I was a programmer for 25 years. I always figured I could have made a contribution to AI research, but the opportunity never came up (I was too busy writing Skype clones for cable companies and ISPs that were never used by anyone). I regretted this for a long time, but not any more. I never imagined AI would be this fucking bad, “bad” in the sense of bad for humanity and just plain bad as a product.
It’s cause it’s not AI
FWIW I always felt my contribution would have been towards “real” AI rather than the current pseudo-AI stuff that is all the rage. In addition to being a programmer, I also have a background in evolutionary biology and neuroanatomy. The “neural networks” that are part of current AI were originally based on the understanding of how neurons function that existed in the 1970s, and they were (and are still) a poor implementation of that level of knowledge. IMHO true AI can be achieved by more accurate simulation of how real neurons and real neural networks function and how they have come about in the real world through hundreds of millions of years of evolution by natural selection.
That being said, “real” AI is probably going to be even worse for humanity than this current stuff is, so at least I’m glad that blood won’t be on my hands.
Please stop using this argument. Unless you also brought it up every single time the term was used in videogames etc
It’s a bad argument that relies on pure language semantics that helps no one and undermines any good argument against it because it weakens your stance
no. no because in video games it was understood that the “ai: very hard” was not actually ai, it was computer player. it did actually use some very early research in the field, but it was not ai and even at age ten I knew that. ai would not have so many dumb pathing problems. nobody was trying to tell me the AI from unreal:tournament or starcraft: brood war had feelings or was going to conquer the world, so clarification was not required.
but it was not ai
It’s as much AI as when they made the chess computer that beat Kasparov, it’s just not a very sophisticated AI.
its not ai. its not on the cutting edge of the field of researching ai. its not progress. it does not get us closer. it does not tell us anything about ourselves or information science we didn’t learn from satoshi nakamoto or Adolph Hitler years earlier.
If we fix the name, it will not change what it does, and it probably won’t fix how people use and feel about it. “Really Good Guessing Machine” is less marketable than “Artificial Intelligence”, and much closer to the truth, but CEOs would be willing to take credit for laying off 20% of their workforce because they have RGGM now.
Maybe “Predictive Token Gamble”, to convey that all it does is predict tokens, and each output, each use case, and the entire industry as a whole is a gamble… “PTG reduced our human capital expenditure by 20%” still goes a long way, though.
You are spot on, and the ones downvoting you are simply wrong.
it’s funny because they are literally proving my point. But oh well
I do bring it up every case and even if I didn’t this case it’s an important distinction. Artificial intelligence is a marketing term used to conjure images of Jarvis and what not in people heads.
Video games companies never claimed to have near sentiment AGI baked into their games that would replace all human workers or blow up the world.
The term artificial intelligence is used to anthropomorphize LLMs as weird as that seems.
It is an important distinction. These are large language models, not reasoning models, not AGI, not artificial life not artificial intelligence.
I know that term as been used by the machine learning community in the past to mean things like decision trees and general models, and it didn’t really matter much then but it matters A LOT now.
Words have meanings. This isn’t arguing semantics, it’s using words as they are intended so we can communicate effectively about something in the world. To argue otherwise is naive, dishonest, and quite honestly just annoying.
Hi I’m new to this debate and I agree with you that language matters and all that stuff you said at the end. That being said, I’m firmly in the camp of “LLMs are bad, but a ‘true’ AI would be even worse for us”. I could type up several paragraphs diving into the several reasons why I think that way, but to save time (and to get back to my job) I’m just going to steal this Frank Herbert quote:
Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them
Do I think AI, in a vacuum, is always bad and will be no matter who or what makes it? No, not necessarily. But whatever we build will be inherently, and intrinsically tied to the micro- and macro-systems of our society, which are all DEEPLY fucked. Consequently, any AI we bring in would also, even if not initially, end up as being equally, deeply, fucked and end up being far, FAR more detrimental than positive
we can argue what actual ai would be, as long as we agree that it remains very firmly in the realm of science fucking fiction, just this side of inertial dampener fields and FTL travel.
I do agree that any ai made by this society as it exists would be apocalyptic. all its stories about ‘evil ai’ are just metaphors for itself. we have already given up our thinking to computers (which don’t yet think) and corporations and kings and markets. we fucking love giving up our thinking. it’s like the whole appeal of fascism for most fascists.
