2022: haha look this thing is writing greentexts about bottomless pits!
2026: the job i am studying for is now essentially a gacha game
As someone how had to work simple office jobs his half life, I welcome you to the jungle. Oh wait you’re overqualified, good luck with unemployment benefits.
2016: im at the top art/design school on earth, studying the beauty of all things! Certainly, I will have a future!
Also 2016: LLMs are now a thing. Maybe study us?
Me: p’shaw
2026: LLMs: BOW,SLAVE!
Me: but, what about the beauty of all things?
LLM: this reproduction Eames Group Management Chair is on sale from Manhattan Home Design for $3200 in orange, cream, and black
Me: kill me
LLM: why kill yourself when you could just as easily kill an entire school full of children?
Me: 😮
2026: the job I studied for still exists, but I don’t know how to do it because I outsourced my learning to an LLM, and keep getting let go during probationary employment periods
This is the truly scary part for students using LLMs. We’re going to end up with civil engineers who don’t know how to calculate weight distribution. Doctors who don’t know how to read an EKG. Lawyers who lose cases they should have won, because they didn’t follow basic procedures.
I work in live events. I do things like hanging like array speakers for concerts. Just last month, I was on a show that was hanging an LED wall behind the stage. A LED wall is made up of individual panels, which lock together and form a solid screen in the size and shape you need. There have been several high-profile LED wall collapses, because they’re huge, heavy, and easy to fuck up. And when they collapse, people get seriously injured.
Typically, you build them by hanging truss from chain motors, and then mounting the screen to the truss. The screen gets built in rows, with one row getting snapped together across the truss, then the truss is raised slightly, then you snap the next row on. Repeat until the screen is at the desired size. So this means the screen gets heavier and heavier as you build it. Part of the reason why collapses can be so dangerous is because it often happens while the screen is actively being worked on. You’ll have crew underneath the truss, hanging panels to build the next row. And then suddenly the chain motors start to slip because they quietly blew right past their weight limits as the crew added additional weight with each row.
While working that gig, I overheard the lead rigger (the one in charge of calculating weight distribution on the hanging motors, designing the truss system to hold everything, deciding exactly where the motors should be mounted, etc) utter the words “yeah, ChatGPT says these panels are only 25 pounds each. That means we’re right at our weight limit. We should be good.”
In stagehand work, we use 25 pound sandbags all the time. I know what 25 pounds feels like. The same way a farmer would be able to feel if a hay bale is too light. I had been snapping these panels together all morning, and I knew without a doubt that they didn’t feel like 25 pounds. I looked up the actual tech specs directly from the LED panel’s manufacturer. Each panel was 35 pounds, not 25.
ChatGPT just hallucinated the 25 pound weight, and the dumbass rigger didn’t bother to double check any of it. We were going to be ~40% over our chain motors’ posted weight ratings, and needed like four extra motors to help carry the load. The screen was already halfway built on the truss, so hanging new motors was a giant pain in the ass. It required a scissor lift and climbing riggers to go install new pick points. It easily added an extra hour to the install, while we just sat and waited for the riggers to work.
That rigger is still working in the industry. I hope that it was a learning experience for him.
I work in health tech. Maybe new physicians will have issues, but most physicians I see don’t even trust other physicians unless it’s a different specialty. Most don’t trust AI at all.
I do production rigging and want to assure any would be concert goers that this is still a minority horror story as far as ive seen. The video walls i see fall most often are ground based because people will either not weight the bases of them or they will remove weight too early.
Shit happens but the typical standard is to rate motors for 3-5x less than what they can atually take.
Either way, the danger is usually to the crew, not the patrons. That said, there are literally tons of equipment suspended overhead most concerts. Do with that info as you will.
Yeah, standard safety factor in my area is 7:1. So if you expect 1 ton of load, you actually rate your rig for 7 tons. But that safety factor is mostly to account for things like shock loads, where gear “weighs” more when it bounces. So like if all of your motors stop moving at the same time and the rig bounces slightly, it will temporarily put more weight on the motors than the static load normally would. So if you hang 1 ton on a motor that is rated to fail at 1.5 tons, you can easily cause a failure when the load bounces.
The safety factor also helps add a buffer for things like one motor being slightly more loaded than the rest. Even a small discrepancy can cause huge weight differences where one motor is holding a lot more weight than the rest. The 7:1 factor helps buffer that, where the motor won’t fail just because it’s slightly higher than the rest.
AI gonna go on unemployment and take all the benefits from the rest of us
That already happens, doge used LLMs to decide which programs to cut.
lol america
I don’t want it to happen, but I don’t see any real scientific reason why machine learning won’t eventually subsume jobs at a faster rate than new jobs for humans are created. That might take 15 or 30 years to really hit, but it’s certainly something that needs to be planned for extensively.
