Lewis Carroll's Nyctograph, a specific tool for a specific job. In his case, it was writing in the dark and without getting out of bed. In my case, it's when I need to write something and either can't get to one of my full-sized mechanical keyboards when I want to or, like Mr. Carroll, I can't be bothered to get out of bed or off the couch at the moment. 

The jury had been out on the iPad since it's a notable difference in screen real estate, but I don't always have one of my Bluetooth keyboards along with me and the on-screen keyboard still has all the same problems as the iPhone. 

As for the paper resolution, that has stuck around as far as taking notes at work goes, though Post-It notes will no longer give me pangs of guilt on the rare occasions I have to use them. For everything else, I do want to start doing more traditional art instead of my usual vector works. I'm even going to try using a fountain pen after either losing or breaking the first one I tried. I don't remember which fate befell it, which is why I'm happy to have discovered that disposable fountain pens exist. In fact, some artists recommend them over their more expensive brethren. 

Welcome to 2026, everybody. 

Bleep You, Got Mine (and I'm sorry)

Probably the biggest mistake you can make when it comes to building a PC is believing it will save you money. Of course, there’s the obvious time sink of researching just what it is you’ll be doing in the first place, learning why some parts will or will not go together, sourcing all the parts in the first place, and the omnipresent possibility of something going horribly, horribly wrong and there being no recourse beyond opening your wallet again. What you might save in money, you’ll lose in time… and possibly a little sleep. 

I’ve built PCs, and I don’t miss it. At this point in my life, I’d rather pay for the peace of mind via professional assemblies than the satisfaction of building something with my own two hands (and having to be my own technical support). It was certainly a learning experience, and I’ll always be grateful for that. In terms of cost, I didn’t really pay as much attention as I should have, but I know at least once or twice I had to basically start over and source a new part because something was amiss, including swapping out an entire motherboard because I plugged something in wrong. To be fair, when I was assembling PCs, it was circa 2010-2015, which meant I had it easy. There’s no shortage of tutorials on PC assembly, and the way PCs are built in this era to begin with is a much different animal than it would have been 20 or even 10 years ago. To put it in perspective, I didn’t learn to solder until 2016 for my job. None of the PCs I built required any soldering, but this was most certainly not the case 20 or 30 years earlier. 

In food terms, you’re assembling a sandwich. The bread is the case, the lettuce is the motherboard, the pickles are the RAM, the meat is the CPU, the mayo is the thermal paste… it’s not a perfect metaphor, but you get the idea. You’ve got individual components that just snap together to eventually form a fully working personal computer workstation. In other words, in terms of time, I actually had it pretty easy. 

Now, about those pickles…

Without getting into a long-winded rundown of just what RAM (Random Access Memory) is, think of it this way: The hard drive is your long term memory. It’s your name, your family’s faces, the streets you grew up on, and anything else you don’t want to forget. The more of it you have, the more you can remember without having to forget something first. RAM is your short term memory plus your ability to multitask. The more of it you have, the more you can do at once. It’s also probably the simplest and most straightforward upgrade you can perform on your computer. I like to tell people that if you can change a diaper, you are overqualified to swap out the RAM in a computer. Even if you need someone to talk you through it the first time, after your first go, you’ll be showing it off at parties.

In my personal experience, RAM has always been very costly. I’ve easily spent more on RAM than I have on motherboards (spontaneous failures notwithstanding). There have been many reasons for this. Games tend to be a bit RAM hungry, especially considering online games in which you’re often running at least two or three other applications at the same time, one of which is doubtless a web browser with two-dozen tabs open. During the pandemic, when we were all stuck at home and relying on our computers to check in at work, start that live-streaming channel, or virtually send our kids to school, RAM prices went up. That old desktop you bought a few years earlier wasn’t going to cut it, but you couldn’t afford to replace the whole thing, so you swapped out a few parts. 

I’m sorry to tell you, but you were over a barrel. We all were. 

There was a demand, so the cost of the supply went up. It’s simple economics. Ironically, something that should have been the cheapest upgrade (sticks of RAM don’t nearly have nearly as much meat on their bones as a motherboard, a CPU, or a hard drive) became a hot commodity. 