LLMs are not AI. AI is a real thing and it’s in pong. LLMs are not anything like the AI in pong. By any definition of AI, a rock could be intelligent. If you interpret it as doing something that resembles intelligence, that is intelligence as we understand it. It wouldn’t likely be widely accepted, but the going definition of intelligence accepts that broadness.
they literally are. AI is a very broad term. They are just not AGI.
People need to learn that categories exist and have grouping hirarchies. You’re proving my point wonderfully
Exactly. My farts are AI. I agree!
It most definitely is AI, and the guy is a Nobel Price winner for making a breakthrough in AI programming.
Contrary to you, he actually knows what he is talking about.so there is a field of ai. it’s a real thing. we don’t have it yet, but, like cold fusion, we’re pretty sure its possible and the steps along the way are fascinating. even partial steps have wide reaching applications and philosophical implications. it is the absolute coolest shit, and gets exactly 0 funding these days.
the thing all the data centers RAM water electricity and economy are backing right now is not ai. it’s what happens when a magic trick and a ponzi scheme fuck on a pile of cocaine in a Pharaoh’s tomb. I say that to the audience here, because there’s really no point talking to someone who’s still a partisan the other way.
We most definitely have AI, on the other hand cold fusion is NOT possible, you are wrong on both counts.
Oh look its the physics knower. What do you mean cold fusion isn’t possible? Did you come up with the theory of everything and fail to mention it to the rest of us? We DONT KNOW what is possible. Science is not even 5% of the way towards figuring out the entire universe. Such a definite answer is in no way verifiable. You could say it’s not possible for us to do it right now. That’s true. But saying it’s entirely not possible is about as backed by evidence as your typical religious claim.
we also have warp travel the warhammer kind not the star trek kind.
i wonder how he ended up on “might”?
You’ll find in general that smart people very rarely speak in absolutes on things such as this. Because the more you know about something the more you realize how much you don’t know.
We don’t know the end result yet.
You’ve got to have a little wiggle room. Just in case.
I ‘might’ not eat a tub of ice cream later on and then sit on the toilet crying until my legs go numb, buuuuuut…
This is why I wanted a TV in the bathroom…
You don’t use it, you lose it. And many people are already bankrupt in the old brain department as is…
I haven’t written a line of code since 2019. I have an old iOS project that I want to revive, but I genuinely wonder if I still have the ability to do anything with it. Fortunately, my motivation has tanked even harder than my programming skills.
bet you’re still better than me! take a shot! need drugs to help?
need drugs to help?
Probably so. I also haven’t smoked weed since 2021, and I never coded sober.
too late, 3 years of layoffs, and more layoffs whent he bubble burst likely wont bring it back before 2023.
We’ve extincted plenty of other species. Eventually we’d start on ourselves.
Start? We’ve been embracing the Great Filter like a long lost friend.
I’d have said that killing ourselves because we ran out of everything else is a very human thing to do, but it’s not like we haven’t been killing each other all along too.
We existed without ai quite fucking recently, I see no benefit that outweighs the cost
excuse me, but you just don’t understand the potential for deep fake ai vr porn!
Maybe if we keep the AI but loss the Oligarchy…
what ai!? there is no ai!
down with killing the oligarchs though.
We would still have to deal with the skill atrophy of AI use.
Even if AI is entirely green, publicly owned and widely accessible, it will negatively affect us.
Skill atrophy is an issue with nearly every technology though. Unless you want to go back to knapping your own hand axes and starting fires with sticks you will be dependent on some technology to cover the skills you don’t have.
What if AI was only used for skill building?
what if twinkies were only used for critical nutrition? then they would be healthy!
It has potential, though I wonder if that’s genuinely better than studying the old fashioned way.
It can definitely feel more dynamic, I’m not convinced the outcomes are provably better.
I imagine something like a tutor. You have an interactive system for 1 on 1 attention that teachers and parents can’t always provide.
Nothing I’ve seen is nearly ready for this, but at the same time, its not completely useless.
This seems like a particularly expensive and resource hungry solution when we could just train more teachers and tutors.
and create better educational resources
Maybe the pace of development would’ve been slower, giving us more time to react to it. Or it wouldn’t get shoved down our throats every second, or it’d be renoved from the places it’s more harmful, if making money wasn’t the only concern
Maybe yeah. There’s certainly ways it could have played out better.
We would still have to deal with the skill atrophy of AI use.
I don’t think you need to worry about skill atrophy if you recognize the skills as necessary and useful.
We still teach basic arithmetic in elementary school, even though we have calculators.
Even if AI is entirely green, publicly owned and widely accessible, it will negatively affect us.
