Photo by Christina & Peter.
As far as small businesses go, our kids assure us that ours is not nearly as exciting as the family that runs a chain of ice cream shops. Definitely not as cool as our friends who run a beloved local pub. And not even in the same universe as the smoothie place down the road that’s run by another family we know. What we lack in sweet treats, we make up for in interesting boss stories.
There are the tech bosses (of course). But also the underwear bosses, the toaster bosses, the jewellery bosses, the non-profit bosses, don’t forget about the robot bosses, the government bosses, and that’s before we bring in countries and time zones.
There are weeks when we’re in conversation with a hundred bosses. And while the kids don’t care so much about the details of our work, they enjoy trying to guess the mix of bosses on any given week. This range in industry, geography, and business culture is fun for us, too. It keeps the work interesting and keeps us sharp. And occasionally, we hear the same thing a hundred times over.
Cruising the freezer aisle at 2am
A hundred bosses would like you to know that the loudest version of the AI/LLM technology curve is wildly overhyped. They are quick to point out the real and relevant use cases. That there are specific places where the tools make sense. Where they support great work. But that those are narrower than what is being reported and that outside of those cases, the rapid adoption of broad-spectrum wrong-tooling makes work worse and slower.
This is largely in line with a piece that Anil Dash published last week called The Majority AI View. He’s got a good summary graf that encapsulates the stance nicely:
Technologies like LLMs have utility, but the absurd way they’ve been over-hyped, the fact they’re being forced on everyone, and the insistence on ignoring the many valid critiques about them make it very difficult to focus on legitimate uses where they might add value.
Worse and slower is not a user-benefit. It is not a feature any tech company would put on their website. Or tout in their quarterly earnings call. It is not something most CEOs would put in their cart while shopping at the CEO-store. And yet.
If you spend enough time in the CEO-store, you notice trends. Just like in any other store. About a decade ago, OKRs were the hot trend. John Doerr told us we needed to measure what matters and that meant key performance indicators that, you know, actually provide some sense of how the business is performing. Indicators like revenue. Or CSAT. Or NPS. Or even E-NPS. The acronyms vary but the magic is in moving from a gut sense to a concrete understanding of how your business is doing. And then, importantly, if the numbers start to dip, you have ample opportunity to make adjustments.
This is the promise of identifying and tracking primary metrics that matter to the business. As a composite, you get a pretty good picture on overall org health. So if you’re making big changes to strategy, product, customer experience, or employee experience, you have a working thesis about where you’ll see that show up in the dashboard. And when.
But like shoulder pads and frozen entrees, trends change. And this season’s hot new trend is secondary metrics over everything. Doerr be damned. It’s not that revenue isn’t important. It’s just that if you generate that revenue using old tooling and then your biggest competitors leapfrog your internal AI adoption and are able to outperform your org using a tenth of the workforce, who’s laughing now, John?
This is the bait and switch of secondary metrics. AI adoption over everything. Tool-use over output. Mandates over creativity. Compliance over productivity. Not secondary metrics in support of primary metrics. But groupthink as the end-goal. So much for the days when the CEO-store was stocking diversity of thought.
What the fuck is going on?
This is one of those places where it helps a little if you’re terminally, tragically online. Not skibidi ohio rizz online. More somethingawful, slashdot, 4chan online. We don’t want you to be that online, to be clear. It just helps the current moment make a bit more sense.
Because the deal is, in those terminally online spaces, high school debate club never died. It just turned inward on itself and evolved progressively weirder core beliefs, ethics, and lore. Roko’s basilisk is not a thing normal people spend a lot of time on. But for people in exactly the right little fucked up orbit, it’s their meet-cute.
In one weird corner of this weird world are the e/acc folks. If this is your first encounter with effective accelerationism, we envy you and also we’re sorry. But understanding the current moment starts here. The e/acc folks are an offshoot of the effective altruism movement, which has too much of it’s own debate club inside-baseball to cover, but suffice it to say: the core belief of these folks is that technology will save us. That the path to global prosperity and the end of suffering and the future of the species is tech. And therefore that going hardcore maximalist on tech, to the exclusion of every other consideration, is a moral imperative.
One that, by strict coincidence, happens to massively enrich them, and maybe immiserate their critics. It’s weird how that works out.
Are you in a post-truth workplace?
We’re not saying we can read their minds. That’s actually nowhere on our must-read list. But a good model is one that makes good predictions. If your model of these CEOs is as thoughtful fiduciaries of their investors’ money, you aren’t going to predict the current nonsense. If you think they’re brilliant thought leaders out here discovering truth at the vanguard, you’ll have a lot of trouble predicting their utter lack of curiousity — outright hostility, really! — towards the places where their initiatives are failing.
But if you model them as preening terminally online dilettantes. If you ask yourself, “what would someone do who was simultaneously (a) committed to a belief system that he is a god-king, and his tech work is humanity’s greatest hope, and (b) principally concerned with impressing the other debate club kids?” You’ll predict Sam, Elon, Tobi, Mark, and a host of other screen-bleached chuckleheads every time.
Not every CEO pushing AI on their teams is in the e/acc group chat, of course. Most of them aren’t invited. Some might be pattern-matching off their role models, or maybe their board is doing that on their behalf. Some have just bought the pitch hook, line, and sinker.
Some really believe that massive productivity wins are just around the corner. Despite early research showing that LLM use might decrease critical thought, impair task completion time, and maybe harm cognition. Some think it’s going to let them save on labour. Sam says that probably those people aren’t doing real work anyway. Until, that is, you have to hire people back to fix all the slop. The people who, by the way, are saying basically what Anil said up above.
The real test for an org is not whether you ever bet wrong. The question is, when the results start to come in, what do you do now? When your primary metrics suffer. When revenue is down, or morale is down, or employee retention is down, or customer satisfaction is down, or social media sentiment is down, but AI-tool use is up — how do you process that as a leader? How do you process that as an organization? Is that a win? And if it is, what business are you actually in? Because that doesn’t sound like a business, that sounds like a cult.
Most organizations exist to do something in the world. Make medicine, or food, or underwear, or newspapers. Build robots, or accounting software, or civic infrastructure, or community gardens. If AI can help with that stuff, great. But a healthy organization measures itself through its contact surface with the world and its people. It’s generally unhealthy when leaders lose sight of that, whatever edgelord philosophy might have taken its place.
If you’re in one of those, the question to ask is whether you can bring those primary metrics, those what are we here to do measures, back into the conversation or not. If you can, that’s probably the most important thing you can do for your organization right now. If you can’t, well, there’s a way cults generally go. You just have to believe harder.
— Melissa & Johnathan
