On AI + Writing

Disgruntled thoughts.

One by one I hear from a friend or family member about their use of AI. Some mention it so casually as if everyone is doing it (I’m learning, they probably are). Others have mentioned it with caveats, knowing the implications for announcing that they are using it in whatever capacity they have deemed necessary. It’s always the people who say they are using it when they also know of the real concerns surrounding its environmental impact that make me tilt my head a bit. The fact that AI has been foisted upon us in so many realms of everyday life so quickly is disorienting. In some ways it makes it difficult to avoid because many programs we are used to using are now giving many pop ups to attempt to force you to use it—I’m looking at you Adobe PDF.

Based on the content I typically see and the chatter I have with online acquaintances and IRL friends, I would like to say we’re reaching some kind of point where things might be breaking a bit with AI but in reality I know we’re nowhere near that. What I see is part of a bubble, after all. However, there have been several big stories out in the last week to month regarding AI that I think will start pushing some people to pause and reflect on the content they are consuming that is obviously created with AI in some capacity. The biggest story in the last few days is that of Hank Green announcing how much he has relied on AI to create content for his YouTube channels, particularly since his cancer diagnosis, treatment, and recovery a few years ago. I think some people had suspected his use of it long before now but others are dismayed at the depth his reliance on it has reached. And that’s not even talking about the realities of AI psychosis, a phenomenon that I believe we will only see become more prevalent in the next few years. In Hank’s case, the easy answer to avoiding AI here would have been to pull back on his end of creating content while in cancer treatment and recovery, to take time off and rest, instead of continuing to feed the round-the-clock content machine in an effort to keep capitalism going.

In the publishing realm, Brooke Warner writes about the publishing industry’s lack of AI guidelines or cohesion about the use of any kind of AI content and it is causing all kind of issues for writers. And even Substack has rolled out a feature, that writers here can disable, that allows people to check for AI content. Regarding the latter, I have noticed or suspected AI use on many of the articles that I end up seeing in my “For You” section on Notes. There’s a rhythm to the writing and they all eerily have the same voice. That, and how many times a week can you publish multiple, thousand-plus word essays? I remember in the blogging heyday where some food and craft writers would publish multiple times a day but that was a typical microblog + photo post, not longform content. And I’m over here beating myself up for not having the time or capacity to get to all of the essays I need/want to write that I tell you readers that I will “write next week!”. No, these people pumping out essay after essay are most certainly using a LLM to get the work done in some manner and there is no transparency about it.

Then there are the AI detection programs. The problem with those programs, as Warner writes about in the link above, is how accurate are they? How many false positives are they throwing? Are we about to enter a realm entirely of AI made content, with or without real human creators behind them? I’ve already seen people positing that we’ll have a whole sector of non-AI derived content, whether that is art, writing, scrolling-to-zone-out-videos, or movies, that is sectioned off and paid for at a premium for everyone wanting to opt out of the AI we’re being forced to swim in.

I have used ChatGPT one time. Back in 2023 when I was working on trying to save Fairfield Lake State Park, I and several others had spent so much going down rabbit hole after rabbit hole on the internet chasing down any kind of information we could find on the developer and anyone associated with them, from elected officials to the water development boards and beyond. I used the Wayback Machine, many county appraisal districts, news archives, obscure Vimeo accounts, you name it, to find information, much of which never saw the light of day because it didn’t really break any kind of particular story but really only supported the one narrative, which was that money and power wins. And the rich have connections everywhere. At this point chatter about AI and ChatGPT was just starting to break into the mainstream narrative and I thought, well, I keep hearing about this miracle of AI, maybe it will be able to pull up something I had not been able to find. Reader, it did not. In fact, I knew far more information than what it spit back out at me, which was a parroting of the most recent news headlines. I closed my account a few days later and realized what a farce it was and have never looked back.

