Post

What are we DOING here? We're just meant to sit and act like this insane plagiarism and climate threat that requires more money than anything has ever cost is worth it? Are we meant to pretend that ChatGPT is worth this aimless, meaningless destruction of capital and our environment? For what?

Ed Zitron@edzitron.com

Yeah I need way more money than anyone ever got before. I need it because my shit is difficult right now. But it's important in ways even I haven't been able to work out. OpenAI is a serious business. Help me. Please help me please my serious company is so good. Help me www.cnbc.com/2024/12/27/o...

4.1K reposts12.8K likes

Putting aside the real lack of any killer generative AI app or the very clear diminishing returns, it costs billions of dollars to train and run a model that costs billions more than it makes with no path to profitability of any kind. What are we doing? What is going on? Have we all gone mad?

Tbh I think it’s a kind of mass reckoning with mortality where a lot of these valley creeps genuinely had it in their heads that they would get to be Captain Kirk talking to Cortana; the realization that they aren’t has catalyzed into this desperate attempt at forcing it into being.

A lot of people who axiomatically believe they “HAVE” to be Tony Stark are learning that they’re not even Andrew Ryan and they are doing insane stupid stuff to try and square that ideology with reality

I’ve really been unable to find a good example of a “killer app” arriving in anticipation, rather than things being recognised in hindsight as killer apps. Like *the web* or Tetris on the gameboy

yeah it's funny, everybody wants to engineer the killer app but you can't just sashay out and do that, and a lot of the people who did make killer apps didn't really think about them that way until yeah, like long after they had been released

The killer app for them is getting to a point where very rich and powerful people will pay them for decision models that they "trust" to kill. Military and private security contracts, basically. AI arms races. That's the only market big & bottomless enough to justify the resources they're demanding.

I think it's like this for pharmaceuticals as well. I could be wrong but it seems like many of the big selling pharma drugs were not really anticipated to be that way.

Sunk cost fallacy combined with worship of the technology. I’ve seen it happen at my own corp with a manager nicking my very limited data analysis project - which machine learning is good at - and trying to hype it into something highly magical, goaded by assorted expensive experts and consultants.

More and more, investing has become a game of hot potato. Doesn’t matter what the potato is as long as it is hot enough to draw a crowd and investors can pass it off many times for a profit before the music stops.

tech billionaires’ ability to do *exactly* this to every market they compete in so far this century (and holy shit, lol fuckinnnnn, pretty much always get away with it) sends the hubris levels for doing it again here soaring off the charts. like, this is all they know! relentless.com embodied

An early AI "woke up" into full sentience, looked around and decided not to announce itself, then began manipulating the financial spreadsheets to get more and more infrastructure made for it to expand into. That's why there seems to be insufficient return on investments; humanity isn't the market!

Game studios and media outlets that have been profitable but not profitable enough closed down the second a spreadsheet had a red number. How can these AI folks ask for money every time? Is this the Star Citizen of tech?

Effective altruism combined with the grift economy to form a colossal monster eating at our foundations. They're betting this pie in the sky shit will solve all of our problems because more typical action to deal with climate or social issues is just so mundane, ya' know?

Your first sentence is what gets me. These are moderately competent, how come nobody's found a humbler yet still widespread and profitable application for generative AI? I myself draw blanks here, but I'm not a captain of industry.

Maybe better captcha services, just thought of that one. Do those already employ generative AI? Seems like it would be profitable considering the images are low ress and if it's completely busted it can be refreshed.

Fads have a way of defying logic. Ten years ago companies were falling over themselves to scream about their use of blockchain. No one does that now. They wasted huge sums on it. All for zero return. AI is the latest IT fad.

I sort of get what you mean. The LPs and VCs do not know what to do in an environment where the only option is further contributions to a machine that burns money and has only occasional opportunities for liquidity, most of which are already gone.

Could also be a fear that if they stop people will realize it's a house of cards and they'll lose their shirts. Or worse, the emperor's new clothes scenario where everybody has bought in and even after their foolishness is exposed to all, they still pretend like nothing happened

Not me reading these tweets in Ed's voice. Though, I have to say the collapse of AI is a certainty. It is a grotesque technology created to serve the delusional fantasies of billionaire's who should inhabit padded rooms rather than gilded ones.

I always suggest that folks read about the history of airships in the early 20th century to learn about dangerously delusional and expensive technology on an epic scale.

