it’s getting to the point where I notice people say it a lot, especially IRL now for whatever reason recently.
And for clarity I’m not in research or anything, so these people just mean ‘LLM/image gen’, not utilities like OCR or (usually not) transcription.
Some have argued it’s just more efficient (which I can kind of get), while others think you’re actively hindering your intelligence somehow.
On the first point:
I’ve tried it occasionally to see how it compares to my own skill, and while it produces a functional result, it’s always very derivative work to the point where you can find things with the exact same names of other ‘public’ (but not libre) works, and often isn’t the ideal solution to what it targets. So I can see how you can get things out of it, but it never felt really that profound to me.
But for the second… isn’t this supposed to be the tool for people to do things they aren’t experienced in? If anything, you probably need to be able to understand how to write pertaining to the task so the token probabilities are biased toward writing from that area.
And even then, if all you end up doing is prompting AI, then wouldn’t you ultimately serve no purpose outside of being glorified QA?
I guess I’m trying to figure out what exactly non-users would be ‘falling behind’ in that affects them more than those who use AI?
They think AI is as life changing to our society as the internet is as steam engines were long ago. They’re trying to promote its use because it doesn’t have a good specific use for day to day. If anything AI has made the internet MUCH worse, there’s so much AI Slop media on youtube and similar that I don’t look at much new stuff anymore because of it.
Ai is more like adding lead to gasoline; it does stuff we could already do ever so slightly better, but for an immense and irreversible cost.
i wouldn’t say better, a better analogy, is cheaper gasoline, with worst results than more expensive gasoline.
honestly, depending on how it affects media in the future, it could really have a similar effect on us, even if it doesn’t directly poison people. Poorly presented information/misinformation has, and will certainly continue to result in so many preventable tragedies.
It’s a desperate marketing strategy based on the classic “peer pressure” technique. “Everyone’s doing it. You don’t want to be the only one who isn’t.”
My experience of a decade in corporate US companies:
Quantity > Quality; no matter what the marketing line says. People who speak a lot in meetings and send a lot of emails are regarded as better at their job, substance is only somewhat relevant. And LLMs are great at producing lots and lots of well sounding slop.
They think there’s some skill necessary to use LLMs for some reason
im going to put it easy: You work doing some kind of research or whatever on daily basis. You used to do that task in maybe 5-6 hours, same as your colleagues. Now all your colleagues use AI and do the same task in about 20min (even if the result is crap it does have some good takes), so the CEO sees you are the only one who doesn’t perform equally because you choose not to, and you’re fired.
So yeah… it’s crap, it will make things worse, but for now you have to use it wheter you like it or not.
100% this in corpo developer jobs.
We need to pin this answer on every post about AI in the workplace.
AI is the first major technology that I have have actively worked against.
I am a person who strongly believe that technology should make us better, and I can only see AI/LLMs making us weaker.
I tend to have a vivid imagination, and all I am thinking of when seeing people use AI is, “ok but could you have accomplished the same thing without AI?”
When the AI bubble pops like the overstuffed bubble it is, and prices are going up, will you remember how to do your job without it?
Or, even simpler, what about if the AI service goes down, will you still be able to do your tasks?
For many, I suspect the answer is no.
I see it with ADAS.
It appears that the more Advanced Driver Assistance Systems a vehicle has, the less competent the driver is and the more likely it is that they will not react correctly (or at all) in an emergency situation.
ANCAP and ENCAP are realising this now and their ratings are no longer weighted so heavily on ADAS.
Yeah, this has been a long problem in government agencies here.
There was a political decision made years and years ago, that in-house expertise was too expensive and that they would hire consultants when they needed specific skills.
That works only as long as you have the in-house skills to determine what external skills are needed.
Same here, without the skills to do the work, you don’t have the skills to correctly evaluate the work of an LLM.
I tend to have a vivid imagination, and all I am thinking of when seeing people use AI is, “ok but could you have accomplished the same thing without AI?”
