Things I Don't Want AI To Help Me With

As we head into spring, I couldn't help but write another AI hater article. Sorry if these are getting old, but I'm enjoying the topic for now. Back at It Again Folks are probably tired of hearing me rant about AI, but I really enjoy all the sort of philosophical questions that the technology has raised. For example, I find the concept of intellectual property to be a really interesting one since generative AI largely requires the left of massive amounts of copyrighted material. Likewise, I enjoy the conversations around the definition of art. Recently, I've been thinking about how much of tech is a solution in search of a problem. Like, do you remember those old infomercials that would show up on TV at night? So many "as-seen-on-TV" products were complete garbage, but they had fun ads. Today, AI has basically taken the absurdity of the "as-seen-on-TV" industry and cranked it up to eleven. We're now seeing AI forced into every possible use case imaginable, from email drafting to employee surveillance. At this point, I am not sure where AI is not going to rear its ugly head, but I can tell you one thing: I know where I don't want its help. I Don't Want AI to Help With Teaching The actual inspiration for this article came from a moment in a meeting I had with some other educators. We were being taught about a variety of active learning techniques, and I was reflecting on all the ones I've tried. At some point, I asked the experts a fun question: say I wanted to try something crazy like "role playing" or a "gallery walk" in my computer science class, how would you recommend I do that? Instead of getting expert advice, I was given perhaps the worst advice imaginable: use Copilot. When I hear something like this, it immediately calls into question the credentials of the supposed "expert" in front of me. What do you mean "use Copilot"? Aren't you supposed to have experience with this subject matter? Needless to say, I was really annoyed with this interaction. Not only did it reinforce the argument I made in “Just Ask Chat”: The Evolution of an Isolating Mantra, but it made me really frustrated with my institution. I was asked to come to these meetings because the institution wanted to improve our teaching. Why should I go to a meeting where even the mildest curiosity is redirected to a chat bot? I think the part that really frustrated me was when the person recommending Copilot told us that they "use it all day, especially when they don't have time to devote the mental energy to a task." Fortunately, before my head could explode, one of my peers jumped in to offer "to spend some mental energy on the task together." I thought this was a sly way to mock the facilitator, and I was very thankful for it. All that said, I really don't think generative AI belongs in education, but I know being this much of a purist is going to turn off other educators. After all, if you're overworked, especially in the K-12 space, it's tempting to use these tools to ease your load. But, I would urge you instead to think about why you're so overworked and consider a more long term solution to that problem: collective bargaining. I Don't Want AI to Help With Academic Research Somewhat tangentially related... but I was recently reading a journal paper. In the manuscript, there was a section detailing how the authors used generative AI to conduct their research. I was thankful they were transparent, but I was also a little bothered by their particular use cases. For example, their most harmless use case was to ask ChatGPT to generate writing schedules for their team. I'm not sure ChatGPT is really meant for that kind of task. After all, how is it supposed to estimate the time needed for specific tasks or allocate the appropriate academic to each section? But as far as use cases go, creating schedules seems pretty tame. In contrast, they admitted to using ChatGPT to look for literature. To me, this raises immediate alarm bells. Part of being an academic is being able to synthesize literature, and I think handing that synthesis process to a chat bot is malpractice. How can I be sure you read the studies and understand them? If you're already using ChatGPT to source your literature, what makes me think you wouldn't also use it to summarize the findings of that research? Just knowing you used ChatGPT at any step in the synthesis process makes me question whether you read any of the studies you cited at all. Hell, you might as well have had the bot draft your argument. Similarly, they admitted to using ChatGPT to generate statistical analysis code. Again, I find this to be malpractice. Educational research is already in dubious territory because you're not required to publish your data; you only have to publish your findings. Therefore, if you collect survey data and have a chat bot analyze it, I'm going to be immediately skeptical of the results. While they didn't admit to doing this, it's only a matter of time before academics start to use generative AI for coding in qualitative research (for my CS readers, coding in this context refers to labeling chunks of text). A quick google search already shows me that there were a few papers published last year with at least one raising some very obvious concerns. In fact, I don't even have to read the article to state my immediate reservation: how can you call yourself an interpretivist, if you're not even going to interpret the data yourself, brother? The authors of that previous article make a similar argument in their "Epistemic Risk No 1: Category Error" section where they state: "it also ignores the interpretive nature of qualitative data analysis, which relies on the researcher's contextual understanding, subjectivity, embodied experience, and reflexivity." If you're considering using an LLM to aid in your research, you might not be cut out for research. That's all I'm going to say. I Don't Want AI to Help With Art Recently, I was watching a "behind the scenes"-style