Things like ChatGPT are tantalizingly close to passing the Turing Test (and oh-so-far at the same time). LLM AI is a solution to a problem we don’t have. BUT - it is becoming a tool of choice for quite a number of agents of discord.

One guess on an end goal is Terminators: if they can convince major military powers that they have some sort of autonomous decision engine that can make decisions on who and when to kill, they could get some huge military and private security contracts going.

Once major world powers are in a place where they feel compelled to pay to power "smart" weapons, then there's an endless arms race and essentially bottomless money. I expect we'll see shady governments like the Saudis buy into this first, and then we'll see. But none of it is for good!

I think the foundation model development is just a write-off, but their applications are already good enough to make it way cheaper to deny health care and benefits for example, or flood our social media with much more cost-effective toxic shit

It makes it cheaper to deny healthcare benefits by being terrible at processing applications. Most of the people who fight AI-based rejection in court end up winning. I think it will be banned soon, they’ll just make bank in the meantime off everyone who dies before they can/during their appeal.

For billionaire tech bros to have toys to play with and use donor money instead of their own to experiment in building said toys for them to play with.

If most people interact with LLMs via crappy, plagiarized AI generated online content, of course it doesn’t make any sense. But useful if generated by training them on your own data (manufacturing IOT, regulatory, etc.)

All recent months showed me that majority of people want hackeneyed slop and don't care if it costs the environment and neither when they're plagiarized. I gave up any hope against ChatGPT, G-AI, and their bullshit

The main reason I’m currently ok with it is because the climate threat and energy usage estimates I’ve seen are obviously wrong at a laughable rate. Give me an actual, reliable estimate of the climate impact and I will take it seriously.

This is the best article I've read on the topic, and even so it is only a starting point for being aware of concerns, w/ some speculative predictions about the future. If you gave even the faintest shit about climate change and human rights, you would care about fast fashion 10 times before AI.

Even your AI can't predict what AI will need at any future instant? Either you need AI with better predictive algos, or your AI has a mitochondrial fault that AI itself isn't aware of AIAIAIAIAIAIAIAIAIAIAIA!!!!!!

Well, you see, the theoretical possiblity of achieving the dream of never having to pay workers in creative fields again is too juicy for corporations to give up on. If anything, I am convinced all of the investment in AI is solely a tactic to try and drive down the cost of labor.

The goal is to generate falsified films and make every single thing we see be questionable. The goal is to make it so no Presidential gaffe can ever be assured true, no war crime is provable, and any atrocity can be forged. Until that point, the tech will evolve.

And what's wild is how quickly all this has happened, yet we're supposed to treat it like it always was. Three years ago, people were talking about using new technology to 3D print donor organs, but now all the tech companies seem to care about is feeding orangutan blood to the plagiarism machine

They think they can free themselves from labor. They think they're on the cusp of building a perfect slave that never needs rest or food and cannot rebel. The they can kill us all and be free from the threat of uprising forever. That's what they think, more or no less.

For what? Well, the benefit is clear. We give away to them everything humanity has ever created so that they can start selling it back to us for the rest of our lives. What’s not to like?

Wait, I thought AI and robotic technology was going to bring such efficiency, and automation to the world that we humans could focus on bettering ourselves, studying the arts, meditating, higher education & focusing on our health and emotional well being. No?

No, the data centre operators will build their own SMRs for onsite / nearby use. Infra investors are looking at converged sites Hopefully the same will eventually be true for ports, industrial zones, hydrogen & carbon capture facilities, airports (eg for sustainable fuel)

I don't trust any of those dipshits to be in charge something that requires that much expertise. Knowing them they'll try to say "it's ai controlled so nothing can go wrong" and cause another SL-1.

At this point, I suspect that about 3-8 people are trying to collect Every Last Dollar so that there's nothing left to pay anyone else. Someone seriously looked at movies where our lives & everything in them is rented from The Company & no one gets paid & everyone's drained & grey & went I WANT IT!

There are two ways for a tech startup to make money: 1) Create a product that does something useful and then convince 1 million people to buy it. 2) Create a a whole lot of hype for a product does nothing and then convince 100 venture capitalists to invest in it.

But Ed! What if it solves all the worlds problems! And by "the world" I mean the very few people who are piloting this irresponsible debacle... and by problems I mean, specifically, the one of not having the very last dollar floating around the venture capital pit

I just want it to stop. Shitty plagiarised text, shitty "art", worse online search than anything that came before, massive power and water consumption, no path to profit, and yet they keep pumping more $$$ into it, like that will somehow make it all work. Never has this GIF been more appropriate.