One company has understood this better than anybody: Apple. Actually, that’s a slight lie; Apple understands the economics, but ultimately plays by its own rules. As a rule, once you’ve purchased an Apple product like a MacBook or an iMac, you cannot swap out or upgrade the RAM, no matter how many diapers you’ve changed nor how much sleep you're willing to lose. When you make that purchase, you’d better be thinking of how it will impact the next seven generations. Otherwise, you’re going to be left wanting down the road. So, do you splurge now and hopefully be content for the next few years, or do you settle and upgrade more often? Apple technically wins in either case. 

I talked before about the pandemic driving up the price of RAM. Well, now the newest plague to befall mankind is artificial intelligence. 

Without getting into a long-winded rundown of Generative AI and Large Language Models, the important thing to understand is that these chatbots and image generators need a lot of processing power and a lot of short term memory. The tech companies behind these AI tools are building more and larger data centers all over the country and even around the world. The demand for RAM is so out of control that Micron, one of the largest manufacturers of RAM decided to no longer sell to consumers and focus on its corporate clients, the ones building the datacenters.

What this means for consumers is that if you thought the RAM prices during the pandemic were bad (and possibly before that) then you haven’t seen anything yet, and the only thing that’s going to stop it is this whole AI bubble finally collapsing on itself because no amount of Sam Altman saying, “Trust me, Bro!” Is going to make this so-called business model sustainable.

This is the part where I humbly brag. As an actor friend of mine once said, “Save your rotten fruit for the parking lot. I’ll have more places to hide.” 

Back in 2020, shortly after I bought my house at the start of the lockdowns, I purchased a Mac mini as a housewarming gift to myself. I’d wanted one since they debuted back around 2005, but the planets never quite aligned just right for me to make the decision. 15 years later, the alignment netted me a 2018 Intel-based model with 8 Gigabytes of RAM. I could have sworn I opted for 16, but I think I talked myself down since I was going to use it mostly for writing, drawing, and at most some 3D modeling in SketchUp. Anyway, despite the low hardware specs, the little gray box I nicknamed Gray Rock served me very well for the next five years. Even after Apple changed their processors from Intel to a homegrown lineup known as Apple Silicon, the little Gray Rock that could was holding up just fine and dandy. 

I knew I couldn’t keep this up forever, though. While the hardware would probably last many more years, software is another problem. With the change in processors, applications were leaving the old processing architecture behind and were being optimized for the new kid on the block. This sort of hardware upgrade/software optimization leap frog happens regardless of paradigm shifts in processor types, but Apple’s new in-house strategy lit a fire under developers to get with the times. 

There was, of course, a new version of the Mac mini, but I have to say I wasn’t too impressed with it. I mean, sure, it’s nice and compact, but it presented a problem for me. While I like Apple’s hardware offerings such as the Mac mini and the iPads, I’m less keen on Apple’s accessories. I don’t like their keyboards, I don’t like their mice, and while their monitors are nice, you can do a lot better for a lot less. As for the mice and keyboards, Apple cares more about form than function in this regard, and that means favoring wireless over having cables crisscrossing what’s supposed to be a sleek, minimalist setup. In other words, my mouse and keyboard needed to be plugged in, and the new Mac mini would have required me to use an adapter. This is a pain, so I searched other options. I guess I’d simply turned into too much of a Prosumer for the Mac mini’s casual demographic. This led me to the Mac Studio, a higher-end desktop with a similar-ish form factor to Gray Rock, the notable difference being the Studio’s doubled height to accommodate a massive heatsink and cooling fan. Needless to say, it had a lot more horses under the bonnet than my model that rolled off the assembly line 7 years ago. It also had more RAM out of the gate, with the minimum being 36 gigabytes. This was starting to look like the perfect upgrade; it had over 4 times the RAM, a new processor, and I could plug in all my favorite accessories without some clumsy adapter. Throw in a special discount that doubled the hard drive space for the same price, and I took it as a sign.

In the first week of August of this year, my new Mac Studio arrived and Gray Rock was shipped back to be recycled. In its honor, the new Studio was given a similar nickname Grimlock (after a Transformer), and it’s been a great upgrade for only taking up so much more room than its predecessor. I may not be pushing its specs with my typical workload, but I’m going for a slow burn rather than anything fast and furious. Even when I’m using a 3D modeling program like Blender, I’m only really using it to make perspective and shadow reference models for drawing. 

