The irony of using an AI generated image for this post…
AI imagery makes any article look cheaper in my view, I am more inclined to “judge the book by its cover”.
Why would you slap something so lazy on top of a piece of writing you (assuming it isn’t also written by AI) put time and effort into?
I know that it’s a meme to hate on generated images people need to understand just how much that ship has sailed.
Getting upset at generative AI is about as absurd as getting upset at CGI special effects or digital images. Both of these things were the subject of derision when they started being widely used. CGI was seen as a second rate knockoff of “real” special effects and digital images were seen as the tool of amateur photographers with their Photoshop tools acting as a crutch in place of real photography talent.
No amount of arguments film purist or nostalgia for the old days of puppets and models in movies was going to stop computer graphics and digital images capture and manipulation. Today those arguments seem so quaint and ignorant that most people are not even aware that there was even a controversy.
Digital images and computer graphics have nearly completely displaced film photography and physical model-based special effects.
Much like those technologies, generative AI isn’t going away and it’s only going to improve and become more ubiquitous.
This isn’t the hill to die on no matter how many upvoted you get.
But people still complain about CGI in film, likely for the same reason it was criticised in the past that you mention - it looks like ass, if done cheaply (today) or with early underdeveloped tech (back in the past). Similarly so, the vast majority of AI-generated images look lazy, generic (duh) and basically give me the “ick”.
Yeah, maybe they’ll get better in the future. But does that mean that we can’t complain about their ugliness (or whatever other issue we have with them) now?
people don’t like generated so bc it’s trainer on copyrighted data but if you don’t believe in copyright then it’s a tool like any other
There are thousands of different diffusion models, not all of them are trained on copyright protected work.
In addition, substantially transformative works are allowed to use content that is otherwise copy protected under the fair use doctrine.
It’s hard to argue that a model, a file containing the trained weight matrices, is in any way substantially similar to any existing copyrighted work. TL;DR: There are no pictures of Mickey Mouse in a GGUF file.
Fair use has already been upheld in the courts concerning machine learning models trained using books.
For instance, under the precedent established in Authors Guild v. HathiTrust and upheld in Authors Guild v. Google, the US Court of Appeals for the Second Circuit held that mass digitization of a large volume of in-copyright books in order to distill and reveal new information about the books was a fair use.
And, perhaps more pragmatically, the genie is already out of the bottle. The software and weights are already available and you can train and fine-tune your own models on consumer graphics cards. No court ruling or regulation will restrain every country on the globe and every country is rapidly researching and producing generative models.
The battle is already over, the ship has sailed.
Exactly!!
Thank God, you get it.This video (which was trending a while ago) explained it pretty well:
https://www.youtube.com/watch?v=pt7GtDMTd3kAnd to add to what you said, people have some huge misunderstandings about how Gen AI work. They think it somehow just copy pastes portions of the art it was trained on, and that’s it. That’s not the case AT ALL, it’s not even close to that.
AI models should be allowed to be trained on copy righted data. If they shouldn’t be allowed to do that, then humans shouldn’t be allowed to do it either. Why do we give such advice to upcoming writers and musicians and artists, to consume the kind of content that they want to create in the future? To read the kind of books that they want to write like? To listen to the kind of music that they want to create? To look at pieces of art that they want to create? Should humans ALSO be limited to only publuc domain content?? I really don’t think so.
Again, Gen AI models don’t just copy paste stuff from their training set of data. They understand what makes up that piece of data. Just like a human does.
Thankfully, reasoning models like Deepseek-R1 have started to show the average person how an AI actually reasons and thinks about things and that they don’t just spew stuff out of nowhere in the hopes that it makes some kind of sense, slapping pieces of their training data set together to write something that’s barely comprehensible. The “Think” tags in such models really helped clarify some huge misunderstandings that some people had. Although, many many people are still left who have a really messed up view of how AIs work, and they somehow speak with such confidence about these topics with no knowledge of the technical details. It drives me nuts.
this post is about programmers being replaced by ai. the writer seems ok with artists being replaced.
Or the picture is a statement for why artists shouldn’t be replaced either. Who can tell.
considering one of the other posts is about “democratizing AI” I lean towards my take.
Oh, it is for sure more likely.