Sure. Protein structure is the clean example. For decades the same thing was done without AI: crystallise the protein, collect diffraction or NMR data, solve the fold. That works. It also costs years and a specialist lab per structure. AlphaFold did not invent a new kind of knowledge. It made the existing problem cheap enough that we now have predicted structures for hundreds of millions of proteins instead of a few hundred thousand experimentally solved ones.
“Could you have done it without AI?” is the wrong test. The test is whether the extra cost of doing it the old way is worth paying every time. Sometimes it is. Often it is not.
deleted by creator
People aren’t talking about alphafold when they’re talking about “AI” in 2026. They’re talking about chatgpt, claude and copilot.
“Could you have done it without AI?” is the wrong test. The test is whether the extra cost of doing it the old way is worth paying every time. Sometimes it is. Often it is not
I would hard disagree. It absolutely applies for super narrow applications like alphafold, but in general, LLMs aren’t remotely close to being worth it. You don’t need to look beyond how hugely unprofitable AI companies are. The only companies who aren’t completely shoveling money into the furnace are the ones selling hardware to the companies that do.
We also see that the consequences of AI use are pretty harsh. Students who use a lot of AI are a full year behind their counterparts that don’t, according the the latest PISA data.
So what we have is something that is unaffordably expensive, with dubious benefit and immense second order effects.
No. There’s no way I could build a full web app on my own in under a day. AI enables me to do that. You’re basing your whole opinion about AI solely on how capitalism is leveraging it. There is a lot of great potential with AI. Medical fields. Non-profit space. The list goes on.
I’ve got some bad news for you. Even with AI you still can’t make a full web app in under a day.
You can make a fun little demo for sure, but that’s a long way away from a full web app. Even without doing any coding or using the computer, simply describing a full web app using pen and paper would actually take more than a day. And when it comes to actually making such a thing, there’s so many things besides the coding and even with regards to the coding you haven’t considered.
This is a big issue with AI. It provides a shortcut to some sort of end result. Since people get a result, they tend to think they have actually achieved the desired result. But in reality most things are much more complicated and people often have entire careers specializing in one aspect or the other. There’s so much that goes on, without people not in the know realizing. Without being an expert yourself, you wouldn’t even consider those things to be important.
Let me explain using an example. I can write words on a page. I could even think of a story and give it some structure. Does that mean I can write a book? Technically yes. Would that book be any good and people want to read that book? Probably no. That’s because I know enough about writing books, that I know I don’t know shit about writing books and would not be able to without years of learning and failing. Now AI could help me vomit out words faster. More words more quickly means I can get a whole book worth of words out in no time at all. But does that mean I somehow gained the knowledge of how to write a good book? No, because I didn’t learn anything. In fact I would learn less from using the AI than I would have simply writing down the words myself. Using a tool doesn’t magically grant skills that weren’t there before.
A lot of people don’t understand this and shit out shitty AI books anyways, which then nobody likes. They shit out apps, which then nobody likes and they get abandoned almost immediately. They shit out “art”, which then nobody likes. And since they got to some kind of a result, they glaze the AI tools and say it’s amazing, even though the results were shit.
There are teams whose whole job is just to describe web apps. They are product managers and UX.
There are people who’s job it is to describe movies to the crew. They’re called directors and producers.
No bad news for me because it doesn’t pertain to me. Sorry you can’t get AI to work for you. Keep trying! I’ve used AI to build things for me that are totally useful. I’m not selling anything. It’s a tool. It can build full web apps. I’m not talking SaaS. Just simple things. And I’ve learned along the way as well. It’s all about how you use it.
You are selling something, you are trying to sell it to yourself.
Good job on absolutely completely missing the point of my comment. I’m not even mad, that’s impressive.
And they say AI doesn’t cause brainrot… “Sorry you can’t get AI to work for you. Keep trying!” sounds like the kind of thing you would expect from someone who only outputs the highest quality of work, doesn’t it?