I’ve been thinking about the timing of my purchase in relation to the spikes in RAM prices. I was under the impression, since Apple’s processors were being made in-house and their RAM works a little differently from the Intel-based models, that Apple was safe from the increased demand. After all, they already charge a premium for RAM when you’re configuring your initial purchase, so I was kind of ahead of the game in that regard. At least, I thought that was the case. Turns out Apple does still rely on these RAM manufacturers for their own machines, and it’s only a matter of time before the price spikes affect Apple’s own price tags (which already have a reputation). 

For whatever it’s worth to you out there thinking of buying a new PC or taking on the risk of building one, I absolutely hate that this is happening. There is no smug look on my face as I sit at my Studio typing this out. I hate the reason for it happening most of all. I hate this far off pipe dream of promise of some kind of computer generated hive mind and all we need for it is to build more power- and water-hungry data centers and for you to keep asking Gemini to make you images of Mickey Mouse cleaning an assault rifle while writing your college term papers for you. 

Apple Intelligence is available on both Grimlock and Sapphire (my iPad). It has not been enabled. I have no desire to enable it. Few things in this universe would please me more than for the parts of their processors dedicated to AI features to be used for something else.

My Slop Could Beat Your Slop

Photo by Fruggo

Let me tell you about someone on Quora, someone we’re going to call J. J had requested my answer to a question regarding YouTube videos. Here is the question verbatim: 

My YouTube channel talks about self development, I currently use stock videos from Vecteezy (I give attribution as instructed), motion graphics and Ai voice over narration to make videos. Will my channel get monetized?

I see questions like this all the time. They’re all worded slightly different, and they don't all involve using AI, but my brain hears it the same way every time:  I want to participate in the Boston Marathon, but I’m really, really slow. If I show up on a dirt bike, will I be allowed to race? 

We can probably have a very deep and thoughtful conversation about the future of AI and how it could potentially be used as a productive tool that aids people in their chosen endeavor. I don’t doubt that. We could probably also have a similar discussion about steroids, albeit the public attitude about those seems pretty clear. Remember when we stopped calling them steroids and simply referred to them by the blanket term Performance Enhancing Drugs? That wasn’t to broaden the definition to include other drugs so much as it was a way for those using said drugs to not sound like they were taking the easy way out. After all, it’s only ENHANCING their performance. They’re still working out and training, they just need that little extra edge because they’ve plateaued in their routine. Is that really so bad? 

Of course, doubtless at least one of you has raised a hand in objection and pointed out that content creation for social media platforms is not a competition like it is with athletics. To that I can only say, “Fair, but when monetization is involved and stated as a goal, you’ve made it into one.” We can’t all be Jimmy Donaldson any more than we can all touch the FIFA trophy. Even if we take monetization out of the equation, you’re trying to gain an audience, and that audience only has so much time in the day to consume content. As a wise man said, time is money. It’s even called the attention economy. 

Before I could answer J’s question, I needed a little context, just to see if I was possibly missing something fundamental. I asked why he couldn’t narrate the videos himself. Maybe there’s a good reason. I mean, I don’t like the sound of my voice, so who am I to judge? Maybe he doesn’t feel it would be a good fit for the subject matter. Maybe he’s got a really thick accent and is difficult to understand. 

J answered in two separate replies, the first being, 

“But with the ai voice over is it monetizable?”

J, I asked you why you couldn’t do the narration yourself so I could understand your circumstances that are leading you to ask about the AI voiceover. I asked as a comment on your question so I’d have more information upon which to base my answer. Repeating the question to me isn’t very helpful. The second was, 

“Usually my voice over produces unclear audio”

This doesn’t really answer the question, either. “Unclear” isn’t terribly specific. In hopes of coaxing a little more detail out of him, I offered the following advice, “That’s an easy fix. Even voice notes on an iPhone can produce clear audio. If your emphasis is on self-development, you need to demonstrate that you’re developed enough to share your message more directly rather than hiding behind a machine voice. It’s all about authenticity. Visuals are one thing, but audio is what can really make or break a video.” There was no response from J to this. What’s “unclear” remains unclear. 

Going back to the response about monetization, this was when I decided to check out J’s profile. There was only this one question on his profile, and he had given only one answer to another question. 