I thought it was intentional AI slop
Yeah, I’m sure they left the spelling mistake in the image on purpose to get increased engagement from pedants like me. I’m sorry, it works on me.
https://defragzone.substack.com/p/run-massive-models-on-crappy-machines
the author doesn’t oppose AI, just programmers being replaced for it.
I work for a fortune 500 company.
just recently lost a principal engineer that built an entire platform over the last four years.
just before they left I noticed they were using AI an awful lot. like…a lot a lot. like, “I don’t know the answer on a screen share so I’ll ask ChatGPT how to solve the problem and copy/paste it directly into the environment until it works” a lot.
they got fired for doing non-related shit.
it’s taken us three months, hundreds of hours from at least 5 other principal engineers to try to unravel this bullshit and we’re still not close.
the contributions and architecture scream AI all over it.
Point is. I’ll happily let idiots destroy the world of software because I’ll make fat bank later as a consultant fixing their bullshit.
There’s also the tribal knowledge of people who’ve worked somewhere for a few years. There’s always a few people who just know where or how a particular thing works and why it works that way. AI simply cannot replace that.
Institutional knowledge takes years to replace.
I don’t disagree with that, but there’s so many “wtf is this shit” moments that defy all logic and known practices.
like for example, six different branches of the same repo that deploy to two different environments in a phased rollout. branches 1-3 are prod, 4-6 are dev. phases go 3,1,2 for prod and 6,4,5 for dev. they are numbered as well.
also, the pipelines create a new bucket every build. so there’s over 700 S3 buckets with varying versions of the frontend…that then gets moved into…another S3 bucket with public access.
my personal favorite is the publicly accessible and non-access controlled lambdas with hard-coded lambda evocation URLs in them. lambda A has a public access evocation URL configured instead of using API Gateway. Lambda B has that evocation URL hard coded into the source that’s deployed.
there’s so much negligent work here I swear they did it on purpose.
there’s so much negligent work here I swear they did it on purpose.
Depending on the place, it’s the “work insurance” - companies would usually think twice before firing the only person who can understand the spaghetti. Now they won’t need said person to generate “working” code
Well, also if the guy was just dumping AI generated code arbitrarily into your product, that pretty significantly risks the copyright over the entire product into which the generated stuff was integrated (meaning, anyone can do whatever the fuck they want with it).
you’re not wrong. unfortunately that’s not how legal sees it.
not sure what they’re snorting, but it must be good shit.
I’m an IP attorney whose been pretty specialized in ML-enabled technologies for a decade now, and have worked in-house for fortune 500 companies so I’m pretty familiar with how these queries are often handled, especially at multinats. There honestly probably isn’t someone in your legal with all three of seniority, understanding and keeping up with the legal nuances, and understanding of the underlying technologies. The overlap in my experience is few and far between.
That’s what I expect if I’m fired and rehired: at least +25% on my salary.
We hired a junior at work from a prestigious university. He uses ChatGPT all the time but denies it. I know that because all his comments in the code are written like some new Tolkien book. Last time I checked his code, I told him it had something like 20 bugs and told him how to fix that because I’m not a bad guy. The next day, he came back with a program that was very very different. Not knowing how to apply my fixes, he used another prompt and the whole thing was different with new bugs. I told my boss I was not wasting time on that shit again.
A reason I didn’t see listed: they are just asking for competition. Yes by all means get rid of your most talented people who know how your business is run.
And can reproduce the whole business in a weekend with the help of AI. There are no moats anymore.
This is prophetic and yet as clear as day to anyone who has actually had to rely on their own code for anything.
I have lately focused all of my tech learning efforts and home lab experiments on cloud-less approaches. Sure the cloud is a good idea for scalable high traffic websites, but it sure also seems to enable police state surveillance and extreme vendor lock-in.
It’s really just a focus on fundamentals. But all those cool virtualization technologies that enable ‘cloud’ are super handy in a local system too. Rolling back container snapshots on specific services while leaving the general system unimpacted is useful anywhere.
But it is all on hardware I control. Apropos of the article, the pendulum will swing back toward more focus on local infrastructure. Cloud won’t go away, but more people are realizing that it also means someone else owns your data/your business.
I think they were suckered in also by the supposed lower cost of running services, which, as it happens, isn’t lower at all and in fact is more expensive. But you laid off the Datacenter staff so. Pay up, suckers.