I’m not saying all AI users are like this, and there are many smart developers that do use it effectively (like Torvalds), but there is definitely a loud group of AI cheerleaders that extremely careless and overconfident, while lacking any sense of self-criticism or oversight of what they say or do.
And by the way, you definitely can build simple web-apps in a day without AI. I was doing that in 2015 by just following the Django tutorial. It took me longer than a day because I tried to read the documentation and understand it. But if I put the same amount of effort that these low-effort AI users do, I could have easily just copy-pasted shit from the finished tutorial and been done in an hour with the same level of understanding (or lack thereof) that AI users have.
That’s the difference between good and bad engineers - you have the tools in front of you, but can you make yourself a better engineer while you use them?
Speaking of which, you know what these AI tools do? They just go to github and copy off some project that already does what you want it to do. Anyone could do that before AI existed. Sure, it’s definitely impressive that they can understand what you want to do and find the right result (usually), but you could do that ever since search engines were invented. AI is just the “I’m feeling lucky” button from Google from back when people had enough sense to be a little critical of the results they were looking at before they clicked it.
Not all, but the 99.999999…999% are, and the 0.000000…001% that doesn’t will be like that sooner than later.
That’s a good ad hominem you have there. Now please make a fucking actual argument. If you can’t, then shut the holy fuck out.
Show us what it built…
You seem to confuse speed with ability.
Ok, so you can’t do the task in the same amount of time, that is not what I meant, I am talking about being able to do the task at all.
Could you make a full web app without AI at all?
Do you have that skill?
llms are able to do a lot of things many people simply can’t (at least without spending a long time learning), although the result can be mediocre. it’s being misused widely, but doesn’t mean it’s entirely incapable of anything useful.
recently i’ve seen it being used for reverse engineering, seemed like a pretty good usecase.
It really cannot. But people who have no knowledge in a field absolutely think LLMs are proficient in that field. As soon as an expert in that field look at what the LLM is doing in their field, that illusion quickly vanishes.
Unfortunately, that’s not true. LLMs have proven new results in mathematics, at extremely high levels, conjectures open since decades.
I am personally very against this use of AI, but let’s be clear on its capacities.
Every one of those mathematical milestones made by AI is followed by an article from a mathematician pointing out the ‘discovery’ was a flawed rehash or not really applicable to the problem. Same for all the claims about finding bugs, they are old and known bugs and occasionally it can identify already known types of bugs faster as long as a human can filter out the wrong results.
AI does do a fine job of filtering large data sets in controlled environments, like how it is used for finding exoplanets, but just for speeding up an existing process.
It doesn’t do anything new.
Look at code security too.
A reference with a reasonable overview: https://arxiv.org/pdf/2608.02859
About some of those proofs…
Yeah, I had read earlier today. But even the author said that his own draft was LLM-based.
names Fremats last theorem proved by Anthropic
Uhm? The one proven in the 90s?
They solved the Navier-Stokes milennial problem just the other day.
We can talk about the AI bubble, water consumption and AI “art”, but it’s undeniably doing some incredible stuff too.
If you can’t wrap your head around something that isn’t entirely good or entirely bad then well I don’t think a discussion can be had with you.
deleted by creator
I absolutely could not build a web app on my own from scratch and at this point in my life I do not have the time to learn how.
There’s absolutely no doubt that AI democratizes many capabilities that people otherwise would not possess, whether the output is “slop” or not. Sometimes you just want to quickly prototype something. Doesn’t have to be perfect. You can do that in minutes to hours. If you like it, pay somebody to do it right, or AI may even be able to do it right depending on the task.
I think you look at a web app too narrowly, a web app could be any page that loads content from an external source.
I am not a web developer, and I built a system that logs the water temp to a database and a webpage displays it to help people know if they want to go for a swim or not.
It took a few weeks of free time to get everything to work, but it has been running for about 5-6 years continuously.
That is a web app, an application solving a need running on the web.
