AI is to Work as Study Groups are to Learning

Thinking to the days when you were in school, try to recall a time when you thought that it made sense in class, but you couldn't do the homework. Or you did the homework, but didn't do well on the test.

I fell down this hole a time or two. It helped me give advice to very bright students in a college program for advanced high-school-aged learners:

When studying as a group, be careful to understand what the group knows compared to what you know. The exam is not a group project.

In college, I had great friends I liked to study with. We would do our homework together. Everyone contributed and everyone knew who the brighter folks in the group were. But that didn't mean we weren't doing our own work.

Even if I thought I was doing my own work, I was later convinced that some of the group studying was, "... and then some magic happens..." and there's the answer, without fully understanding what the magic was, or that it had been there at all.

I'm nervous for an exponentially worse experience with AI, in school and at work. For the purpose of this sweet sweet blog, the focus is at work.

AI-yi-yi

Don't believe the hype. The hype being "artificial intelligence." The thing we call AI, is generative, not general. According to Coursera (the best concise definition I could find, not provided by AI, probably) is that current AI is generative and uses what's fed to it. Whereas AGI, or artificial general intelligence, would need the development of, "sensory perception, manual dexterity, and social and emotional understanding."

In short, we're putting a lot of trust into a system that is a very fast, quite good, reporting agent, not a thinking partner. "Oh, you're right, you did tell me to exclude all references before 1950. Thank you. I will revise that now."

And we have folks out in the wild talking to robots as though they're people, falling in love with them, and, as I understand it, marrying them in some way or capacity.

I don't blame the people. They're getting something from the AI they're not getting from humans in their proximity.

Same is true for work... except for the amorous relationship, I hope.

People use AI and it gives them an Easy Button. Yeah, I could learn to code, or I could ask Claude for it and use whatever it spits out because Claude is "smarter" than I am at writing code.

Claude has a massive dataset from which to pull to follow the instructions I'm informed enough to give it. Claude's expertise stops where my understanding of how to ask for what I want ends.

I was writing papers during the transition from hand-written or typed to word-processed. The quality of papers presented, according to some teachers and faculty, went down and they didn't know why. Even as a young lad, I thought I knew. Students could produce what looked like a high-quality final product that was clearly legible, had great margins, was orderly, and looked, frankly, like a book.

The word-processed papers had the appearance of quality without the matching quality of content.

With AI, the output might look like a final draft, but if it's not, we're collectively starting to lose our ability to interrogate the quality AND then make adjustments to take the rough draft to a functional product.

Workplace Culture and AI

There is no plausible way that AI is not negatively affecting workplace culture. People are turning to their favorite robot to help with slide decks, emails, proposals, and so on. How do people interact meaningfully if they've got a robot as their cognitive intermediary?

Not rhetorical. Literally how do they interact person-to-person?

They do it less.

Part of the reason may be convenience. Then add in fear of making a mistake. Sprinkle on some anxiety, and top it off with a bit of missing confidence. Now we've got a recipe for, "I asked ChatGPT and it said I should..."

This also removes some of the accountability for a person knowing, thoroughly, what in the world they're talking about. If they wrote the proposal from a blank sheet. Odds are good that they know what's in it. If CoPilot wrote it, okay, sorry, that would probably never happen, if Gemini wrote it, the author-turned-presenter might not know the detail of the proposal nor the underlying assumptions and data that fed the outcome.

Be a Part of the Rebel Alliance

We're not in the "revolt against our robot overlords" part of the timeline just yet. One way to avoid gettign there is to rebel, appropriately.

When I get stressed, I make a run for Peanut M&M's. Well, I used to. That was my Easy Button that, as it turns out, didn't actually solve anything.

When you're stressed or even just pressed for time, the AI-Easy-Button can beckon. Avoid it as much as possible. Not everytime, and if you use it, fine. The goal is less, not none. The reason is, the more you let AI determine your course of action, the less you will have agency for your own decision-making.

More to the point, if you cannot clearly and relatively easily articulate why you made a decision, then your credibility is in question. So is your mental acuity.

Keeping your brain in the game will ensure your own skills remain sharp. This is true for your work product as well as your workplace relationships. When people can interact as people, the workplace culture is more productive, rewarding, and predictable.

On this particular timeline, the last thing any of us need is a way to be less human.