Here is that other question verbatim: 

If I use an AI generated image in my video and add voiceover to the video and upload it on my YouTube channel, will it get monetized?

Here is J’s answer to that question: 

“It is best if you go through YouTube's monetization policy.
From your question, your videos might fall under -LOW EFFORT”

So, for those playing at home, we’ve got one content creator that is using stock videos and wants to use an AI for narration, and another content creator that is using AI generated images and a potentially non-AI voiceover (that’s important). The daylight appears to be measurable in seconds, doesn’t it? Curious if J has actually gone through YouTube’s monetization policy to know that this particular combination of sound and vision is ineligible. 

When I brought this up to J, this was his response, 

“Yes but there was a significant difference in our content type
That person said they wanted to use still images+ai voiceover only in their videos
But my videos use videoclips, edits and motion graphics+ai voiceover
Our content type is totally different”

Actually, J, that person didn’t say their narration would be rendered by AI. They said they’d “add voiceover to the video” after mentioning using AI-generated images. You made an assumption and tried to insist that using stock assets was more effort-intensive than using AI-generated ones, which is a healthy enough discussion we could have. After all, you’re both using something you didn’t personally create. Someone else did the work and offered it willingly to be used for other people’s videos. The AI-generated assets are a product of data scraping the work of others, regardless of their choice in the matter, but those results are also tailored to a specific input prompt. We could split hairs over who’s putting more effort into the visual portion of their videos until doomsday, but it’s certainly fair to say they’re both low effort compared to people who produce their own visual content, from the humble vlog to the elaborate and collaborative animated story time video. 

I should point out that there are many content creators who integrate stock assets into their videos along with their own video and audio content. The important distinction to make here is that the stock footage is not being used as a crutch, much less a foundation. It is supplemental to the original portions of the content. The same goes for something like music from YouTube’s audio library or other stock music resources. These are parts of larger works and their contributions are ultimately secondary to what the content creator brings to the table. 

The problem with what you’re doing, J, is that you want the backup band to be more than backup. You’re trying to pile up enough supplemental material that there’s no longer any primary content from you beyond possibly the barest bones of a script and overall vision. Given that, this is why I point out there’s barely any daylight between what you’re trying to do and what you called out that other content creator for asking. 

The point I’ve been trying to make to you is that you need to put more of YOU in what YOU are producing for YOUTube. It’s all about authenticity. The reason you’ll hear so many people complain about AI Slop content is that it’s all so impersonal and lacking in heart. It’s designed to chase a trend and feed an ever-changing algorithm, not actually appeal to anyone. It’s junk food, and it’s not even good junk food. The flavor’s gone in an instant and if the calories were any emptier, they’d collapse in on themselves and form little black holes. If that’s the best you can bring to the table, then all you’re doing is getting yourself lost in the noise. Why should anyone give your work attention if you’re not going to give it your own attention and leave a machine to do nearly all of your heavy lifting. 

My advice to you is that if you can’t take that step to make your content more personal, then don’t make your content. If you can’t be yourself, why should anyone care about you? 

Boof Jark: My AI Pledge

A wise man once said that one man’s software is another man’s AI. As such, it’s important we establish some kind of clear, working definition of what’s being referred to when it comes to AI-generated content. For example, the ghosts in Pac-Man are AI. That’s not what we’re talking about. Spellcheck is arguably a kind of AI, but it’s not what we’re talking about, and I’ll even give a pass to predictive text for reasons that will hopefully become clear as we progress. Case in point, I only had to write “prog” and follow it up with the tab key to write the rest of “progress”. Another writing tool I feel I should disclose use of is an oldie called Grammatik. It launched around 1981 by Aspen Software before eventually being absorbed into Corel's WordPerfect, which was how I was introduced to it circa mid 1990’s. Bear in mind, this was not a Large Language Model trained on all the text in the world. I have absolutely no idea how it worked, but it’s not important because outside of obvious things like run-ons caused by a misplaced period, I never relied on it more than I trusted my own two eyes giving my paper a once-over (and that was only the first line of defense). It was also oddly fixated on the idea of passive voice, a grammatical concept that I think I understand and I know I’ve had explained to me countless times, but I couldn’t spot it in the wild or stop myself from writing it if my life depended on it. If anyone knows a really good mnemonic for it, I’d love to hear it. It’s also worth noting that Grammatik couldn’t really learn as you wrote. At best, you could tell it what to ignore, but it wouldn’t remember what you told it to ignore the next time you reopened the document, so it was only useful if you wrote in crazy, uninterrupted marathon sessions a la Jack Kerouac. Most importantly, Grammatik couldn’t offer suggestions for how to revise something beyond “You’ve started every sentence in this paragraph with the same word. Avoid repetition.”, it certainly couldn’t offer suggestions about tone, and it definitely could not fabricate a full essay on my behalf from a briefly descriptive prompt. I was still doing the work, and that’s the key distinction to keep in mind. 