video for KPop Demon Hunters with my daughter, and there was a moment in the video where a couple of animators were talking about extremely small details, like timing and character movements. For instance, I remember one of the animators talking about Jinu's head lift as like his signature move. Honestly, I was a baffled by the level of detail incorporated that the average consumer is never going to appreciate. Even I hadn't noticed the head lift motif after watching the film over like a dozen times with my daughter. That's when I sort of had an epiphany: this is what art is all about. Art is about focusing on details that no one will ever appreciate. I remember doing that sort of thing in band in high school and college. Generally, the audience was filled with our family members who knew nearly nothing about music. We would spend entire rehearsals working on things like dynamics and articulation; things that our audience would never appreciate. See, art isn't about consumption. It's about you. You get to obsess over all the details. It's something I even do on this site. I spend more time than I would like to admit on things as small as sentence structure. Surely, as the reader, you don't care or notice. You're just reading to learn something, but I'm writing to develop my voice and deepen my understanding of a topic. These are things that generative AI can never hope to replicate. These models can produce technically good outputs, but they have no soul. I don't mean that in some supernatural sense. I mean that literally. Any "art" that is produced by generative AI is inherently devoid of soul. There was never a single conscious decision made by the model: just data in and data out. So, I have no interest in using AI to help me with art. In fact, I think it would actually get in the way of the artistic process. Using it for brainstorming would rob me of the joy of serendipity (e.g., the sudden realization that I have a great idea for an article). Using it to write would rob me of the ability to develop my voice and style. Personally, I don't like being robbed. I Don't Want AI to Help With Basic Tasks If there was any topic in this article that would get people fired up, it's this one. I just have no interest in using AI at all. Every time I hear about someone using AI for some menial task, I'm always a little put off. It's not that I think that people are being lazy or that I have a problem with laziness generally, but it's such a cop-out. For one, if you're so busy that you can't get your work done without the assistance of AI, then it might be time to develop some class consciousness. Don't turn to the chat bot. Turn to your coworker and form a damn union. You deserve to be able to maintain some level of quality of life at your job that doesn't require you to compromise your individuality or morality. Likewise, if you start offloading basic tasks like emails, our social fabric is officially cooked. I really can't put this into better words than Northernlion, so I'll just drop his clip here: https://youtube.com/shorts/kiFa494bl5c?si=J4hukzZwz5pUf-ua Here he's talking about love letters, but I don't think it's much of a stretch to say that the same folks sending ChatGPT emails are probably also using ChatGPT to communicate with their supposed loved ones. The Naysayers One thing that has been really annoying about the advent of generative AI is that a ton of people make the case that it's essential technology for folks with disabilities. This drives me completely insane because it provides the perfect armor to any criticism: if you don't like generative AI, then you're ableist. And, the frustrating part is that I can't even make the "what were you doing before ChatGPT?" argument because it sounds like I'm saying "what were you doing before wheelchairs?" Like, perhaps ChatGPT is a major innovation for some disabilities. Maybe you do need it to craft professional emails. But, I'm highly skeptical of that. Society should be moving toward widespread acceptance of folks with disabilities, not creating tools to make them more "normal." If you have social skills issues due to a disability (e.g., autism), you should be able to send an email without risk of it not seeming professional enough, or as one Reddit user put it: as an autistic person i didn’t ask for AI to help me, i just want people to be nicer and try to see how it is to be in our shoes, like actually research and get educated so we can change society to be less hostile to us, it’s so very hostile to us I guess my point is that any argument that props up generative AI because it might help some disabled folks seems disingenuous. It honestly feels like someone put in the prompt "how do I sell this technology in a way that makes people feel bad if they're against it?" Or again, as one Reddit user puts it: I hate when tech bros use the intellectually or physically disabled as an excuse for AI grifts. I see that shit all the time among AI "artists". It feels just as fake here. A man who has no experience in the creative arts talking about things he doesn't know anything about. I feel for his daughter but this article reads like an ad for Google and Open AI born out of wishful thinking rather than actual reality. If you want to see a really interesting debate around disabilities and AI, check out this Reddit thread. At a quick glance, it seems like disabled folks really don't like being used to market AI. Perhaps my favorite quote from the thread reads: "Accessibility is about more than personal convenience." Again, I didn't have a great way to end this article. I was enjoying toying with different places in my life where I'm not interested in AI-integration. Then, I stumbled into the comment section of that NL video, and I say the naysayers using disabilities as a shield. So, that's where this article ended up. Anyway, I hope you enjoyed it. Since I plan to hide this article behind the paywall, I won't bore you with the usual links. As always, thanks for reading!