Neat toolsets though.
The cloud provides incredible flexibility, scale, and reliability. It is expensive to have 3+ data centers with a datacenter staff. If the data center was such a great deal for the many 9s of reliability provided by the cloud, company’s would be shifting back in mass at this point
Oh no way. It was a year(s)-long process to get to the cloud, then the devs got hooked on all the toys AWS was giving them and got strapped in even further. They couldn’t get out now if they wanted to. Not without huge expense and re-writing a bunch of stuff. No CTO is going die on that hill.
They jumped in the cloud for the same reason they jumped into AI - massive hype. Only the cloud worked. And now % of the profits are all Amazon’s. No app store needed. MuwAHhahahAhahahaa
What most people forget is that as a programmer/designer/etc, your job is to take what your client/customer tells you they want, listen to them, then try to give them what they ACTUALLY NEED, which is something that I think needs to be highlighted. Most people making requests to programmers, don’t really even know what they want, or why they want it. They had some meeting and people decided that, ‘Yes we need the program to do X!’ without realizing that what they are asking for won’t actually get them the result they want.
AI will be great at giving people exactly what they ask for…but that doesn’t mean its what they actually needed…
Great points. Also:
… AI will be great at giving people exactly what they ask for …
Honestly, I’m not even sure about this. With hallucinations and increasingly complex prompts that it fails to handle, it’s just as likely to regurgitate crap. I don’t even know if AI will get to a better state before all of this dev-firing starts to backfire and sour most company’s want to even touch AI for most development.
Humans talk with humans and do their best to come up with solutions. AI takes prompts and looks at historical human datasets to try and determine what a human would do. It’s bound to run into something novel eventually, especially if there aren’t more datasets to pull in because human-generated development solutions become scarce.
AI will never not-require a human to hand hold it. Because AI can never know what’s true.
Because it doesn’t “know” anything. It only has ratios of usage maps between connected entities we call “words”.
Sure, you can run it and hope for the best. But that will fail sooner or later.
that stuff should really get worked out in the agile process as the customer reacts to each phase of the project.
Getting the real requirements nailed down from the start is critical, not just doing the work the customer asked for. Otherwise, you get 6 months into a project and realize you must scrap 90% of the completed work; the requirements from the get-go were bad. The customer never fundamentally understood the problem and you never bothered to ask. Everyone is mad and you lost a repeat customer.
yeah but with agile they should be checking the product out when its a barely working poc to determine if the basic idea is what they expect and as it advances they should be seeing each stage. Youll never get the proper requirements by second guessing what they say.
Also, LLM doesn’t usually have memory or experience. It’s the first page of Google search every time you put in your tokens. A forever trainee that would never leave that stage in their career.
Human’s abilities like pattern recognition, intuition, acummulation of proven knowledge in combination makes us become more and more effective at finding the right solution to anything.
The LLM bubble can’t replace it and also actively hurts it as people get distanced from actual knowledge by the code door of LLM. They learn how to formulate their requests instead of learning how to do stuff they actually need. This outsourcing makes sense when you need a cookie recipe once a year, it doesn’t when you work in a bakery. What makes the doug behave each way? You don’t need to ask so you wouldn’t know.
And the difference between asking like Lemmy and asking a chatbot is the ultimative convincing manner in which it tells you things, while forums, Q&A boards, blogs handled by people usually have some of these humane qualities behind replies and also an option for someone else to throw a bag of dicks at the suggestion of formating your system partition or turning stuff off and on.
Yesterday the test team asked me for 3 new features to help them. I thought about it for a few minutes and understood that these features are all incompatible. You can get one and only one. Good luck finding an AI that understands this.
I haven’t seen anybody point this out yet. The owners of tech were never in it for the “tech”. It’s just a tool for them to wiggle their way up to the top. Trying to hit the jackpot so that they can wrest control of society from the current “old rich”.
Although I agree, I think AI code generation is the follow up mistake. The original mistake was to offshore coding to fire qualified engineers.
Not all of offshore is terrible, that’d be a dumb generalization, but there are some terrible ones out there. A few of our clients that opted to offshore are being drowned is absolute trash code. Given that we always have to clean it up anyway, I can see the use-case for AI instead of that shop.