To put it another way: Who is driving the car? The one in the driver’s seat with the wheel in hand, or the one in the passenger seat giving directions? The officer who pulled you over already knows, but he’d really like to hear it from you. 

I don’t know if it was on Medium or Substack where I saw about a half-dozen AI generated images in a row while scrolling and saw red for a moment before closing the tab, toppling a chair, and storming out of the room grumbling about the decline of Western Civilization. That’s a slight exaggeration; I would never do that to a perfectly innocent chair. Besides, the last thing you want is to make a chair mad at you. They won’t stand for it. 

What upsets me most about this trend is the fact that these writers are often the first, loudest, and yet most eloquent at bemoaning the fates and predicaments of artists, creatives, and professionals in general in the face of this ever-changing landscape that AI is slowly consuming. It makes me want to grab them by the shoulders, shake them like they’re the first character to go hysterical in a disaster movie, and scream, “YOU ARE NOT HELPING!” and finish off with a slap across the cheek. For the record, I have the kind of upper body strength that makes Mr. Burns look like Sugar Ray Leonard at the height of his career, so if you’re afraid of this potentially becoming reality for you, rest assured you’ll barely feel a stiff breeze. I’ll also likely collapse from the exertion while breathily asking if you’ve had enough, so the only pain you’re guaranteed to feel in the whole ordeal is the pain in your sides from laughing.

I made a note on Substack (their version of a Tweet) to essentially complain and vent, vowing that if I saw an AI header image to an article, I wasn’t going to read it. I wasn’t addressing anyone in particular; it was not about naming and shaming. As you may have heard from your parents at one point while disciplining you, I’m not mad at you, I’m mad at what happened. Moreover, I'm saying this because I want you to do better, and I know you’re more than capable of it, whatever your self-doubt is saying about you right now. 

As for doing better, these are your options.

1. Learn to draw and/or take photos. 
2. Use a stock image service; you’ll be paying real artists or at least using works volunteered by willing artists. 
3. Commission a human artist directly. 

This is the part where I acknowledge that I am fully aware of the potential pitfall of my pledge in that I may well mistake something genuine for something AI-generated, therefore missing out on something authentic, inspiring, compelling, and any other praise I could throw at it. That said, let’s be real, a lot of what we collectively refer to as AI Slop is pretty samey. I think the best description I ever heard for it is, “Well-rendered but badly-drawn” with tells such as heavily-detailed shading and color gradients while the proper lacing of shoes and/or number of fingers on a hand is spotty at best. Basically, a painterly once-over of a hasty sketch. Even more basically, polished shit. There are other descriptors, but these are the most polite. 

“Well, those AI generators can make some really photorealistic results that are harder to detect, especially if you know how to—“ 

If you’re willing to put that much effort into hiding your use of AI, you have even less of an excuse to not employ any of the alternatives I’ve listed. 

Before I go, I’ll reward you for getting this far with an explanation of just what in Hel’s Realm Boof Jark is. I don’t know where it was posted, maybe Instagram or LinkedIn, but it was an image of two boats on the shore of a deserted island, one filled with fruits and vegetables (along with eggs and a rooster because MidJourney doesn't know how poultry reproduction works), the other filled with all manner of meats, including a package of Boof Jark. I mean, you seriously couldn’t be bothered to take the most basic pen tool and write some block letters on a white box? 

Speaking of handwriting, which I’m notably not using for this, I’ll close with a story about Bob Ross. He once received a letter from a fan saddened that he couldn’t follow along with the painting sessions because his hands shook too much. Bob Ross, being the absolute legend that is Bob Ross, assured this person, “Well, then you can draw trees better than I can, because trees are rarely straight.”

Behind the Scenes