The Gospel According to Shumer can be found here:
https://shumer.dev/something-big-is-happeningShumer begins with a deeply faulty description of the world of February 2020, which I remember vividly and not really at all like he describes. I went to DisneyLand that month, and in the leadup to the trip, I was already making jokes with a financial adviser about the coming plague while debating the wisdom of actually going on that trip. In the event, I did go on the trip, and my husband got incredibly sick. I was mysteriously exhausted and lost my sense of taste and smell. We had no idea it was covid at the time, but we definitely agreed that the hospital sounded like a terrible idea. I’d been monitoring travel cases in the US already and we’d been discussing whether there was any local transmission yet but at that point we still naively believed that there wasn’t, or wasn’t much. I also remembered the fizzle of the previous Sars scare years earlier and flu scares that had “disappointed” (happily) and was thus prone to downplay the risks. We always stock TP (costco), but when I heard people were buying out masks everywhere they could find them I was mad because I thought those should be left for the people who actually need them and hoarding can create supply shocks all by itself (beef scares of my childhood, for example, as documented by Bill Rathje’s Garbology project, which you can read about in Rubbish!).
He then describes how AI has already displaced many people in Tech, who never imagined their work could be automated. There’s substantial evidence within the text that Shumer was never more than what we would once have called a script kiddie, but is now an app developer of the most rudimentary sort. Here in Massachusetts, it’s hard to avoid stumbling into a conversation with older folks who want to reminisce about front panel programming a PDP-11, and even Facebook serves up stories of the women who hand coded stuff for the early moon missions. The Typing Pool was drained by people hunting and pecking in word processors, because we no longer needed someone who could type quickly and accurately — slow and inaccurate was Good Enough, especially with spell check. Back in my early college days, Pattis showed everyone a curve of telephone adoption that indicated clearly that everyone would have to become a telephone operator soon, and of course we all did, but via dial tone and rotary dial and later push button and still later Hey Siri Call Mom. In the meantime, there were far fewer jobs for telephone operators, and they did very different work than the women who used cords with jacks to complete circuits … literally.
Shumer is very impressed by the ability of Claude agents to write code, and on one level, it is remarkable.
https://arstechnica.com/ai/2026/02/sixteen-claude-ai-agents-working-together-created-a-new-c-compiler/But if you read the whole thing, you realize the lie embedded in calling this “clean room”. This is just more theft, renamed as “creation”. Shumer looks at enshittified stolen goods, and believes that genuinely new things that are of value can come from this. It’s possible; composting really does improve the soil. But it’s much more likely that the current round of AI will be digested in much the same way we digested calculators: rapidly, where accuracy and speed in calculation truly matters, creating an entire job skill and job category in the form of “data entry” or “10 key”, and slowly and with great resistance, in areas where speed and accuracy were competing with other cultural values (k12 education and doing the family budget).
Shumer’s But It’s Getting Better Fast is an effort to undercut the observation everyone has made: AI quality is pretty bad for many purposes, and you have to have someone quite expert already in order to rapidly sort through what’s usable from what isn’t. AI is good at increasing the productivity of the already productive; it’s not good at enabling someone who is new or not very good at a job to do it at even a consistently mediocre level.
Shumer’s advice to pay for subscriptions to Claude and similar, and advice to lawyers to feed contracts and accounts to feed tax returns into these tools ignores the fact that trade associations for both groups specifically say not to do that for reasons of confidentiality which he entirely refuses to engage with. These fields and others will develop models that have rigorous (enough) walls around them, but rushing into that has dangers he does not engage with. They are also regulated fields which will further slow their change.
Shumer also flattens competency, which requires capacity and experience, across multiple realms. “If AI is smarter than most PhDs, do you really think it can’t do most office jobs?” I mean, we have an entire stereotype around the absent minded and/or nutty professor, who can’t even get dressed properly without supervision.
I don’t necessarily disagree with the idea that if your job mostly is at a keyboard, your job is going to evolve a lot over the next few years. It evolved a lot over the last five years, so, expected! But Shumer has no real understanding of previous waves of job evolution. He writes, “When factories automated, a displaced worker could retrain as an office worker.” Of course that is not what happened. They stayed home and their wives got jobs in the service sector. “When the internet disrupted retail, workers moved into logistics or services.” Retail had a lot of logistics workers already, so, sure, they moved to another company with a slightly different orientation, doing roughly the same work (load, unload, drive stuff around, pick). But retail workers who were not already doing logistics work — the ones who were there because they had the ability to interact with randos in person answering provocative and/or insane questions for hours at a time — did not go into logistics. They just didn’t. They went somewhere else where their excellent emotional regulation and other people skills were valued. Child care, health care, hospitality, education and similar.
If Shumer really thought it worked the way he said it worked, why isn’t he advising people to pivot to health care or building trades? I think the main reason is because I don’t think Shumer knows anything at all about these activities, and he has some fuzzy-minded idea that Optimus or some other humanoform robot will be taking over those jobs Real Soon Now.
