It was supposed to cure cancer and other diseases, not make us homeless and surveil us.
“It” is very ill defined. The AI they are selling you was never supposed to cure cancer. It was meant to make money. The AI which cures cancers is of a completely different kind than LLMs and media generation. And it does not have any of the same drawbacks nor any of the same funding despite it’s far greater usefulness.
The AI which cures cancers is architecturally pretty similar to LLMs in that it’s based on transformers and GPT in particular is also based on transformers (that’s the T in GPT). It is also generative AI, you give it tokens that describe proteins and it generates folding structures. It’s like an LLM except the tokens aren’t words or syllables like in LLMs.
It’s used very differently and the energy cost is a few orders of magnitude less (you can run it on a single H200 with 141 GB VRAM for ~9000 tokens or an A100 or H100 for ~5000 according to this university and it even has a slow CPU/IO heavy first stage preparing everything so they can minimize the GPU time in stage 2).
There is yet hope that such AI applications will eventually benefit from all the research that’s been put into LLMs. Obviously not everything will carry over, but there may be architectural improvements to be made in research models too.
Fair
Funny thing, right now federal funding is hyper focused on AI because the people who bought Trump expect a return on their money. The things is, the people on the federal side more or less fall into two camps- political appointees who know nothing about anything and old guards who know that “AI” has been around for decades in various forms. This is creating fun situations where a funding opportunity might as well read “please find a way for fElon to make money using AI”, but they never specify “LLMs” so submitting a neural network or machine learning proposal fits the guidelines. Time will tell how that plays out.
Yeah, and any “it will cure cancer” usually boils down to “uhmm, if we feed it enough training material, it will turn into a sentient superintelligence, which will cure cancer”.
It’s a tool. Blame the humans who use this tool to make you homeless and surveil you.
Leave the orphan crushing machine alone!
Lol I kid. You’re right though: how we use it matters. How we build it matters. I wish it wasn’t destroying the environment a little bit more each month
Or better yet: Use the tool for one’s own goals.
This is exactly my thoughts when I’m trying to say “misused” to people who don’t seem to understand the sheer resources that would require to solve certain problems which would be a perfect fit. But it’s just not being used for those(not to the extent it should be supported). instead, it’s being funnelled to suck every value out of living a fulfilling life.
It’s like plastic. Certain medical procedures would be unthinkable without it. But then capitalism happens and it becomes beyond reproach real fast.
This is the important difference between defining something as ‘evil’ and using something that is truly innocuous and neutral with evil intentions.
We always think of not working as something negative, we should desire not having to work and machines giving us the opportunity to do so. But the system is scuffed and capitalism works against us rather than for us.
We need a new system where it’s the goal to automate everything and then we only have to work a few hours a week and spend the rest with our loved ones, getting good at a hobby, enjoying life, singing dancing creating art all the fun stuff.
I’d be fine with that, as long as there was some academic aspect involved.
A bunch of idiots doing nothing for a long time isn’t going to accomplish much of anything.
You frame it with ambition as a virtue to be desired, but that’s really not a given… ambition is often super toxic. I can imagine a future of luxury automated space communism™ where the ultimate purpose is fulfillment. To be Zen is to release ones self from ego and put desire and ambition in context.
The toxic aspect generally comes from the consequences of pursuing ambition by undermining others in a competitive context. If, hypothetically, the competitive pressure is off, then “ambition” is not nearly as likely to have negative consequences.
Here the person was saying we really need to actually put our minds to use, and that’s not really likely to be the ‘toxic ambition’ you fear.
And even worse, one of the thing they desperately try to automate is the arts and the humanities…
How about instead of automating everything to use machines that require electricity, maintenance and resources that wreck the planet, we make it so people worldwide only need to work four days out of the week at most while still being able to afford hobbies, holidays, and luxury items such as food that isnt poison.
Well, there’s livelihood and work, and so far no one has exactly inspired confidence that the current labor class will receive a livelihood. We’ve let systems solidify that basically demand that the labor class have some semblance of leverage to participate in the benefits. So far the practical result has been that the hypothetical scenario erases that leverage, while avoiding any concessions that may win over the people that would be screwed over. It’s nice to hand wave and say “some new system”, but without more concrete things to look at, there is the risk of “well, let’s remove the labor leverage and just assume that the utopian outcome results instead of the dystopian”.
One thing that concerns me is that a common short story held up as an example of awesome utopia brought on by post-scarity due to AI is “Manna”. The plight of the dystopian half is very plausible, but the “utopian” half has a lot of problems in mechanism of how it happened and issues when “realized”. If even in fiction we can’t make a convincing narrative that lands in an unambiguous utopia, then my hopes for reality is pretty low.
So I guess in short, I agree with your sentiment, but too pessimistic to see the upside being realized and afraid that the optimisim around the upside results in rationalizing the downsides.
I implore young people to get into ai resistant fields and push for a return to basics and inefficiency.
I’m almost 49 and my exit plan from my corporate job is to have a small farm just big enough that it can be run simply with man power alone and then market to a premium.
