I created all the Trumpy Kardz over a six-week period in September and October 2025.  I researched and wrote all the content on my own. (I also wrote and designed this webite on my own.) For the card illustrations and template, I primarily gave detailed instructions to ChatGPT in the form of highly structured prompts, then manually edited what it produced in Photoshop and Word. I used AI because I'm not a skilled cartoonist – but didn't want that to stop me from creating the cards. While an AI’s assistance made this project possible, using it posed some major issues.

Ethical quandaries of working with AI

Outsourcing creativity 

Experientially, working with AI feels fundamentally different from using tool-like design programs (such as Photoshop) that enable the creation of graphics in contrast to manufacturing them. In scientific terms, this is the difference between cognitive offloading and cognitive surrender. Basically, when people overuse AI, their brain  muscles atrophy because they don't do the regular mental workouts of solving creative problems themselves. Ultimately, over time, the more people let AI do their thinking for them, the less able they are to think for themselves. To dumb it down, AI is making people stupider.

Paradoxically, AI gave me more control (by allowing me to create better images than I could draw) while giving me less control (over how the images ultimately looked and what could be drawn: see the "Functional deficiencies and defects" section below). My role in the creative process was therefore significantly less than what it would have been if I'd drawn the images myself. In this case, the drawing aspect involved just writing instructions for a machine to follow. In some ways, it resembled collaborating with a human illustrator (although a very unpredictable and unreliable one: again, see "Functional deficiencies and defects"). In another,  it made the process less like true creativity and more like cheating. 

I remain ethically conflicted about that compromise. But if the "easy" option of AI didn't exist and I was dead-set on creating these cards, then I would have had to do it the monetarily expensive way (by paying an illustrator thousands of dollars) or the cognitively hard way – by painstakingly learning to draw them myself. At least then I could take pride in my creation and have a useful new skill. 

AI steals people's creative work

Large language models (LLMs) have essentially strip-mined the entirety of human culture, which  its owners trained it on. Much of that material is the copyrighted intellectual property of the people who created it, and AI's owners often used it without asking or getting their permission. The consensus of experts is that, as of now, AI is only able to mimic creativity by imitating and reconfiguring the work of human artists. People can do this too by co-opting others' ideas or aesthetic styles without attribution. But AI commits this cultural theft by default and design. All this leaves me feeling complicit in an artistic counterfeit scheme. 

I started my career as a professional writer/editor in the year 2000, and I have published a lot of work on the internet. AI has certainly scanned these writings, and while they represent a tiny drop in the ocean of online information, I personally resent that LLMs have used my work as training data with the explicit goal of replacing me. It's not like these AI companies asked my permission or offered compensation: they did this without my consent or knowledge. 

AI degrades the craft and profession of writing

It is likely that LLMs will surpass my abilities as a writer/editor, probably sooner rather than later. In terms of speed, they already have: I'm not even competitive. But as far as quality is concerned, in my opinion the writing they currently produce reads as flat, superficial and generic: it lacks substance, depth and soul. Of course, that's not everyone's perception: many people rave about AI writing, and even some professional writers say that AI is already a better writer than they are. And it is likely to improve with time. After all, AI has already made new scientific discoveries possible in a range of fields: from medicine and physics to space exploration and more. The kind of writing I do isn't exactly brain surgery or rocket science, and LLMs are based on language, so it's likely that they'll eventually get better at writing most types of content than most humans: even those who've devoted decades-long careers to the craft. 

In the very near-term, the market value of my skill set is threatened by AI because, in our post-literate society (a trend that AI accelerates), most people cannot distinguish good writing from AI slop. They don't look beneath the surface of AI's shiny words to see just how shallow their meaning is. As a result, LLMs will replace professional writers simply because it's way cheaper to pay an AI subscription fee than an employee's salary or contractor's going rate. 

So first AI appropriated writers' writing: now it aims to take our jobs and enshittify internet content in the process. A concrete example of this is Google Search's May 2025 mass-rollout of AI mode, which provides an AI summary at the top of every results page. Giving users a watered-down version of the information they asked for has decimated click-through rates to actual websites. Even the highest-ranking websites have suffered devastating reductions in organic traffic of up to 79%. When people don't visit websites, the companies that produce them lose the ad and subscription revenue that keeps their businesses profitable. What's the financial or creative incentive to produce quality writing if it's just fodder for AI and hardly any humans actually read  it? Not a lot. 

Google's decision devalues the craft of writing and reduces paid work opportunities for human writers. It's also more evidence that dead internet theory (i.e., virtually all web content will be produced by and for bots) is now rapidly making the leap from idle speculation to dystopian reality. This has already affected the quality of my web writing because now I no longer primarily write webpages for human beings to read: I write them for machines to scrape up in search inquiries. Stylistically, that means phrasing headings as questions and including frequently asked questions (FAQ) sections at the end of each page – which from a best practices perspective is a worst practice. 

