Dude donated a shit ton of money to slop Linux (omarchy). What did he expect?
Well, yeah, the problem with LLMs in a professional context is exactly that the second order effects of using them are so bad that they wipe out the direct effect gains (the time saving in creating the text).
It’s fine for entertainment if all you want is to be distracted for a while with some somewhat bland chewing-gum for the brain.
CEO Tobias Lütke… told employees last year that using AI is “a baseline expectation,” and they should first prove they “cannot get what they want done using AI” before asking for more resources.
Isn’t this an example of Play stupid games, win stupid prizes?
It looks like an example of Play stupid games, win stupid prizes.
Just playing uhm angels advocate? Not sure how to call it, the naive optimist perhaps!
Not necessarily: the statement (in isolation) is a call to learn and become capable to explain limitations. The criticized effect is basically “people not being able to explain the limits”.
I suspect that “finding limits” wasn’t incentived though (and I mean things like positive feedback, not money). And then you get shit-in-shit-out factories.
Or to rephrase your statement: If you want to play a game then stick to the rules you’ve set up - otherwise everyone’s allowed to cheat.
how else is ai slop going to be monetized? more slop more hallucinations more bubble.
Lmao I am living this right now at work 🥲
Yep. We have a guy on the team who generates document after document for us to review and it’s all slop that he has no understanding of (not that anyone could understand the mostly nonsense)
Did you not get to reject all of that yet?
This is happening everywhere. My job is about 2 months behind schedule at the moment and theyre still pushing for AI solutions
“SHIP MORE CODE!!! NO NOT LIKE THAT!!!” - CEOs
More lines = better
max-line-width: 20
Ok then why did you give money to your friend DHH for Omarchy? it’s literally a slop vibe coded “distro”.
Because of the complete vibe coding of Omarchy it takes longer the configure it than any other distro with a WM. I mean hell the AI cut the Hyperland config up into 6 damn LUA files AND THEN you have to unbind everything and rebind it as opposed to ONE file that you can literally just swap out binds. And yet you gave the guy a couple million for it.
So I’m so sorry Tobias but you did this to yourself. you were one of the first to embrace AI completely and use it completely wrong and now you’re whining about it and you continue to fund it. you can’t have it both ways bud.
Has anyone noticed that when a teammate uses AI, their teammates also have to as well? Honestly it might be because they don’t want to bother reading the machine’s code when the original author didn’t put much thought into it. Another might be the mental fatigue from reading code that grows faster than you can parse.
He raises exactly that point in the interview when talking about slop grenades.
I prefer the term “Meat Proxy”, but I get that it’s more of an ad hominem. I use both terms.
I spent half my day dealing with a slop grenade.
Someone came in and dropped a (bunch of) AI generated PRs. Said person knows nothing about my world, but they want cost optimization and Fable runs through logs and gives them code. Second Fable review agent tries the code, it doesn’t work. Gives suggestions to first. New commit comes in with workaround, I decline this because it’s a shit and had another unnoticed critical integration bug. This goes on a few more times.
It comes back a fifth or so time with something that is now technically correct, but fucking horrifying. Great, now I don’t know what to do, because I’m getting told it’ll fix the billing issue and this is a huge problem, but the code is insane.
This is a cost optimization PR, advertising massive savings, but the numbers seemed way off. It turns out it looked at one day with a runaway process that racked up insane billing (90% of year-to-date total project cost) and “annualized” it, and that was the savings. But further, it calculated all the costs using the wrong billing system, so all the numbers were wrong.
When I recalculated and corrected it, the whole thing saves us $6/month, because we don’t have runaway outages very often.
This is on top of every ticket I get is Claude generated, they never just state the fucking issue, and 100% of tickets in 2 weeks plus today I’ve received were inaccurate and wrong.
I hate the slop. I hate the non technical users throwing the slop on me without knowing how it works.
Take whatever prompt you put in the AI and write exactly that for the ticket content, then spend 30 seconds summarizing it into a title. I don’t care if it’s “customer wants a button that shows a cat”, just write that.
I cannot take another “client has a load bearing contract dispute with sales. Genuinely the verdict here will secure the shape of the contract. Critically in foo.py function get_foo_blorp there is no reference to felis catus to satisfy hard customer requirement. Acceptance requirement: return felis catus when module controller clicked and client is logged in and client’s birthday is in same month, otherwise return error codes 412 404 526” word salad ticket that doesn’t just state the problem.
I think a good way around this is that any project submissions using AI must include draft material to show how much data was used vs generated. We will never get away from AI at this stage but we need to get to a trust but verify stage. Even if they built the whole thing with AI they should be able to show the amount of tokens used in the entire to generate it (more generally means they have spent time refining whatever it is)
We need to get away from AI=Bad and rather use AI as tools and don’t help it replace us by showing 100% generated work as your own. Me using AI to make something is going to be much more polished and refined vs your general office worker. We need proper AI classes to help show people how to use it responsibly and collaboratively so we can cut down on the amount of duplicate work being generated as if we can cut down processing power it also cuts down the pollution generated.
Wasn’t Shopify one of the first companies to go all in on an AI infrastructure? I seemed to recall reading that about a year or two ago.
Yep. Same guy said managers will ask why they can’t use AI for that job. It seems that he has answered that hypothetical question here.
Great, can we now abort this asinine “AI” bullshit and go back to normal GPU prices?
No, we’re going to continue on this path until we have to suck the liquids out of your body to cool the data center
Not a fan of the guy but I’m stealing “slop grenade.”
I’m not a fan of a lot of his politics but I honestly thought that interview was worth listening to. He has some interesting nuanced takes.
dw he stole much more
The “purpose” for AI (the reason it is being FORCED on us) is that it allows “wealth” to access “skill” while at the same time keeping skilled people away from wealth.
NOW what billionaire would not want that?
What poor person would want that?
See the problem yet?
Defenestrate Mark