Shumer’s so bad at communicating with and understanding other people, that he doesn’t even realize that’s a skill set that is valuable. He thinks that lawyers sit around analyzing contracts and finance folk sit around creating financial models (I mean, he must not read Matt Levine, because if he did, he’d know that only the entry level folk do that shit; the higher tiers take valued clients out to dinner, right?). There are people who have jobs that involve putting together complicated legal and/or financial deals and then attempt to sell those to plausible participants (LBOs and similar), but that’s not what most lawyers and accountants are doing. Most lawyers and accounts are answering email and talking on the phone and occasionally meeting people in person to do something with an estate or a trust or a closing or file something with the court or tell people it’s not worth it to sue over something or trying to figure out whether a settlement is good enough or helping someone figure out whether they should take a plea offer or moving contentious folks through a divorce process or explaining to people what their rights and responsibilities are in some document like a lease that they signed or are considering signing. Show me the AI that is going to do any of that well. No, really. Show me.
Even when all of this is done in words, drafting an agreement, circulating it, incorporating any responses, recirculating it, and then getting it finalized and filed if needed is a remarkably tricky bit of business. Probably AI can help with this now, and in offices large enough to have paralegals, this will probably impact the paralegals work load and work flow.
”Writing and content. … The quality has reached a point where many professionals can’t distinguish AI output from human work.” Let me just say, I HOPE that an AI wrote most of this piece by Shumer. I’d hate to think he crafted this repetitive turd by hand.
”I’ve already watched people begin relying on AI for emotional support, for advice, for companionship. That trend is only going to grow.”
I actually believe this. It’s the one thing here that I find scary. And I don’t think Shumer finds it scary at all, even though there’s a lot of indications in this piece that he enjoys his time with AI a lot more than he enjoys his human time (fair!) and is now evangelizing that as a choice that has urgency, and anyone who doesn’t make that choice will suffer (not cool!).
”Eventually, robots will handle physical work too. They’re not quite there yet.” This is odd, given the massive changes occurring in logistics with semi-autonomous / supervised robots, and of course robots are everywhere in manufacturing. What physical work does he mean? Wiping butts? Setting up a central line? Adding a new electrical circuit in housing stock that was created before the Vietnam War? Replacing plumbing with lead in it with lead-free plumbing? Remediating leaded paint on old housing stock? “But “not quite there yet” in AI terms has a way of becoming “here” faster than anyone expects.” Does it? Or is Shumer’s conception of work so limited that he only sees the things that are susceptible to doing poorly without significant loss?
Under the heading, “What You Should Actually Do”, Shumer moves into Cult Recruiting. I have no difficulty spotting this, because I was raised a JW, and my mom was born into a Holdeman Mennonite community. Call it the ultimate exposure therapy. I might lose people to cults, I don’t lose me any more.
”I’m writing this because I think the single biggest advantage you can have right now is simply being early.”
It’s true for MLMs. Is AI an MLM?
“Sign up for the paid version of Claude or ChatGPT.”
He mentions only the lowest paid tier, not the highest, nor does he mention that anyone who uses the lowest paid tier rapidly escalates to the highest paid tier.
”Instead, push it into your actual work.” This one really bothers me. Cults and MLMs and similar want you to use your friends and family as resources for the cult. That’s what Shumer is doing here, too. This is where he specifically recommends that licensed professionals like lawyers and CPAs do things that their trade associations are all over the place saying do not do under any circumstances. Without acknowledging any of those trade associations concerns.
”if it even kind of works today, you can be almost certain that in six months it’ll do it near perfectly.”
Promises, promises. He goes on to create urgency “That window won’t stay open long. Once everyone figures it out, the advantage disappears”, and promise that the unbeliever will be punished: “The people who will struggle the most are the ones who refuse to engage”. He brings in prepper/goldbug thinking. And he acknowledges all the things that WILL slow adoption (right or wrong), but says any time created by resistance must be used to adapt to the AI future. Even if reality shows you his timelines and promises are all wrong, he insists he is right, and you are wasting time. No matter how long the delay.
The JWs are still trying to reinterpret their failed 1914 prophecy. Most of their other dated prophecies they’ve tried to hide.
Like all cults, Shumer wants you to indoctrinate your kids. And like all cults, every desire or fear you have will be fixed by AI. You can write that book! Cancer will be cured! In a thread pulled straight out of 1984, “The specific tools don’t matter as much as the muscle of learning new ones quickly. AI is going to keep changing, and fast.” Kind of like the alliances that you must be loyal to while they are in place and forget completely when they are gone.
Finally, the admonition to commit a substantial amount of time daily: “Here’s a simple commitment that will put you ahead of almost everyone: spend one hour a day experimenting with AI.” Simple commitment? 365 hours is over 9 40 hour work weeks. _2 months_ of a full time job. If you have kids, a job, a spouse or anyone else who cares about what you do with your time — and I hope you do! — I wonder what you are going to think about this. And I wonder what Shumer did when they pushed him to spend less time with the AI.
Shumer concludes: “If this resonated with you, share it with someone in your life who should be thinking about this. Most people won’t hear it until it’s too late.”
Gross.
Never forget that 1 hour out of 24 hours is a little over 4%. So if someone is asking for an hour a day of your life, they are asking for more than 4% of your life. I mean, pause and reflect before making that commitment.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
walkitout
