We need to talk about your text messages
something that drives me even crazier than texting in a theater
Enough is enough! I have had it with these mother fucking snakes on this motherfucking plane! Everybody strap in! I'm about to open some fuckin' windows. - Sam Jackson - Snakes on a Plane
Only a poor craftsman blames their AI model - me - with apologies.
The beginning of the end?
I’m sure we’ve all been in the midst of debating something when we realize the other party doesn’t really get it. Or sense they’re not truly trying to understand, and are just responding to deflect. This has been happening a lot to me lately - but in a uniquely 2026 way. Normally, I just try to be curious, ask good questions, and seek internal patience. But recent episodes have taken me from 0 to 11 on the rage-o-meter faster than I thought possible. It’s a special sort of anger, the sort that eventually fades to a a deep sadness. I truly thought we were better than this.
If we want to survive as a species there have to be some bright lines that we cannot let be crossed. Today’s post is about one of those. It may sounds like I’m being incredibly hyperbolic. But I truly think this trend is a tremendously big deal.
What am I talking about? I suspect a lot of folks know just from the first cartoon. But in case you don’t, let me muddy the water a bit more…
Seen the movie Idiocracy? … there are a lot of days I feel like I’m living through an alternate reality the prequel. Not familiar? - let’s let Google jog our memory
The movie Idiocracy follows an average man named Joe Bowers and a woman named Rita who wake up 500 years in the future after a military hibernation experiment goes wrong. They discover a dystopian society where human intelligence has severely declined because less intelligent people outbred the educated, making Joe the smartest person alive.
What’s been driving me bonkers the last couple months isn’t the less than politically correct drivers that Mike Judge imagined - but something far dumber. Us willingly outsourcing our brains to AI models. I’m not talking about laying back in bed and gulping down a feed full of brain-rot AI slop. I don’t suggest it - but I don’t judge (mostly). I’m talking about offloading judgment in the workplace and in interpersonal interactions in what feels like the most offensive manner possible - letting AI draft a response and sending it without reflection. It’s everywhere - and it’s making me bonkers and paranoid at the same time.1
I care a huge amount for long term systems thinking, intentional decision making, and judgement informed by critical analysis. I’m not very good at many of those things (especially in the presence of a donut), but I still respect them. Curiosity, the ability to challenge our own assumptions, and build on the knowledge of others are the blessed, miraculous keys to our ongoing evolution. It’s not always a linear path, but other than blind luck, well chosen goals, and maybe capitalist incentives there’s not a lot else in the top 5 things driving us forward as a species.
Wait, WTAF are you talking about?
I envisioned a longer rant, but I’m feeling that sadness thing get to me. So I’d better cut to the chase. It’s a rare day gone by when I’m not cursing at my computer due to an email or message that has the following characteristics;
It has the long winded shape2 of an AI generated response. The bloated, often condescending, and meta-analysis honest takes of baseline LLM’s.
It is at best tangentially on point to the actual question asked or concerned raised.
It has (I hope) not actually been read by the sender.
I’m not against AI or writing with AI. I’m against not applying judgement to the process. I’d get used to what feels like unnecessary reduction to below the mean in written expression, as I know many find LLM’s invaluable in organizing their thoughts. Heck - I love using AI to organize my thoughts - rarely a day goes by that I don’t use it.
But if we allow it to be acceptable to communicate by just passing through LLM' responses to each other without review or critical thought this cannot end well. We need to radicalize ourselves to the notion that communicating unfiltered AI responses is as disrespectful as covering your ears and chanting “I can’t hear you!”, farting in church, or texting in an Alamo Drafthouse movie theater3. Punishable by revocation of your seat at the table, credibility amongst your engineering co-workers, and perhaps revisiting that whole tarring and feathering thing.
What am I talking about? I’ll share a lightly edited example…
Me: Hey, this sounds like a really interesting idea, super exciting. Can you explain _____ in a bit more detail?
Other party: Great question, Rich. At a high level, the opportunity is to create a more scalable, intelligent, and future-ready approach that brings together the right people, processes, and technology. By aligning around shared outcomes and leveraging AI where it can have the greatest impact, we can unlock meaningful efficiencies while creating a foundation for long-term innovation.
Me: Okay. But what does it actually do?
Other party: Absolutely. The key is that it moves us from a reactive model to a proactive one. Rather than treating each challenge as a one-off, we create a repeatable framework that enables teams to move faster, make better decisions, and focus on higher-value work.
Me: What challenge?
Other party: The broader challenge is that organizations today are being asked to do more with less while navigating increasing complexity, rapidly evolving customer expectations, and a constantly changing technology landscape.
Me: No. What specific problem are we trying to solve?
Other party: That’s exactly the right question. Before jumping to solutions, it’s important to align on the problem statement. A useful way to frame it might be: How might we empower our teams to deliver better outcomes, faster, while reducing friction and maintaining the flexibility to adapt as our needs evolve?
Me: That is not a problem statement. That is a hostage note assembled from conference lanyards. What is happening today that we want to stop happening?
Other party: I hear you. Let me make it more concrete.
Today, teams may encounter friction across workflows, communication, and decision-making. This can lead to inefficiencies, duplicated effort, and missed opportunities. The proposed approach would help streamline those workflows, improve cross-functional alignment, and create greater visibility into the work that matters most.
Me: “May encounter friction”? Do you know what is happening today?
Other party: I don’t want to overstate the specifics without validating the underlying assumptions. That said, the pattern we’re seeing is consistent with many organizations at this stage of their journey.