The only other capacities that I use AI knowingly and willingly is through the now default AI search results at the top of Google and through my use of iNaturalist.¹ (I’m sure someone will be able to point out some obscure way I’m not thinking about—please do!) On the former, I use it with a heavy dose of skepticism. I typically scan what it says and then scroll down to real search results, and depending on what I’m looking for, half the time I end up on Reddit looking for answers from people over there. Often, I find, what the AI results suggest on Google are not the full truth or even what the link they use to backup their claim actually says. The AI results may say, “XYZ was first made in Paris in 1745” but you click the link (which is usually also in the main search results further down if you actually scrolled beyond the spoon-fed AI results) used to support that statement and find there’s absolutely no statement within that page to back up that “XYZ was first made in Paris in 1745”. And once you dive into it further you find out that actually “XYZ was imported into Paris in 1745 but had its roots and beginnings in Medieval Germany in 1388” or some other such nonsense. You don’t have the full story with the AI synopsis and so, so many just people stop there.

On the latter, iNaturalist, I see this as one of those intended uses of AI, though I do think it has its problems too, as the vast majority of people using iNaturalist and its sister app Seek, are not using it to its full capabilities. By and large, most people start using iNaturalist to identify a couple of things in their yard or on a nature walk, upload it via the app on their phone, click whatever suggested identification that it brings up, agree that maybe that’s it, and hit upload, only to never return to their account after the fact to double check if it is correct or to see if others have corrected the observation to its rightful classification. It is very beneficial to log onto the computer and re-look at your observations because the suggested species may actually change on the computer compared to what was suggested to you on your phone (iNat sometimes throws errors on the phone for whatever reason) and there are also many avenues to filter through suggested species, even by county checklist, to deduce what it is you even saw. The vast majority of the public isn’t going to do this unless they become iNaturalist nerds, though. I am an iNaturalist nerd and spend a lot of time IDing observations for people and that is how a lot of my flora and fauna knowledge has expanded over time. Paired with field guides, online and paper taxonomic keys, herbarium records, and communication with other iNat users, this is where the real gold is found with using the program. Yeah, the “computer” has slowly learned over time how to identify something in a photo, but it is based on humans identifying and correcting identifications to accurately identify that photo someone uploads via the app. And sometimes the computer is wildly wrong because subspecies exist and photos are poor and sometimes it takes a human eye to look at a geographic location and habitat to realize that what another user uploaded was identified incorrectly. And even then, someone who thinks they know better will still argue with you about it!

---

There’s clearly a difference between what we used to call computer generated or computer assisted tools and AI as it is being promoted and used today. Computer generated codes or programs that would run an algorithm to analyze results or maybe even transcribe something to be edited by a human later, to me, are vastly different than a would-be writer typing into an AI platform a potential idea for a novel or essay, adding a few descriptors to personalize it with the human’s own thoughts, and then copying it into a newsletter or Word document and passing it off as if that person had spent hours (days, weeks, months, years) doing research via reading or interviews and writing it down themselves. It’s the same as every writer on here (and elsewhere) who uses an AI image to illustrate what they wrote—why isn’t a photo you took on your phone worthy enough to illustrate it? Heck, there’s free stock imagery available here and plenty of free Creative Commons licensed imagery on the internet if you took even a sliver of the amount of time you used to prompt an AI LLM to produce some abysmal image to attach to your essay. Better yet, why not let your non-AI writing stand for itself? After all, open a novel and tell me how many photos and illustrations are in there! (Graphic novels not withstanding, of course.)

At the heart of this all I see is consumer culture. “Creating” more = more revenue in some capacity, whether that is just more eyes on a document or paid subscriptions to a newsletter, or if on a photo/video platform more followers and views which can entreat advertisers to solicit brand deals to people who have become their own brands as they create more and more content. The Hank Green Issue. We already can’t keep up with the amount of content being created and I certainly do not (and cannot) read every single newsletter that is posted that I subscribe to. The more time we’re watching the latest Reel or reading the third AI derived essay this week from the Hot New Writer, the less time we’re taking to read actual books, journal articles, or highly detailed blog posts by writers who have spent their time researching their topic of interest and dedicated the time to synthesize the material in a way that our brains were meant to absorb and learn knowledge. We’re also not doing the work of creating the work ourselves by teaching ourselves how to make our own art and improve our skills. Sometimes it is the act of doing something that is more important than becoming the best at it.

