Lately it seems that all some people want to talk about is AI. Granted, there are plenty of people who can go through their day without discussing it at all, but when people do have it on their mind, they are not shy about letting you know. One day in particular last week, I had several conversations about AI before being asked to comment on a recent AI/employment study on KVVU-5. Having had to string together coherent thoughts in response to some great questions (not all of which made it into the final story), I thought that I would use this space to sketch a few potential AI Implications that we might want to think about.
When people learn that I work at a university and that, among the things I do there, I teach, they immediately ask my thoughts about AI’s potential as a cheating tool. While I think questions of academic integrity are always worth discussing, I don’t think that this is the most pressing issue that AI presents to us, because cheating didn’t start with AI and won’t end with it. There are so many ways to be dishonest academically, and I don’t see much difference between farming an essay out to an AI agent and paying someone to write it for you (although the former offers fewer barriers, probably making it more widespread). Though AI use might make cheating easier, it doesn’t fundamentally change the idea of what it means to be academically dishonest, and it comes down to a tale as old as time: using an unapproved outside resource to produce work that the student passes off as their own.
Perhaps you think I am being overly broad in lumping AI use with cheating. Yes, I know that some instructors authorize the use of AI to brainstorm, create a rough draft, and polish the final draft (which makes me wonder what exactly the student is contributing here; I’m thinking of the Solider in White from Catch-22, because at some point it’s more honest to hook the two jars directly to each other and “eliminate the middleman,” as it were). But assuming that the instructor hasn’t authorized it, I struggle to differentiate between AI-powered outside help and old-fashioned outside help.
To me the more significant challenge with AI is that it can become indispensable to people very quickly; the day that I was interviewed, I spoke with someone who said they use AI for everything, work, school, and home. I don’t know enough to know whether AI can be addicting, but that sounds to me like someone who is giving it an oversized role in their life. We all rely on things like oxygen, water, and nutrition every day, so there’s nothing wrong with dependence per se, but having a product of a large corporation become so intimately essential to you–well, I just don’t think that story usually ends well.
But over-reliance on a new technology isn’t, I think, the most concerning part of AI usage. As I alluded to in my interview, the chief danger of AI might be that it comes between people.
We can be cynical and say that insulating people from each other might be a positive social good; I’m thinking of how Douglas Adams envisioned a galaxy where all impediments to mutual expression were swept way: “Meanwhile, the poor Babel fish, by effectively removing all barriers to communication between different races and cultures, has caused more and bloodier wars than anything else in the history of creation.” From that perspective, filtering contact through an AI tool that tells us “what words to use” might help de-escalate conflict, and that’s something to consider if the goal is to keep things lukewarm. But if we want to communicate with others about what is important to us and what we really want, putting another layer between us and the other person going to, ultimately, be an impediment to the kind of raw exchange of thoughts that, while momentarily uncomfortable, can actually get us to where we want to go.
So if remaining locked in a awkward status quo is that important, by all means use AI to “soften” your language and conceal what you really want and how important it is. But don’t be surprised when you aren’t as happy as you think you should be at having “avoided conflict.”
Yes, using AI to draft an email can save some time. But if the message you are sending isn’t worth your investing some time to write it, why would it possibly be worth the receiver’s time to read it? Let’s think on that: if you can’t stomach the emotional strain and time investment of writing something, maybe, just maybe, it isn’t something you really have to—or deep down, want to—say.
And as far as using AI to write and/or grade assignments in school goes, I find something personally dystopian about a world where students submit AI-authored work for instructors to grade using AI agents; even worse is the prospect of students trying to ingeniously skirt anti-AI rules by hiding their AI use while instructors use AI tools to try to “catch” them. Now I’m thinking of Blade Runner. We are treading close to a classic Phillip K. DIck scenario, though I think that in Dick’s telling, the student’s AI would deliberately output a bad paper because it didn’t like the student, or the instructor’s AI tool would report false negatives because it felt sorry for the students. Or maybe the tool reports every submission as AI to point out the futility of it all. Or the instructor’s AI reaches out to a student because, in reading their paper, they find a kindred spirit.
As entertaining as Dick’s stories are, I don’t think any of his characters enjoy being in them, so while it might sound superficially cool and exciting to star in your own sci fi tale, I imagine that once you are actually neck-deep in it, the charm will have long worn off.
The alternative to AIs catching AIs and AI-penned emails read and responded to by AI is an admittedly radical departure from the current groupthink (at least as relayed through the incessant pop-ups and notifications I’m getting demanding that I use AI to do everything from watching my front door to summarizing a three-line pdf certificate of achievement), but one that has a long history of admittedly mixed results: thinking for oneself. That doesn’t always lead to the smoothest of words and maybe not even the most-desired outcome, but it served our ancestors (as I said, with mixed results) for millennia, and enough of them muddled through to pass their genes on to us, so it can’t be all bad.
In short, I’m not the person who can give you an argument for the responsible use of AI in professional or educational scenarios, but I feel confident in saying that, when you want to get through to another person, it is probably going to get in your way, and it might even degenerate into a conversation between two AI agents with humans as merely their button-pushers. I like to talk about communication within conflict resolution as being primarily about a “meeting of the minds,” and AI intermediaries seem designed to subvert that meeting.
So until next time, expect the unexpected, stay informed, and I’ll stay informal.
30