Humans are not meant to be machines. Inefficiency needs to be valued more than it is.
Have you ever grown a tomato successfully and eaten it?
Nothing beats a garden tomato. It really is hard to eat store bought produce once you have grown it yourself and it is so much better tasting.
Well yes, but I was thinking more about fantasy vs reality.
Reverse centaur.
Increasingly, they’re trying to get it to design the building.
I’m having to double-check credentials in my plan reviews because contractors have started turning in plans with AI-generated architect and engineering stamps because they don’t want to pay a professional to prepare their plans.
How can they use AI for an engineering review? Wouldn’t an actual licensed engineer have to sign off on the building?
I work for the city. Contractors are faking stamps to pretend there’s an engineer to get a permit.
Lots of municipalities are overloaded and they’ll just accept a stamp and approve the project. I do not. We review the engineering.
can you fine those who fake official stamps?
Not really. We can refuse to issue the permits.
If it were to slip by us and the house fail, even then it’s difficult. Almost everything in my city is custom spec homes for the mega-rich, and the developers set up an LLC just for building the one house and dissolve the company as soon as the house sells.
What I’ve learned doing development review for houses being sold to billionaires is that contractors will try to do cheap, shitty, unsafe work no matter how rich the client is.
Fascinating. Insane, but it makes sense in today’s economy. Thanks for sharing
It’s one of the big issues that capitalists continually overlook. In most states starting an LLC is nothing. My state it’s like $800 and a 1 page form plus getting a tax id (free and another 1 page form). There’s tons of shitty contractors that will start an LLC, do garbage work and get trashed online, then just fold it and start a new one every 6-12 months. Then if you need warranty work or whatever it’s like “oh sorry, company went under”. But the illusion of the small business being viable means this loophole will never be patched
You can and should check licenses and not hire unlicensed workers, but even that is almost impossible now as many contractors are licensed but never do any real work, they just come out, assess your issue, and then hire subcontractors for a couple hundred bucks per day. Your $30,000 roof replacement is like $1-2k in wages to workers, 8-10k in materials, and $20k to line the pockets of the fat shit “running the show” so he can keep up on the lease payments for his 4 million dollar house and 110k ford f150 that’s never seen a day of work
interesting. It’s so weird they can get away with manufacturing counterfeit documents
What they used to get in trouble for was using an actual engineer’s seal without permission. We’d tell that engineer and they’d go nuclear on the developer. So now they fake a stamp to keep from getting sued.
No one wants to pay anymore
Part of the problem is corporations have squeezed so much money out of all our systems that there is none left unless you start cutting corners. So that is what people will do.
All jobs like this are going to pay minimum wage once the rest of the workers are replaced with AI and there are lines of people willing to work for no money.
If only the construction trades could develop some kind organization to protect members from race to the bottom wages and also promote training that improves work quality.
I think we need bolder strategy. We could develop a 2 team approach where one team is made of up people who make one dollar amount and the other team is made up of everyone with less than that dollar amount. I think we will be much better distributed results.
I’m not expecting to see unions continuing to work let alone remain legal.
yeah, trades have pretty strong unions in my area. only, well my experience with the trades unions have put me in a position to see lots of corruption. it’s not been a great view.
See, that’s why strikes work. They don’t have to be legal.
As AI improves it will program hand like robotics or tool attachments to be able to do this work. This is in the pipeline. They already have workers in China, Indonesia, Vietnam and others wearing cams to capture what they do with their hands so it can be consumed and replicated by AI. It’s only a matter of time. Then there is the replication of an issue we already had. Everyone is going into trades now same as everyone went into colleges before. Long term it will lead to surplus workers in that given field. Best thing to do is be agile and willing to learn and adapt as things shift. Link: https://deepmind.google/blog/gemini-robotics-2-brings-whole-body-intelligence-to-robots/
In the 1950s, Isaac Asimov predicted that eventually, nearly everything on Earth would be automated, leading to mass unemployment. He then predicted that the “excess population” would move to other star systems, which served 3 purposes: reducing unemployment on earth, providing a meaningful existence for the extrasolar colonists, and spreading humanity throughout the galaxy.
I’m down
The funny thing is it can’t do white collar jobs either.
Construction will be 1000% automated in the future. It’s already happening slowly.
it will not be made by LLMs, because it’s upmost important buildings are built exactly to spec. You will want a proper list of instructions for the machine, not something a model comes up with on the fly.
Important to the people who want to use it, sure. Not so important to the people that just want to sell it.
its not that important, plenty of countries dont give a fuck about building code
Regulations, just like health and safety laws, are written in blood.
Some countries allow deregulation AHEM
Some of you may die but thats a sacrific im willing to make
The challenge is that the current approaches that are very novel broadly have proven relatively less useful for things like robotic manual labor.
The crux of the current AI craze is machine learning, which needs all the sensory information the human perceives that is relevant to the task and all of the things the human does in response in every detail. You can’t tell a model to “just feel out how that screwdriver is going and make sure you don’t strip the head or cross-thread it by mistake”, it has no abstraction, it can only derive the most likely inputs from training data based on the sensory information.