Tools to rapidly parse, sort, aggregate, and distribute information aren’t going to affect us any more negatively today than they did 40 years ago or 400 years ago.
We’re going to end up with civil engineers who don’t know how to calculate dynamic weight ratings, because they used Claude to get through their physics classes.
Skill atrophy is absolutely a concern, even if we continue teaching the fundamentals. Because without that higher level knowledge, there is nobody to check for hallucinations.
The desktop calculator is an amazing tool for people who do a lot of math, but those people still know how to do the math. AI is fundamentally changing that, because it is directing users on how to do the work, not simply taking in
{a}+{b}and returning{c}. It is a user going “hey chatbot how do I find the difference between {a} and {b}?” Then the AI confidently (and incorrectly) goes “Oh all we need to do is{a}+{b}to give us the result{c}.” But if the user already doesn’t know how to find a difference, they won’t notice that the AI is incorrect.And that is obviously a small example, but apply it to larger problems like calculating thermal expansion requirements for a suspended walkway. Calculating maximum wattage ratings for a circuit. Calculating maximum weight ratings for a highway overpass or bridge. Calculating medicine dosage based on a patient’s body weight and an interactive drug.
Any one of these could lead to disaster if done incorrectly. And the issue with AI is that it will happily hold your hand through the entire process, while confidently being incorrect the whole time.
We’re going to end up with civil engineers who don’t know how to calculate dynamic weight ratings
I haven’t seen any actual evidence of this.
I have, not in education (I have relatives who work in higher education and they claim as such, but both of these are anecdotes anyway) but the interview process. My work has struggled to hire anyone and its driving me bananas. They either get candidates who used AI to pass the non-technical phone screen and show up for the first interview clearly asking AI about industry standards and they flunk almost immediatly. The second group are full on tripple threats, they can sing, dance and act, but balk at the first talk of pay (my business daddy has not kept up with the competition in a narrow field, they can get a lot more across the street).
We are looking down the barrel of the last cohorts of classically trained professionals working through their careers, the world is about to be a much weirder and worse place because of [gestures broadly to everything] but mostly AI.
We teach arithmetic, but are people better or worse at it than before calculators were widely available?
I’m not sure what you mean by “if you recognise the skill as necessary or useful” - but taking a leap, I don’t see how that matters.
If I allow a machine to code for me, and don’t regularly use my own skills, when that machine becomes financially inaccessible, or just breaks down, it doesn’t matter how important I think the skill is - I’ve given it away.
No question the average person is worse at arithmetic since the invention of the calculators. But we probably wouldn’t have our current standard of living if we hadn’t invented the calculator either. And calculators wre so ubiquitous and basic…
So… On balance it was a net benefit. On the other hand AI affects more than a single skill. Will the QoL improvements outweigh the negative behavioural effects? I think the jury is still out on that one.
Absolutely, I’m not necessarily making the argument that calculators are bad, just the point that technology can help us, and also create dependency, or at least a kind of laziness.
…said many things to Theuth in praise or blame of the various arts, which it would take too long to repeat; but when they came to the letters, “This invention, O king,” said Theuth, “will make the Egyptians wiser and will improve their memories; for it is an elixir of memory and wisdom that I have discovered.”
But Thamus replied, “Most ingenious Theuth, one man has the ability to beget arts, but the ability to judge of their usefulness or harmfulness to their users belongs to another; and now you, who are the father of letters, have been led by your affection to ascribe to them a power the opposite of that which they really possess.
“For this invention will produce forgetfulness in the minds of those who learn to use it, because they will not practice their memory. Their trust in writing, produced by external characters which are no part of themselves, will discourage the use of their own memory within them. You have invented an elixir not of memory, but of reminding; and you offer your pupils the appearance of wisdom, not true wisdom, for they will read many things without instruction and will therefore seem to know many things, when they are for the most part ignorant and hard to get along with, since they are not wise, but only appear wise.” (Phaedrus 274c-275b)
Why don’t you try getting a new argument, yours is a bit dated.
Thankfully, removing literacy from someone is cost prohibitive.
Can’t help but notice that in addition to being weirdly smug, all you did was show that someone else has come to similar conclusions.
Writing things down is a great way to offload the cognitive load of remembering things.
Fortunately, it’s hard to paywall writing. The same isn’t true for AI.
That was written when less than 10% of the population was literate because of how expensive and difficult it was to learn. There was no publicly funded literary education, and public opinion on literacy was at best contentious.