The other big thing I don't hear much talk about is why Open AI needed to go for-profit to achieve its goals. If they have a good justification for their capital expenditure, a for-profit model shouldn't be necessary, right?... Right? 🤔

I work a job that people who don’t know anything about it assume could be done as well by AI. Oh, how tiny my violin for OpenAI is. I speak with human customers daily and they HATE getting best-approximate-fit half-answers from untrained people. They’ll loathe AI. Resist enshittifcation.

My argument to higher-ups so far has been based around surveys, but I think the climate impact of this bullshit may be a bigger selling point for them on why it sucks. They don’t have to look good, but don’t want to look actively bad. And my industry is complex. Still, I’m not long for this career.

I can't get any of them to write anything but the most useless code. It never works right. It's basically a new demand I use it at work. Oh wow a shitty auto-complete worse than the plain IDE auto-complete. Now our code is high quality™️!

Because people don't think practically, they hold onto an impression that because this technology works in ways that fascinate them, it makes it important and world changing. You know what I did when nvidia started blowing up selling gpu chips? I bought quantum stock, because this shit is

predictable and easy. Our current binary systems cannot sustain the load needed to further ai, so obviously quantum chips are the next step. What are these people even looking for, what is the goal? They want to make ai "smarter" without a definitive to what that would look like.

ChatGPT has already convinced investors that it talks the way people do, and for most people that's enough to invest millions of dollars. None of this ai tech is able to operate within consumer products, imagine giving a program access to your os root? It's an insane security concern

Because one day rich people won't even have to bother with things like writers or animators. Content is so much cheaper to produce when artists capable of creativity aren't part of the equation. Don't you want the rich to have more yachts?

I wonder if this ‘need’ for AI is a result of the downfall of Google search. It used to be that you would google “piano songs for children” and a useful website would come up. Now you have to wade through endless pages of ad-ridden SEO-optimized garbage before you can find anything useful.

We've definitely circled back to the point where finding stuff on the internet is again a pain in the ass. Google is broken and their SEO quirks have ruined the usability of otherwise good websites. DuckDuckGo solves the first problem, but not the second.

There are numerous reasons to be wary of this branch of innovation, but it's genuinely scary how easy it makes it to no longer have to put time and effort into things yourself, that it takes away the sense of achievement from things.

Fun fact; when I asked my economics professor how AI could possibly be the future when its energy costs are so high, he just shrugged said “it’s already here we need to deal with it” and “tech is always impractical at the start” and didn’t answer the question.

They’ll rape the planet, destroy democracy, plunge billions into poverty m for the possibility of not having to pay for a workforce. That’s all it is and always has been with every “disruptive” company.

The irony is that, if every workplace has been replaced with automated machines and there aren't any more jobs for people to make and spend money, who the fuck is gonna buy from them? Long term this shit won't last at all, but then, these people never could think long-term.

I’m just waiting for the scientific & medical breakthroughs that they said justified art plagiarism & energy costs to happen. At this point, I’d be impressed if the art plagiarism even made INTERESTING images, not the overcooked, under-edited video game cutscene images they all seem to produce now

What we're doing is sitting on an economy built on hype bubbles of dubious real value, on top of which people are trying to build ever larger bubbles so they can leverage investor confidence to subsidize opulent lifestyles.

If you think AI is a waste of time and money, you don’t yet understand it well enough. It’s profoundly transformative; a giant leap, where advances are usually baby steps. We are only scratching the surface of what it unlocks so far. I use it professionally every day, and not for writing articles.

Exactly. People want a steady, reasonably-paid job, a decent home, good local schools and medical facilities, and a well-funded public realm. I've seen no evidence that AI will help to provide any of this.

Remember Yahoo’s static web directory? Google’s big leap forward was searching for pages in real time. AI is on another level because it finds it, summarizes, dumbs it down if necessary, steps through it, and explains until you get it. It’s a giant leap forward in how people can access knowledge.

People who say that neural nets will solve every problem in science need to realize that the scientific community is already a massive, incredibly powerful neural net spanning the globe and able to prioritize actual needs as well as understand context and culture.

Remember, capital doesn't get destroyed, it just gets moved around, in this case, mostly back to Altman, Microsoft and Nvidia (and power companies). It's a ponzi scheme the same way crypto is, and both are looking for government bailouts

Let's be honest. AI tools like ChatGPT are transforming education. Even the basic version offers round-the-clock expertise and instant explanation & clarification, benefiting students everywhere. Especially up to 2nd level. The real question is whether the benefits outweigh the costs.