I think the core takeaway is your shouldn’t outsource core capabilities. If the code is that critical to your bottomline, pay for quality (which usually means no contractors - local or not).
If you outsource to other developers or AI it means most likely they will care less and/or someone else can just as easily come along and do it too.
The core takeaway is that except for a few instances the executives still don’t understand jack shit and when a smooth talking huckster dazzles them with ridiculous magic to make them super rich they all follow them to the poke.
Judges and Executives understand nothing about computers in 2025. that’s the fucked up part. AI is just how we’re doing it this time.
Companies that are incompetently led will fail and companies that integrate new AI tools in a productive and useful manner will succeed.
Worrying about AI replacing coders is pointless. Anyone who writes code for a living understands the limitations that these models have. It isn’t going to replace humans for quite a long time.
Language models are hitting some hard limitations and were unlikely to see improvements continue at the same pace.
Transformers, Mixture of Experts and some training efficiency breakthroughs all happened around the same time which gave the impression of an AI explosion but the current models are essentially taking advantage of everything and we’re seeing pretty strong diminishing returns on larger training sets.
So language models, absent a new revolutionary breakthrough, are largely as good as they’re going to get for the foreseeable future.
They’re not replacing software engineers, at best they’re slightly more advanced syntax checkers/LSPs. They may help with junior developer level tasks like refactoring or debugging… but they’re not designing applications.
…shouldn’t outsource core capabilities.
This right here.
I wonder if there will eventually be a real Butlerian Jihad
The jihad starts with the tech bros’ butlers, that’d be very poetic
Find one named Iain and radicalize him! So it would be the Butler Iain Jihad.
Maybe after Herbert’s idiot son dies and someone else gets the rights
His books are so shit and totally miss the point of Frank Herbert’s books.
Literally anybody who thought about the idea for more than ten seconds already realized this a long time ago; apparently this blog post needed to be written for the people who didn’t do even that…
You underestimate the dumbassery of Pencil-Pushers in tech companies (& also how genuinely sub-human they can be)
To your point at my last company party i got drunk and kept complimenting people by calling them human.
MBAs are like surgeons; their every solution is to cut.
Stop giving me bad ideas lol
e: I suck at markdown
I was a mf’ing hard core rider of the tech boom, was a sought-after consultant, and I and my colleagues rode the razor’s edge of what was possible in online gaming for 2 decades… and I can tell you now, AI presents to creative individuals who have a clue, the greatest opportunity ever handed to them. Look at how AI destroys things and “invent” solutions and you’ll pay yourself well.
Now more than ever a “programmer” is a guy that can plug other people’s modules together and pray it works. Notice that now and git gud at what you do.
You spend all your work day in meetings bragging about yourself while never actually doing any work, aren’t you?
I’ve been retired for a decade, moron
Look at how AI destroys things and “invent” solutions and you’ll pay yourself well.
Yeah, I’m seeing the absolute deluge of AI shovelware games. I know it generates money due to sheer volume, but to me that’s just like all those online courses of “how to dropship”. You’re being one of the worst literal definitions of “waste of resources”.
None of you can hear. You’re all so afraid. There is OPPORTUNITY EVERYWHERE but you’re so locked into your script there’s no talking to any of you. It’s so sad to see you limit yourselves. But in a way it’s revelatory of the truth I’m speaking… the “i’m a porgammer” because ya downloaded other people’s work is over, and the path is open to those ready to work and innovate. Good luck, but you don’t need that because you’ve already decided you’ve lost.
I don’t think the point you’re trying to make is nearly as clear as you think it is.
I love how you spent so much effort to say literally fuck all, instead of presenting a real use case that isn’t pumping out AI slop
I just hope people won’t go back to these abusive jobs. The oligarchy that runs the US has shown it is more than happy to lay people off to cool wages and the Fed is more than happy to blame workers getting paid a reasonable amount as the cause of inflation.
It’s hard for people who haven’t experienced the loss of experts to understand. Not a programmer but I worked in aerospace engineering for 35 years. The drive to transfer value to execs and other stakeholders by reducing the cost of those who literally make that value always ends costing more.
It’s utterly bizarre. The customers lose out by receiving an inferior product at the same cost. The workers lose out by having their employment terminated. And even the company loses out by having its reputation squandered. The only people who gain are the executives and the ownership.