Y’all, I don’t even use Siri. I can see how that technology is incredibly beneficial to those with disabilities who need to use that to make their lives run more effectively. But what about when people who can certainly set their own alarm or reminder begin offloading their brain onto services they were never meant to be using, what is that tech doing long term to our cognitive skills? Most people don’t even remember phone numbers anymore because they are saved into a contact list in our phones! Forget AI psychosis, what kind of tech related dementia will we be looking at in 10-50 years? I know I experience a return to “normal” when I spend a considerable amount of time off my phone from just scrolling Instagram—what about everyone who is tied to using AI for asking basic questions or who are stuck in a loop having conversations with a computer? What about all of the kids going through school now and who will enter the workforce over the next decade who have been served AI on a platter and find themselves faced with having to use their own brain for decision making and are unable to do so?

I’m perplexed by this issue. I have spent too much time lately thinking about this, namely because I put a lot of literal brain, sweat, and tears into writing my book, but also into these essays. I’ve taught myself other creative skills over the years by reading how to do it, or more rarely, watching a YouTube video of an actual human doing the thing I want to learn—not a computer generated/AI derived video of how to do it. Or by having someone in person show me how to do something and then learning by making mistakes and trying again.

I think about our increased use of GPS and how many people now struggle to navigate without it, even to places they have been to hundreds of times. Of course there are always going to be people who may be directionally challenged and I’m not discounting any real issues like that, but I think the research shows that for most people, there is a real decline in spatial memory with dependent use on GPS, even in people without inherent spatial issues. I think back to my time working for a tribal government in south Florida for six years and how many tribal elders navigated the Everglades by airboat without a GPS. They knew how to read the land and where they were in relation to one tree island from another by spatial memory. Eventually I got better at knowing where I was without a GPS, too, though I certainly wouldn’t have been able to navigate the 15-20 miles through sawgrass and recall every single airboat trail because I wasn’t doing it every day and didn’t always drive the airboat. I wasn’t making the mistakes and learning to embed it into my brain.

We aren’t learning from our mistakes and embedding it into our brain when we default to using AI to write or brainstorm or research. How are you managing to grasp what you are reading if you are only catching a synopsis and summary instead of the full details of the information. There is article after article coming out right now about the rising rate of illiteracy and how “no one is reading”—they aren’t reading but really, they aren’t thinking critically either, which is a huge part of literacy. If everyone is letting ChatGPT or Claude (or whatever LLM is out there you may use) tell them how to outline and brainstorm an essay, basically hand feeding them the information without using the skills they probably have to do the work themselves, what are they learning in the process? Not much. So then, how are they able to discern when they are being fed a bunch of bullshit? They aren’t. And it seems, even smart people like Hank Green are falling prey to this because the tech oligarchy is making it easy to do so.

I’ve been learning a lot about my brain in this last year since my ADHD diagnosis, understanding how much I think in an associative way. It has always made sense to me and I know it probably drives people insane when I really let loose and set the ADHD free as I’m talking—it’s something I have learned to keep under wraps over the years for most people, so it doesn’t always come out. But, I can’t help but think this heavy push for generative/lazy AI in schools, in work places, across social media—this quiet acceptance that it’s normal and ok for everyone—is tied to the sudden explosion of Flock cameras and surveillance tech, to this decrease in literacy, to the push for data centers in rural areas where thereby roads will be expanded and/or created new, the increase in residential and commercial development into these areas, the reduction in environmental regulations associated with this construction, new pipeline and transmission corridors into these places, and eventually water, water, water—the commodification of water and draining of our aquifers.

It’s all intentional.

---

Of course, there will be someone who may comment here about how they use AI in some some manner and it absolves them of any negative consequences for some reason or another. Someone else will inevitably point out how we’re all hypocrites because we drive cars, we are consumers who shop at Big Box Stores, we eat meat, insert additional grievances here, and if we wanted to be purists maybe we should move to the bush and live off the land. That’s not something worth arguing with someone over because it is only a distraction to the issue. But also, yeah, please drop me off in a small commune in the middle of nowhere that at least has clean water but definitely no cell service.