This works reasonably well for vehicle operation, where the human generally only has sight and hearing, and generally only operates a few relatively simple controls, wheels, pedals, knobs. Even with that advantage, it needs a mountain of sample data. Millions of miles of drive data to work with and the result is still a bit janky, kind of like a human with about 3000 miles of total lifetime driving experience.
Once you want to extend this approach to anything involving a more comprehensive use of hand dexterity and everything you can feel, then it becomes a problem. If you can imagine yourself tele-operating a robot with VR headset and controllers to do a task, then they have a path to get training data. That’s how a lot of these ‘chore-bot’ ambitions are trying to be realized, workers tele-operating for demos and training data, in hopes of getting enough training data to do things competently one day. But imagine trying to use a manual screw driver in that tele-operation scenario, making sure it’s lined up, doesn’t cross thread, and you don’t strip the head. I couldn’t imagine doing something even that simple in the real world without just being able to feel the screwdriver in my hand, to feel how the screw is going in or not going in, if the resistance feels right and consistent. If it’s a situation that makes sense to use a driver, then probably could tele-operate it, but in cases where we need the finesse of a manual tool, those are going to be super challenging to address. Even using a driver, I always manually start it, either with my fingers or a screwdriver, and the driver comes in when alignment is well assured.
Now the long standing approaches are of course already a big factor in automating construction. The new approaches can also make it easier to take a task that was almost within the reach of robotic choreogrpahy but foiled by something like unexpected weight and wind, so the relatively newer AI approaches help make it easy to correct for deviations to provide a steady environment to execute the programmed movements. But the versalitility of a person with full use of their hands and feelings is going to be a tough road unless some breakthrough happens.
I was at enough construction sites to know it would be quite trivial to automate at least laying bricks with some programmable robots. Less skilled workers are already taught on the construction sites how it’s done, to help out masons. Same with roof tiles, you only need to look out for chimneys and edges.
And in our hellscape, it will just lead to forcing out a large portion of workers, make them work even more for even less, then a second round of layoffs can go.
Bricklaying robots are already here: https://m.youtube.com/watch?v=5ZHRbnK1jt4
…this assumes that no one will ever connect LLMs to machinery, or generate code to run machinery (gcode, for example)?
I’ve seen some AI designed STLS. They are…not good
Yep, CAD and house building are still human jobs, for now.
Physical systems are a lot more expensive to train. It’s questionable if they can safely finish buildings before some better new form of AI is invented.
Also to some extent, impossible to train. If a person needs to feel the situation or it demands the full dexterity of the hand, then that is not going to be captured in training data. It’s why the robots are being tele-operated, yes in part to demonstrate the mechanical bits work, but the tele-operation is the only way they can get training data. If you couldn’t use a VR controller to do a task successfully, then they can’t even capture the training data needed.
Just gonna plug it into the house building machine?
Maybe
Fuck yeah
Ah don’t worry. With robots we’re getting there.
By the time AI is capable of finishing buildings it will choose to eradicate us instead.
The construction industry is one of the most heavily automated and mechanized industries in existence. We just don’t think about things like “back hoe”, “crane”, “nail gun” or “concrete plant” as forms of automation because they have been around for so long, but they have all replaced more manual aspects of construction processes.
Well… Automation usually describes process automation and you’ve just listed tools. So yeah, I agree, people don’t think of them as automation. Because, colloquially, they’re not.
The process of producing lumber is largely automated.
The process of producing concrete is largely automated.
The process of producing drywall, mud, paint, nails, tools, wires, switches, pipes, are all largely automated.
The process to create the architectural designs and blueprints for the house are largely automated.
The process to create the electricity, fuels and vehicles needed are all largely automated.
To make a house, people used to draw plans by hand (if at all), chop the trees down with an axe and/or saw that was made by hand by a blacksmith, shape the timber with hand tools made by the blacksmith, fasten things together with hand-crafted joinery or nails hand-made by the blacksmith. All the materials were transported by hand or by horse and lifted onto the structure by hand or manually operated hoist. Every part of it, including the production of all the tools, was previously done by hand.
The invention of the assembly line opened up the possibility to eventually automate virtually everything except the final assembly, and we have robots that can be made capable of doing that autonomously as well. You can buy pre-fab homes where all the major sections are built in a factory by robots and shipped out to the construction site and attached together.
Construction is far more automated than what you make it out to be.
So when you hit assembly line I was with you. "That* is automation
I mean i could have just as easily described the process. Cutting a trench is a process, most of which was automated by the back hoe, conveyor, dump truck, and laser levels. Before, 20 or 30 people would be required for trench work which is now often done by a crew of two.
Right but whilst the thing is easier with better tools, and indeed the process has changed, to me ‘automation’ of a task refers to its process being reduced to, at best, a daily 30s review. Tools are… Ways of optimising and changing a process. Automation obviates them. It’s a personally lexical stand, you can sanely differ.
And further automation will continue. We probably don’t even need humanoid robots. For instance, a robotic ironworker, instead of a humanoid robot, could be a module bolted on to a back hoe or other hoist.