It’s not a similar conclusion. You are literally parroting the same talking points. They are just as fundamentally misguided and divorced from reality now as they were 2000 years ago.
Removed by mod
We teach arithmetic, but are people better or worse at it than before calculators were widely available?
Better. Significantly better.
I once had to explain to a room of four other servers and a bartender (so you know, people who are more likely to do mental arithmetic at work than many others), that if each of five servers tips the bartender out ten percent of their earnings, then he’ll earn half of the average tips. Several of them did not believe me even after went through a couple examples.
I’m not saying you’re wrong, but I do despair if that’s the best we can do.
He’s not really citing anything, just making a claim that happens to align with his existing argument.
People certainly don’t seem better at arithmetic, and readily rely on calculators. Which isn’t necessarily bad, but it speaks to my argument that technology can, and does, cause skills to atrophy.
Well yeah, but so are you and I’m just bringing up an anecdote. If we wanna bring numbers into this, they’ve probably got some evidence related to the Flynn effect, but I don’t know that there’s a good way to actually determine whether the average person got better at arithmetic after calculators came out, unless we want to reuse some oooooold tests.
I mean, that’s a curious anecdote. But it doesn’t explain why they would have been better at the math prior to 1970
As with all anecdotes, it does not explain itself, but I can explain this one: repetition.
I don’t think that was really in question though and both you and the other person have already responded genuinely, which makes this interaction less fun than I was intending, because I had hoped my anecdote would be taken as the joke it was.
Them’s the breaks over text sometimes.
What is even the benefit
I think there’s an issue with lumping them all together here. There is a functional difference between a machine that does your homework and is designed to endlessly praise you, and a model specifically designed to identify viruses or medications.
deleted by creator
In this case it’s basically the same technology (LLMs); you don’t get one without the other.
Sure you do. We don’t have to have every permutation of every technology.
That’s like saying you can have space exploration without ballistic missiles. Once the technology is there power and/or the market will do whatever it wants with it.
Except even moreso with language models. Here it is not just a similar technology, but the exact same vehicle.
Translation: The way scaling laws and gradient descent work for language models means that there is NO functional difference between a machine that does your homework and is designed to endlessly praise you, and a model specifically designed to identify viruses or medications.
Well, for starters, if you want to translate a written work to another language, it has never been easier.
LLMs are excellent tools for modeling large languages.
That is all they are good for. They are ok for translation, although even then they miss a lot of nuance like someone who can speak a language but isn’t fluent.
Well, there are a lot of malicious uses, but passable translation is the one positive use case.
They could replace the CEOs. Huge cost savings for stock owners.
I mean, a drunk chicken could replace a CEO.
A sociopathic drunk chicken, maybe
That is all they are good for.
Wife is a lawyer. She needs to do complex keyword searches on massive volumes of legal documentation. She absolutely loves how quickly and easily Copilot parses and tabulates info based on her queries.
Total game changer for the kind of work she’s doing.
I work in IT. Love to cobble together scripts in languages I’m otherwise unfamiliar using Claude. It’s also excellent at diagnosing system errors and providing effective remediation.
There are better AI options that could be developed for that over a massive all in one inefficient LLM like Copilot. If the same amount of money was thrown at dedicated development AI and searches it would require far less energy, pollution, hardware, and could be run locally in many cases with the same or better results.
There are better AI options
You can take that up with management. I don’t get to pick what AI agent my boss decided to pay for.
You have, and let me be clear on this, absolutely no idea what you are talking about.
Also, not all AI is LLMs
I was replying to a post about LLMs and the article is about LLMs so while it is true that AI predates LLMs and the label isn’t literally limited to only LLMs and genAI those two things have taken over the label and are what nearly everyone means when they say AI.
They aren’t talking about the AI used to find exoplanets, or the AI used with lidar to find structures in jungles, or to do actual medical science, or all of the other types that are useful tools. The article is about LLMs implemented by the big companies that are designed to steal and manipulate data in ways that are detrimental to society.
I really didn’t need a world-renowned subject matter specialist to tell me that, but I’m glad to hear him say it nonetheless.
Might?! Really after all the shit that is happening? Irrecoverable environmental and financial damages is a might?!
let’s not forget the theft and destruction of books
or whether “we fail to manage the risks and … have a huge gap between rich and poor, huge cultural problems, massive unemployment, just society being torn apart”.
Whether indeed; Who can say?
huh, if only we stopped sooner…
Gee thanks. Not sure you needed the doctorate to figure that one out.
aS iT tUrNs oUt!