It was developed by me without any AI, and I have only limited experience with web development.
I think you can do more than you give yourself credit for without AI.
There’s absolutely no doubt that AI democratizes many capabilities that people otherwise would not possess,
There is so much wrong with that.
For starters, you don’t possess it. You still can’t do it because you haven’t learned.
Second, it’s not something you have inherent access too. LLMs are completely unaffordable to anyone using them, which is why the companies are running insane losses. Your use of LLMs depends entirely and fully on the availability of venture capital to burn.
Third, it was always within your ability to learn. The information to acquire the skills has been available for free.
Fourth, thanks to LLMs (and enshittification) it is becoming more and more difficult to find and access that free information.
Please send me a web app you have made without the assistance of a computer. I will wait.
I have a degree in electrical engineering, happy to decode, DM for the address.
I mean you could say the same thing about the power loom. Labor-saving technology does matter.
The issue with this comparison is that a power loom is like a computer script, a person built the logic, it just replaces physical labour.
AI/LLMs replace cognitive work while the outcome is not guaranteed.
That’s one thing it can do, but there are other, actually useful things you can have it do. For example I’ve used it as a pseudo-search engine when Google refused to cooperate/when the thing I was looking up was too complex before.
No, LLM’s CAN replace cognitive work. You don’t have to give up all your thinking to an LLM. Think about what you want. Create very specific instructions. Review the output. Challenge it. Refine. Ask questions. Make decisions. Teach it something. It’s a tool. It doesn’t have to replace your brain.
Calculators replace cognitive work. Are you anti-calculator?
AI/LLMs replace cognitive work while the outcome is not guaranteed.
Calculators… are devices designed to produce deterministic results that can be proven (and yes, with a seed AI is technically deterministic, but that doesn’t make it easily provable) as accurate very easily. If you removed calculators from a task, either the result would be completely the same, or it would be worse due to error introduced by something unpredictable.
past like really, really early education, I don’t really see a point to go without a calculator since it uses up so much time for little gain.
The issue is that most people seem to enjoy offloading their cognitive work to LLMs and take their results at face value, that is what I dislike.
Didn’t they offload it to TV, then tiktok etc before that?
It is highly unlikely, but possible, that the majority of the people you are talking to are wearing clothes they made themselves without aid of machine.
It is highly likely, that at least one person reading this DOES believe we should destroy all the factories and return to the trees.
How they manage to have that argument on the internet without imploding due to cognitive dissonance is beyond me, but I have seen more impressive feats of confusion in my days.
Well, good luck making sure your page doesn’t have huge security holes, that it performs well (and by good performance, I mean that it can run reasonably well even on a 2010 or even 2005 computer), that it handles memory properly, and that it’s not just a skeleton that works by selling your soul and destroying your device. You’ll need infinite luck for that, I assure you.
You can delude yourself all you want that the AI is enabling you to do “x”. But all it’s doing is ensuring that you can never do x and that you keep getting worse at everything related.
And science is proving it. In just a few years of usage, llms are making people stupid and insane.
The brain is like a muscle. Each time you get in a car to travel 5 miles in 6 mins instead of running there in 30 minutes you get weaker. After 10 years your body will be in a horrible shape and much closer to death.
Yes I’m against calculators too for most things. Because of avoiding them I can do small calculations faster in my mind than most people can type into a calculator.
And guess what I’m not left behind. Because of my philosophy I am stronger and sharper than most people half my age.
The recent PISA data confirms that students who use AI a lot are about a year behind everyone else. https://www.oecd.org/en/publications/pisa-2025-results-volume-i_73451bc5-en.html
I mean you probably could do a decent range of web apps in a day as a minimum viable product, if you already have a solid plan ready. But that’s kind of the thing with AI.
AI… really seems to just make derivative stuff. I tried asking it to replicate one of my terminal text editors (via a description of it), and it ended up making… essentially a 1:1 rip off of different editor but with more comments. I have a feeling that the massive amount of training data is what makes it seem like it can write lots of applicable code, while it’s still suffering from quite literally being just a larger version of this half-decade old Google Translate update under the hood.