I’m only half joking.

Yes, societally, we have a choice in many of those options. Some, we really don’t, even if we are given the illusion of thinking we have a choice. But I feel like this recent push to infiltrate AI into every fiber of how we operate as a society is certainly a choice we have right now. Which way are we going to go? Are we going to allow the tech oligarchy to continue to push AI on us and blithely accept the use of these platforms in very dumb ways that result in more data centers being built, more electricity being generated, more ecosystems being destroyed? In a discussion with my friends in our group chat yesterday, one of them brought up that they had recently read about how China was testing out building data centers underwater in the ocean as a means of keeping the infrastructure cool, instead of using ground or surface water to do the job. Ok, so, what happens when there are thousands of AI data centers in the ocean globally? They do increase the water temperature where they are located and I am sure more conglomerated in a region will only increase it further. Not to mention the sound generated by these data centers and how that will potentially impact marine mammals that are already sensitive to human generated sounds like shipping traffic and seismic exploration. What are we even doing here? Is all of that worth Hank Green or any other so-called internet creator having constant access to ChatGPT spitting out information that may or may not be correct for those creators to constantly be churning out content? Sorry whales, you can’t migrate this year because we needed to make sure ChatGPT could outline our latest YouTube video. And that’s to say nothing about what happens if that equipment starts leaking into the water. Don’t worry, the University of Florida already found that this method is already highly subject to cyberterrorism!

Jumping back to iNaturalist for a minute before I wrap this up, because I did mention choices. I did look into iNaturalist for a minute to see what was out there regarding their digital carbon footprint. It certainly isn’t going to be the same size as all of the other generative AI platforms but it isn’t nothing either. I couldn’t find much but there are several threads in the forums about it, such as this one from 2020 that talked about the carbon footprint of iNaturalist. I am sure those numbers have changed in 6 years. I could certainly delete my iNaturalist account. Maybe even better would be for me and everyone else to delete old emails we don’t need and to take excess items off the “cloud”. And even then the comparison between the server data storage we’re already familiar with and what is needed and used for AI computing is on an entirely different scale. So, deleting emails and taking photos off the cloud is really just a drop in the bucket.

---

I offer no real solutions here because I am not in charge of the world. All I can say for myself is that I will resist generative AI as long as possible until it is foisted upon me in some work platform I use with no way out—which may be an inevitability the way things are going. I know this is an incredibly judgmental piece and that many people who have used or played around with AI in completely frivolous ways may be upset about it. This isn’t meant to hurt anyone’s feelings or lay blame at anyone. You may not know about the impacts or implications. Or maybe you do and are ok with it. I doubt I’ll change your mind and vice versa.

We are at a very interesting and odd time in technological development, just like other big technological developments in the past that had major consequences for the environment. This one also happens to affect our cognition as well as a whole host of other things, including intellectual property.

I’m not willing to let it unfold and happen to us without putting up a fight, though.

---

If you made it this far here are a few things to read/watch.

• “I’d go so far as to say that without social media — or, rather, the need to compulsively post on social media, to be digitally productive — there’d be almost no reason for generative AI to exist at all.” - Hank Green’s anti-AI AI psychosis via the Garbage Day newsletter. H/T to cass marketos for the link over on Notes.

• I read the techno-dystopian novel The Circle by Dave Eggers last fall and loved it. Despite being written in the early 2010s it was right for the moment. This is a great interview with him on NPR about the use of AI in writing.

1. Late edit before I publish: To my annoyance, it appears Instagram captions that are transcribed over a video may be considered generative AI. I am seeing conflicting information about this on whether it is based on historic voice to text transcription tools or is now generative AI, but knowing Meta, I would bet on the latter. This is one of those tools that I think is beneficial to use for obvious reasons but at the end of the day if we’re using it to create content on videos for the social media hamster wheel…well, it is certainly something to review as to its value. So, add this to the bucket of potential generative AI tools I use. Would love some clarity on this, though!