His argument was literally that
- artists, writers, and programmers are “labor aristocracy” and deserve no respect because they’re sitting around making art and writing code while other people are toiling and laboring,
- genAI is a “free resource” (“it’s free for me to use at work”) that “helps everyone make things” instead of having to rely on skill and expertise and therefore makes everything equal,
- it’s hypocritical of people to be worried about the environmental and societal impact of genAI because “only white people are worried about it now that their hobbies are being threatened and everyone else is catching up”.
Pointing out there there are artists and programmers from every race and economic class, as well as perfectly ordinary people in every walk of life who are all actively being harmed both by the theft of ideas and labor and by the physical effects of gigantic data centers ruining communities, was met with much hand-waving and incredibly impenetrable philosophical jargon.
I’m particularly incensed because I’m as leftist and anticapitalist as they come and this felt like the fuckery some conservative cockwomble would try to come up with to parody socialism, except this guy was completely serious.
I do wonder if some of this anti ai sentiment is a psyop to keep the proletariat off the tools and away from ai while the bourgeoisie learn to use them to consolidate power.
And with this recent news about Nvidia inflating the price of their chips by paying/guaranteeing neoclouds to take them makes me even more sus
The last computer revolution had to be distributed because the Internet wasn’t good enough yet, but this time they can be ahead of the curve gobbling up all the hardware so they can decide what you can do with these models.
I imagine another world where openai stayed open and Nvidia stayed a consumer company and everyone just has their own little box that can order their shopping or whatever
At the end of the day, these models are compressing all knowledge into something much more accessible, and knowledge is power, so it would be a smart gambit to limit access and spread fear…
That’s the thing though: they aren’t. Information has a fundamental maximum level of compression.
In most cases what they’ve captured is the ability to sound plausible, and confident.
If you’re being conspiratorial, it’s worth remembering that they’re much better at being persuasive than accurate. You can tell them anything is the truth, and they’ll confidently repeat it, and automatically fill in the most plausible sounding explanation.
I can ask it how relativity works and it will tell me, I can ask it to explain something differently and it will, I can ask it to eli12 and it will dumb it down. Of course there is a maximum compression, that doesn’t make what these learn not useful.
And you can train it to do whatever you want, be confident, be persuasive, be accurate, nothing inherent about the model makes it go one way or another. And if you glue them into another system containing verified facts (database or a wiki or something) then you can very easily get very high accuracy
And we are literally 5 years in since chatgpt dropped and these things have gone from “can finish the line of the function your working on” to “can do basically anything on a computer code wise” …METR stopped doing their horizon benchmark because it got too expensive to develop longer horizon tasks and frontier models have saturated what they already have
I think this is something anti ai crowd miss a lot too, these models are currently as bad as they’ll ever be, all the issues people have are just bugs that can and will be fixed … Or left in depending on the wants of the wannabe overlords running the thing
You’re pointedly wondering if anti AI is a psyop to keep people from using it while we actively have people using it to push agendas. Simpler of two conspiracy theories.
So how do you know it’s explanation of relativity is correct? With a library, you have a notion that there was review. With the traditional Internet research, you have multiple sources. With an LLM you have its unwavering confidence, and when you catch it being wrong a conciliation “you’re totally right to push back on that”.
The models are inherently biased to emulate the patterns in their training set. That leads to communicating like people do in the training set, which if they’re including factual information means “stating facts” as a demeanor, regardless of accuracy.
My program to climb a ladder to the moon is going great. Every day I’m buying new ladders, climbing higher, putting the new ladder on top of taller and taller buildings and my massive stack of ladders. My progress has been consistent, steady, and relentlessly forward, so there’s no reason to think I’ll ever hit a roadblock.
Training these models is incredibly expensive and it’s far from clear that it’s a financially viable venture.
The overt data processing requirements sets limits on what the minimum infrastructure requirements are.
The big names in the field are aggressively unprofitable.
There’s this thing called an AI winter. After a big cycle of hype around AI, people get sick of it and others feel let down or disillusioned by the reality vs the implied benefits. Then the field collapses back to researchers and it repeats in 20-30 years.
Anti ai being psyop and people using it to push agendas are not mutually exclusive
People are being paid large amounts of money to build these datasets, and the builders are incentivised for the training set to emulate real data, our OP is a clean example of this
For your point on my relativity comment, Llms are just as capable at producing referenced text as a human. And if the knowledge representation is air tight then hallucination is detectable. Eg llms solved fermats last theorem solved in lean.
Idk how ai winter fits here, did you check out the METR long horizon benchmark I mentioned. The bottleneck now is finding tasks it can’t solve.
If you build a ladder to the stars, you don’t have to achieve that goal to land on the moon.
Imagine yourself in 2020…what tasks would you have said was unachievable by a computer in that time? How many of those tasks are trivial in 2026? Fermats last theorem? proof against np hardness of 3 sat?
Anyway guess my point is, don’t kid yourself, these things are powerful, ignoring them and pretending their capabilities haven’t grown since 2022 is naive. And if we let oligarchs run this shit then the next century is gonna be fucking grim. Knowledge is power…arm up
That’s a part of it, but the proletariat are useful idiots, too. They can take the Anti-AI sentiment to get them to back “anti-AI” representatives and senators.
Those legislators will support AI “regulation” written by-and-for the top-companies in the industry.
This will make it incredibly cost-prohibitive for anybody to enter the AI industry new, and cements the top companies at the top, while the smaller companies run around spending every last cent they’ve got to meet the regulations…and when they fail, they will be scooped up by the big cats.
Don’t be surprised if you start seeing the big guys supporting (or, at least, not disagreeing with) policies that initially seem counter to their interests. It’s a long game thing.
That’s my prediction, at least…based upon, well, nearly every other industry that everybody hates.
Luckily China is slightly on the backfoot here and so it’s in their best interest to make all this stuff as public as possible
My hope is that (and I think somehat grounded based on engram and MOE) research will congeal towards fully modular networks that can run in a couple gigs of vram and connect to some neural memory store where all the knowledge is encoded…then the weights only have to learn algorithms to manipulate data, rather than learning the entirety of human knowledge in a single weight set…kinda like how computers went from running in warehouses to lounge rooms from the 70s to now