Of course it’s derivative, but just like any tool, the user plays a huge role in the output. You just work with it. It can do lots of things and it can’t do lots of things. Am I gonna sell anything I make with it? Probably not. I use it to help myself and to prototype things really fast.
While I disagree with exactly what you’re saying, I agree with the part that AI CAN make things better. But I’ve seen too many sloppy vibe coded apps that don’t work to think that having it make an app all on it’s own is a good use of AI.
You’ll fall behind in the race to the lowest IQ as they all turn their thinking over to AI and get lazier while you continue to exercise your brain.
Fuck 'em.
its already been shown HS and college grads have even poorer reading writing skills due to it.
The people who espouse such nonsense as “if you don’t use AI you’ll fall behind” believe that using AI will become an absolute necessity in the future, and that those who use AI now will have more expertise in using AI than those who don’t, giving them better prospects. Those same people end up saying “you’re just not prompting it right” when you point out flaws in the LLM’s responses.
Prompting is obviously a skill that you can improve, but if you compare it with the skills they are trying to replace it’s a very basic skill. Going through a computer science program and a few years of work experience vs figuring out what prompts it tends to fuck up and learn to safe guard against that (as much as you can) are different leagues, different sports, different everything.
It’s like saying that if you don’t start using lemmy now, you’ll fall behind when the year off the fediverse is upon us, when you’ll just pick it up as well if need be.
Prompting reminds me that most people around me don’t know how to use Google. If I want a recipe for an apple pie, I type “apple pie recipe” (3 words). Other people would type “i want to make an apple pie and I need a recipe or something.”
And now it’s worse with chatbots because they would say “dear ChatGPT, I have been thinking of my grandma and I would loooove to make an apple pie like the one she did, it would mean a lot to me, and also the sun is shining, thanks, kiss, goodbye, have a nice day.” It’s horribly inefficient and shows that most people cannot use computers properly, and don’t understand that they are fast calculators and not magical brains.
It is even worse, since LLMs talk back like that, which probably encourages people to be like that. I had to use one to solve a Linux problem, for which I could not get help any other way, and it was just awful. I do not want the agent to spout useless phrases like a human, it just wastes so much time to read them, when you have to figure out are the answers even close to being useful. Maybe I could have prompted it to just reply more efficiently, but I tried to keep the usage minimal anyway. That discussion made me sound more like the robot. “Blahblah not working” “Oh I’m so sorry you are having this problem. Lets start with…”
Fucking awful.Right! These people are now “super human AI prompt masters”, nut my natural ability with written language from a young age and early exposure to technology has made me (and anyone with good vocabulary and grammar mechanics) better than a lot of these idiots could ever hope to become with years of practice. The thing about corporate cultism, though, is in your mind it always makes you “superior” somehow, if you can stick your head far enough up your ass. Corporatism is all about enthusiastic support of your own exploitation, and this shit is the ultimate expression. We need fucking LLMs more than clean drinking water to hear these robber baron fucks tell it.
Levergism. To get what you want in the world you need power to get power you need leverage to the keys to power.
For working class folks that tends to be in the form of withholding labor, providing labor both of which are dictated by supply and demand. So learning a specialty in an in demand field or getting/making tools or skills that make you more demand is common technique. (Organizing with others to be noted also a huge way to do this)
They are arguing that learning AI is like learning to drive a forklift or an excavator. Like sure you could use a shovel or set up pullies but it’s just faster and thus more valuable to use those tools.
The difference being we don’t actually know what tools and varients of those tools actually make sense. So it’s like saying you need to learn to use motors to a 1900s factor worker.
This is so spot on. VFD’s revolutionized manufacturing, but telling that to someone in the 1900 would have been crazy.
They mean that they have lost all ability to think for themselves.