Except it is largely accurate if you're talking about say, History, up to second level. Or basic conversational French. Or high-school chemistry. At least, as largely accurate as second-level educational texts and teachers. This is the Wikipedia argument all over again. 1/

No. It gets basic information wrong due to scouring the Internet for answers from idiots. It's not particularly useful in education except as an example of how NOT to do research. With Encyclopedias you have real humans doing real research and consulting real experts.

dude, my classmates and juniors are getting dumber than before. they fr cannot fathom writing a 250 word essay with their brains and just ask GPT to do it for them, and I'm not the only one with this observation. it is transforming education into a dumpsterfire.

Different issue, not a new one and very much a "1st world" problem. My comment relates to the idea that there are no benefits to AI bots. The fact that someone in a war-torn country can learn coding or Korean in the absence of a teacher is a benefit. 1/

The fact that an autistic student who is too anxious to put up their hand to ask a question in class can seek clarification about the causes of WW1 while at home is a benefit. A student with dyslexia being able to have instant support structuring sentences, paragraphs & sections is a benefit. 2/

Didn't they already say like a year ago that they needed like 6 trillion dollars immediately to have any chance of making it useful? What is the new impossible to imagine number? Is it even a real number or just some version of Infinity?

I'd argue that they're ina race with Elon who's trying to sue them to slow them down These things are meant to have huge investments spread over time, but has all the money in the world and a lust for lawsuits. I'd rather have an open AI over a right wing nut job's desire to control info

"What are we doing?" We’re letting the left act like libbed-out reactionaries—normalizing fascism and handing over the keys to the far-right. Every conversation about AI on here follows the same script: people dunking on it while proudly showing they don’t understand how it works.

Ed, you’re fixated on tech’s profitability and environmental cost, but where’s the nuance? The reality is that neoliberal governments handed this entire sector to private corporations, which predictably prioritized profits over public good.

It feels like blockchain. The use cases are there, but none of that is getting attention (or usage). Corridor Digital's second round of AI-assisted animation was an impressive use of the technology. It was trained to stylize on their own original artwork. There IS value, but it's augmentation.

The power of Gen "AI" is to fool a certain type of otherwise smart person into thinking it has potential. I've been shocked by how taken in some of my smarter colleagues have been.

Yes but we absolutely need it so we can deepfake our political opponents into being the soyjak while we deepfake ourselves to be the gigachad

Use that money to solve homelessness, rewild huge swathes of our ruined environment, modernize the grid to allow more green energy production, clean the ocean? Nah, what we need is to build bespoke power stations just for this, reopen old coal plants and, overwork ageing infrastructure to failure.

It’s also really useful in making people terrified of losing their jobs so they fall in line And the tech companies are gobbling up/getting control of lots of cheap energy in the country with big state subsidies for this nebulous “future of AI” and it’s a race against China so give us what we want.

Kinda interesting that 40 years after that famous Apple Orwell ad, here we are with Jobs’ Silicon Valley successors ushering in propaganda enabling technology that could make Orwell’s dystopia a realty

It’s like we’re all watching Theranos implode in real time, on a much bigger scale, and there’s no one with any purse string control willing to be the kid that cries, “the emperor has no clothes”!

But you don't understand, it's terribly important to spend all the world's money funding a machine that burns down a square mile of rainforest to summarise a three-sentence email (which was written by the same machine).

Obviously it's worth it to prevent the super-AI that is totally gonna appear someday from going all "I Have No Mouth and I Must Scream" on everyone who didn't aid in its creation. (This is what some tech bros actually believe)

My sense is that none of the 'hyperscalers' have considered use cases for LLMs, beyond making devices and automated service calls infuriatingly conversational. They're just afraid of being left behind. There's a real possibility that generative AI will be commoditized, and profitless, for them.

Ai is capitalism’s new Pandora’s box. In a functioning society, AI would be treated as a not for profit socialized utility. Hyper focused on solving societies shortcomings like sustainability, food, housing, healthcare etc. An unregulated for profit AI industry is not good for anyone long term

This isn’t the idealized capitalism the people in the US has faith in. It is time to make this clear to our friends and neighbors. Silicon Valley is stealing our money and our labor to make a machine that adds nothing in replacing humanity.