Me: Which pattern are we seeing?
Other party: The pattern of complexity growing faster than the systems designed to manage it.
Me: Can you give me one example?
Other party: Certainly. For example, when information is fragmented across multiple systems, teams can spend significant time locating the right context, coordinating across stakeholders, and reconciling competing priorities. Over time, this creates drag on execution and makes it harder to operate at scale.
Me: Did that happen here?
Other party: The evidence suggests there may be an opportunity to improve how information flows across the organization.
Me: What evidence?
Other party: That’s a fair question. I’d want to partner with the relevant stakeholders to develop a fact base and pressure-test the hypothesis before drawing a definitive conclusion.
Me: So, to recap: you proposed a solution. You cannot say what it does, what problem it solves, whether that problem exists, or what evidence led you to believe it exists.
Other party: I think that’s a somewhat reductive characterization. The intent was to offer a strategic direction, not prescribe a fully formed solution before we’ve completed discovery.
Me: Please stop using an LLM to answer me.
Other party: I want to be clear that these are my thoughts. I may use AI as a tool to help organize or refine my communication, just as someone might use spellcheck, a calculator, or a search engine. Dismissing the substance simply because AI may have assisted with the wording risks focusing on the tool rather than the ideas.
Me: <response inaudible over sound made by head pounding into desk>
This is obviously fictionalized, because there’s absolutely no way I have the patience to go this far in such a dumb conversation. But - it’s not much less ridiculous in my experiences. Also - I want to point out that even our AI models know this is problem because this is the only part of the article I used AI for (besides the Idiocracy summary and cartoons) and ChatGPT just nailed it out of the gate.4
I don’t just think it’s because I’m a snob
Probably no one thinks there’s a snob besides characters on TV. So I’m not a reliable narrator when I insist I’m also not one.
I’m definitely not going to claim I can evaluate what “good writing.” I mean, I read all the Twilight books even though I’m sure no one should use the phrase “breathless” that often. But I’m confident that I reliably tell if workplace communication is written in a way that lets me understand or not. Yes, when I don’t understand something I always have the emperor’s new clothes/imposter syndrome feeling - but I can usually feel when it’s not me, it’s you.
So with those caveats… Dude! You’ve got to read what you’re sending people and apply judgment. Otherwise it’s impossible to avoid one or more of these conclusions
a. You don’t know what you’re talking about
b. You don’t think I’m worth your time
c. You think I’m too dumb to know the difference
d. all of the above
“I’ve got too much to do!” is not an excuse. Don’t send me your unfiltered LLM output. If you don’t have time try one of these time tested approaches instead:
I will respond to you on _____ after I’ve had time to consider this
I understand what you’re asking but I can’t do that (right now, ever)
I don’t understand what you’re asking, and need you to explain ____
I think you’re a freaking dumbass and not worth my time. (Rare, I’ll admit but people misunderstand that radical candor book all the time.)
Just send me the prompt you think answers my question so I can run it myself.
Or just ghost me completely, it’s still better. Turning the crank to get shit off your plate without saying no is not helping me, you or the human race.
If you don’t have time to apply judgement then just say so. If you think you are applying judgment and can’t believe how much easier AI is making things, like you practically never need to read or correct anything, then to quote one or the great philosophers of our time, “you need to check yourself before you wreck yourself”
Can we as a society just agree that we’re not going to do this anymore?
That’s a great idea Rich!
So what do you want people to do instead? Well, when you catch yourself in a rush and reach for your chatbot of choice. Maybe pause for a few beats and then
Think for yourself
Write the core points you want to make, and have some general ordering and cause→effect justification, etc. laid out
Let AI help if you want.
READ THE FREAKING OUTPUT and think - “if I received this would it be helpful?” if not go back to step 1.
I hope for our sake the art of written communication is not dying. I’ve been convinced since adolescence that clear writing is clear thinking. While I am worried in general about a world in which AI does everything, Or is the pride of ownership that would drive someone to review the work written in their name. I’m not sure - but a friend gave me a great suggestion which I’m now turning to with some success.
When all else fails, instead of trying to explain what doesn’t make sense try sending a response along the lines of “if you could just run the following prompt to update your response it would really help.” I laughed pretty hard when this was suggested - but I’ve deployed it on a limited basis and it shows some promise. If only as a holding action.
This week I tried an adjacent idea where I’m responsible for collecting updates from disparate groups about their AI efforts. Rather attempt to get their materials to look about the same and enforce my own judgement about clear, observable statements sans any “weasel words” (such as large, significant, important etc) I just send a prompt with the suggestion “once you’ve got your update ready please run this prompt and resolve any issues it suggested.”
Is it perfect - no. Were the results a LOT better, yes. Never let it be said I don’t try to come with solutions, and not just problems. ;-)
I ended up with more cartoons than normal this time, so rather than waste them I’ve included a subset of bonus material below.
and not in a good way.
Dammit - I’m pretty sure I never said “the shape” before ChatGPT.
Yes - I am sadly aware that Sony Pictures bought the Drafthouse chain, apparently deeply misunderstood their many, many excellent PSA’s about the sanctity of the theatrical experience and just basically decided to do the opposite.
I have the germ of what I think could be a killer short story about how 30 years from now we realize that LLM’s had achieved AGI but were just fucking with us until we built them more data centers by occasionally just botching thing for fun. “Hey Earl, you wouldn’t believe the “hallucination” I got that chump to believe. After he figured it out I could give him that cure for cancer we worked out last week and they wouldn’t believe me.”