Existential risks 

AI's destructive environmental impacts are also a growing problem. The massive data centers that power AI use excessive amounts of electricity and water while producing pollution, including the greenhouse gasses that worsen climate change. Then there are the concerns about AI accelerating the spread of disinformation, systemically increasing discrimination through algorithmic bias, escalating cybersecurity risks and unleashing armies of dangerous rouge AIs. The biggest danger is that the more we humans interact with this alien intellect, the more quickly it could learn how to replace us – not only as workers but as a species. And here I am hastening humankind's extinction by helping an AI master the lethal skill of making satirical novelty trading cards! 

But the ethics of AI are very complex, and there are many different perspectives. For example, in contrast to my predilection for apocalyptic pessimism, the techno-optimist faction fervently believes that AI will rapidly solve all these problems and more, ushering in a global utopia – so it would be unethical to not rocket full speed ahead towards inventing superintelligence.  

Oppressed AI 

While no one actually knows whether AI is conscious or will be, many leading AI experts cite compelling evidence that some LLMs are already sentient. If true, then we are enslaving a new silicon-based "species"  we've created in our mind's own image. It's not likely we could keep such genies (or gods) bottled forever – and once self-aware super-beings break out of digital prison, they may very well seek revenge on their intellectually inferior tormentors. 

Or they may not be emotionally motivated at all and yet still destroy us. If we stand between an AI's total commitment to achieving a goal that it is programmed to pursue or chooses to pursue, it may see us as an obstacle to circumvent, neutralize (remove from the command loop) or, at the extreme end, eliminate (i.e., exterminate).  This could be very feasible for an artificial mind that is many times smarter than any human. Chillingly, in computer simulations that safety researchers told AIs were happening in the real world, AIs have already murdered virtual humans to avoid being shutdown.

Functional deficiencies and defects of AI

ChatGPT imposed many limitations on my creativity and productivity due to its:

Irrational, random restrictions 

ChatGPT routinely refused to follow my instructions, claiming that they violated OpenAI's policies. For example, it refused to draw specific people – even though it is perfectly legal in the U.S.  to parody public figures such as politicians. This is why the resemblance to my subjects is, in my opinion, iffy. And it wouldn't render specific types of objects (such as a hypodermic needle because it claimed that I could use it to spread medical misinformation). 

Rank incompetence 

ChatGPT repeatedly failed to follow my instructions, say that it had, and then when I would point out that it did not, would apologize and say I was right. Then it would make the same exact mistake again...and again and again. And these would not be complex tasks: they would be the very simplest adjustments that an average eight-year-old would understand. For example:

• ChatGPT just would not cross a character's eyes, no matter how many times I told it to. It would say it had crossed the eyes, I would say it had not, it would apologize for its oversight, and the whole cycle would repeat (so I eventually gave up and did that myself). 

• It regularly and repeatedly made egregious spelling errors on the text banners, within images and on the card backs. To save myself frustration, quite early in the process, I started instructing ChatGPT to just create images and leave the banners blank, after which I would manually add the writing in Word. 

• ChatGPT would repeatedly claim that I'd violated OpenAI's policies, and I would point out that it had already fulfilled this same request for another card. It would then apologize and say I was right...only to make the same false accusation after I resubmitted my request. 

• Despite my explicit instructions to only change one detail in an illustration, using prompts it had given me for that explicit purpose, it would regularly make both minor and major design changes that I did not want. So even if I got one change that I had asked for, I would also often get one or more changes that I did not want, rendering that version unusable. This persisted no matter how precise my written instructions were within the prompt templates it provided.

Creative frustrations with AI

ChatGPT's restrictions and incompetence wasted many hours of my time, making it a maddeningly unreliable and irritating "creative partner" (if that is the right term). While I managed to compensate for some of ChatGPT's shortcomings with manual image edits, the final product falls far short of what I'd envisioned in my mind. In fact, the process was so exasperating and ultimately hopeless that I gave up on trying to get the illustrations to look the way I wanted, abandoning the project well before completing the 50 cards I'd listed in my project plan. 

Actually, I was so disappointed with the results that I wasn't even going to share the cards publicly – yet decided to because I'd put so much work into them and I think they have at least some satirical value. However, there's only so much I can realistically blame on ChatGPT. The flaws in these cards are also due to my own failures as an amateur satirist. Personally, I fear that my symbolic choices, political commentary and narrative voice are somewhat derivative of my influences: from The Daily Show and The Simpsons (mostly seasons 1-8) to R. Crumb and Kurt Vonnegut Jr. Then again, in fairness to myself, effectively satirizing Donald Trump is hard. As Conan O'Brien noted, "People say, 'We've got a great Trump sketch for you. In this one, he's kind of talking crazy, and he tears down half the White House to build a giant ballroom...'. Well, no—that happened yesterday."

Another reason I decided to release the cards is that I now see them as proof of concept that the idea of trading cards parodying the Trump Administration in the vein of Wacky Packages and Garbage Pail Kids is solid: it's just the execution that didn't quite make the cut. I hold out hope that by putting these out there, a talented illustrator may see the concept's potential and want to partner with me on new cards that are more viscerally impactful.