This is absolutely by design. The corporate raider playbook is well-read. See: Sears, Fluke, DeWalt, Boeing, HP, Intel, Anker, any company purchased by Vista (RIP Smartsheet, we barely knew ye), and so on. Find a brand with an excellent reputation, gut it, strip mine that goodwill, abandon the husk on a golden parachute, and make sure to not be the one holding the bag.
What happened to ankor?
They were acquired by Opta Group in 2023. Since then, the quality has declined while prices increased. And around the time of their acquisition, they started doing some shady stuff when claiming USB-IF compliance. The cables were blatantly not USB-IF compliant.
Another example: I personally love my Anker GaN Prime power bricks and 737. Unfortunately, among my friends and peers, I am the exception. The Prime chargers are known for incorrectly reading cable eMarkers and then failing to deliver the correct power. This has so far been an issue for me twice, but was able to be worked around.
those executives act like parasites. They bring no value and just leech the life from the companies.
executives act like parasites
WE MAED TEH PROFITZ!!!1!!1
which is ironical since without them the profits would likely soar. Doing bad shit 101 is to pin the consequences of your actions on others and falsely claim any benefits others have managed to do as your own achievements.
IMO without execs, employees would get paid for a greater percentage of their labor and profits would go down.
what profits? the ones that end up in the pockets of the executives?
<cough>Boeing<cough>
Everyone. But Boeing did a pretty fucked up job of it.
Well, yeah, but those costs are for tomorrow’s executive to figure out, we need those profits NOW
Executives think they are the most important part of the company. They are high level managers, that is all.
I’d argue the CEO is the most important person, usually. We see dipshits like Musk and turn around and bag on all of them.
Think of a business, doesn’t matter if it’s local or national. How do the employees act? Are they happy and seem to be doing useful work? Are they downcast and depressed looking?
Sometimes it’s the local manager staving off corporate bullshit, but company culture mostly rolls down from the CEO. They saying, “Shit rolls downhill.”, works both ways.
I’m just a dabbler at coding and even i can see getting rid of programmers and relying to ai for it will lead to disaster. Ai is useful, but only for smallest scraps of code because anything bigger will get too muddled. For me, it liked to come up with its own stupid ideas and then insist on getting stuck on those so i had to constantly reset the conversation. But i managed to have it make useful little function that i couldnt have thought up myself as it used some complex mathematical things.
Also relying on it is quick way to kind of get things done but without understanding at all how things work. Eventually this will lead to such horrible and unsecure code that no one can fix or maintain. Though maybe its good thing eventually since it will bring those shitty companies to ruin. any leadership in those companies should be noted down now though, so they cant pretend later to not have had anything to do with it.
Even if I ask AI for how to do a process it will frequently respond with answers for the wrong version, even though I gave the version, parameters that don’t work, hand waving answers that are useless, etc.
I find it’s the most useful when asking it for small snippets of code or dealing with boilerplate stuff. Anything more complicated usually results in something broken.
give it a few more years
Yeah, like bitcoins, NFTs, and Tesla driving themselves.
Bitcoins are there, aren’t they?
yeah, there are many things its easier to just give up having the ai do it. even if you somehow succeed it will likely be such mess it gives you its not worth it
AI mostly seems useful when you don’t know a specific concept and just need the base ideas. That said, given it’s often confidently wrong and doesn’t involve humans actively steering you toward better ideas, I still find Stack Overflow more helpful. Sometimes the answer to your problem is to stop doing what you are trying to do and attack the problem from a different angle.
I also find it is best when I’m very specific and give as many identifiers as possible. App name, version, OS, quoted error code, etc.
yeah, though for questions people on stackoverflow would consider stupid or beneath them, ai was superior as it never gets angry with you. How i see the ai is kind of like being able to ask the knowledge it consists of questions. But since it has also been fed with garbage and its all there like ingredients of a soup, who knows what affects what so you really dont want to trust it too much.
As a software engineer, I’m perfectly happy waiting around until they have to re-hire all of us at consulting rates because their tech stacks are falling the fuck apart <3
The jabronigrammers before me seem to have made a fine mess without the aid of an AI tool as it is…
Article about bad AI decisions
Thumbnail is AI
Lmao