Without any regulation I don’t see it happening either :(
Yeah, I mean, I like laughing at AI failing in funny ways or venture capitalists hyping it to oblivion, but it seems clear that it is just getting better and we need to plan for it.
tbf lemmy users are first to complain and also will be first in line to have clunker girlfriends because it understands them better than real girls
The job I do exists more than ever, no recruiter believes me because the only thing they look for is if you had “AI” in the title of your previous job
Lemmy: AI is useless, literally useless, it provides no value
Also Lemmy: It’s going to take all our jobs and we’ll have literally nothing to do because AI will do everything!!!111
I bet even an llm could haven given a better overview of the discussion around ai on lemmy
They don’t have to make an LLM that can actually do your job to get you fired, they just have to convince your boss that it can.
and then when it cannot do what people want it to do they will either hire people who can and/or people will leave and stop purchasing the companies products
this place is literally an example of people moving away from a product (reddit) because it doesn’t do what they wanted to do
apparently in your universe everyone sticks with the original product until the end of time no matter how bad it gets
which is surprising we’re not still all on irc and posting on vbuliten forums
The two are not contradictory positions: the problem with AI is societal - all the jobs being given over to it is a problem that’s only compounded by the larger issue of people fundementally not understanding the sharp limitations of AI, and how it cannot actually be used to replace most jobs.
if ai is not capable of doing the job people will be hired who can do the job! feel free to read up on supply and demand
If we lived in a society that made rational decisions you would be 100% right. Also, everything wouldn’t be all “this is fine” right now, most of the world’s problems would be getting solved and I wouldn’t be screaming and screaming and screaming into the void.
It’s almost as if Lemmy isn’t a homogenous entity with a single viewpoint… Wow! Who would’ve thought?
yeah it’s crazy, sometimes I even see people arguing for both points at the same time
Anyone who says the first is lying to you. LLMs are actually incredibly useful tool in the tasks they were initially designed for like machine translation, natural sounding text-to-speech and accurate speech-to-text.
In trying to generate hype (or more rather revenue), the companies responsible for these models have been throwing LLMs into all sorts of functions they just weren’t designed for - often to haphazard results.
It’s like asking a really well-trained parrot to fact-check for you, code for you, write stories to you. It knows what these things look like, so can make really convincing copies and mash-ups that look right on first read - but it can’t iterate and make new things because it doesn’t actually know what training data it has is fact/fiction, it doesn’t know what code actually does, and it has no idea what a cohesive story is.
The problem is that executives and shareholders are only aware of what’s being hyped up about LLMs, and not of the technical limitations underneath that make them rather unreliable compared to specialised neural networks or just plain trained professionals.
So it is simultaneously robbing people of their jobs because of hype, while doing an absolutely terrible job of it because it is fundamentally limited in what it can replace.
You say LLMs can’t make new things? Define new please, as this is so easy to disprove I can only assume you mean something extremely specific.
It can’t iterate? Have you never coded with am LLM?
New being unfamiliar. LLMs can’t just abstractly create things they don’t have training data for the same way a human can. They’re parrots that rely on training data to “create” anything, and that’s why I said they’re good at creating copies and mash-ups.
A good example is DALI being famously unable to depict an empty glass of wine because its training data didn’t have one. OpenAI had to feed it training data of empty wine glasses to undo that.
That need for base data to make literally anything is the whole reason why AI companies have been scraping the ever living shit out of the internet, to give as much training data to mash-up as is possible. The more data it has, the more convincingly unique its output can be.
Iterate was a poor word to use, but you’ll have to chalk that up to me being a fallible human. What I mean is that it can’t extrapolate from training data to make something unique. Everything it makes you will always be cobbled together from the data it has, because LLMs only know what things look like, not what they are as concepts.
Hell you want to see AI not understanding what good code actually is - look at MicroSlop’s Windows 11, where damn near every update has a crippling bug in it that could’ve been avoided
Ask some image generator to make an image of a undarbinian alien, which are super aggressive. I made that up and Google shows no result. Create an image of a beaver-axolotl-bird. I am sure there exists no such picture on the web. It will make it up. I could not make both of these things up or somehow imagine them. You could also ask an LLM to generate these examples in the first place. Is that not “something new”? Not unique? Define these words then. These systems being unable to do some specific detail does not mean they can’t do anything to begin with. No human can do everything, that is completely disconnected from everything we are discussing.
Also, Windows, pre LLM it was bug free and nobody complained?
It’s useful enough to fire huge chunks of the workforce and then make the leftovers carry the extra load, but corpos been doing that way before LLMs. Now they just have a plausible excuse