They do it to protect their feelings. They’ve normalized the use of something most people hate, so they want everyone else to adopt it, too. That way, they won’t feel judged, when they 100% deserve to feel judged.
It’s very similar to learning how to use a computer in 1995. I’m just old enough to remember meeting these people.
There is this thing, which, supposedly, can help you do your job better, it’s being pushed by management, and it is sometimes expensive, often confusing, and obsoletes the skills you spent years to acquire.
Fast-forward 30 years, and it’s very, very, very difficult to find people to talk to who still argue computing doesn’t belong in the workplace, and that using a computer isn’t essential for virtually all work.
The reason it is difficult to find these people, is because they were left behind.
Blindly trusting google results when copy/pasting into the terminal is sketchy. Doing that with AI is even worse.
Go ahead and blindly follow the LLM, have fun.
Remember NFTs? when people were buying pictures of cartoon monkeys and people were saying, if you don’t buy it now you will be left behind
NFT was literally a foreshadowing of what was to come. You know that villain hero beats up at the start of the film but main villain is someone else?
Yeah, BUT only YOU own the gorilla with the pink cowboy hat! You just gotta understand NFTs man! /s
Ah yeah, remember when NFTs solved a millennium prize? That was crazy. Such a fad tho.
AI didn’t solve the navier stokes equation; the mathematician using an LLM did.
jfc grow up.
What mathematician?
OpenAI heard that a millennium prize was potentially about to be claimed by an Anthropic researcher and assigned it to a swarm of 10,000 autonomous agents at a cost of an estimated $15 million in inference.
We used a system of coordinating agents powered by our internal model. The agents had access to tools such as the ability to read from a cached version of the internet and the ability to run code. Agents were subdivided into groups with the ability to communicate within the group. The groups varied in size, and the group that produced the Navier–Stokes resolution involved on the order of 10,000 concurrent agents.
There isn’t a mathematician sitting there working with a LLM. OpenAI’s solution seems to have been a brute force and mostly autonomous event to try to get ahead of other “mathematicians with their LLMs.”
how fucking dumb are you that you think there wasn’t a mathematician involved in solving one of math’s greatest challenges?
“Mathematicians Tristan Buckmaster and Levent Alpöge spent about a year studying a related version of the Navier-Stokes equations with the assistance of AI. Their work showed that a solution in the friction-free version of the equations can develop a singularity in finite time, a result often described as mathematical “blow-up”.”
furthermore, OpenAI’s solution is almost certainly built upon this work.
“The most consequential part of OpenAI’s Navier-Stokes claim remains unverified. If its model has genuinely produced a valid proof of the full problem, it could represent an extraordinary moment for both mathematics and AI. But until the proof is made available for scrutiny, questions over its origins are likely to remain just as important as the claimed breakthrough itself.”
OpenAI can end the discussion by making the proof available but hasn’t.
and I’ll add:
It does not appear to be directly built upon it, and they are allegedly in the process of submitting “substantial progress” on the Hodge conjecture and setting aside the inference to try the same approach on the other prize problems.
Give it a week or two and let’s see.
If even a single other problem falls, it seems really unlikely the process by which Navier-Stokes was solved relied upon allegedly copying the work of the mathematicians involved in it, and that instead massive amounts of inference with current internal SotA models is sufficient.
You don’t suddenly solve two different problems by relying on the work of a mathematician working on only one of them.
Edit: And just to explain Tao’s comment — he’s not saying “oh, they copied the work of my colleague.” He’s saying “holy crap, if tech companies are going to be adversarial and spend millions of dollars on inference to scoop my field when they hear we are getting close, it becomes risky to share that we’re making progress on a given problem because then the lab can say to their AI ‘here’s millions in tokens, go work on X problem there’s probably a solution.’”
His quote strengthens my point that this was accomplished without human mathematicians working on it, as that’s effectively the core of his point.
If like you can’t drive a car, that makes it harder to find a good paying job.
