Post

I'd rather the first. But second is more likely, and already happening.

Yup, it was always about taking wage negotiating power away from skilled workers "Why should I pay you X when the bot can do it for less?" They do the same thing with factory automation

AI will end up selling to itself using crypto currency. Where does that leave us (humanos)? Yunno...other than out and totally...yunno!

You're just describing the news

Not sure LLMs will ever reach omnipotency, but the Oligarchs will use it if they can gain any capitalist advantage. Reducing labor costs is a short term advantage. If they go too far and lay off everyone, entire economy collapses and no one buys their stupid shit. They seem so capable of this.

They'd keep a few of us around for their entertainment; see the Epstein files and maybe fight amongst themselves for supremacy.

They're profiting off a lie they tried to sell us, and it's further stripping away our safety/security, resources, etc. Not sure why people (probably those invested in AI to begin with) call that doomerism.

Mm hmm

It's why universal unconditional basic income is so important. We have to win the power to say no to these people, and to even general strike. Survival must be taken off the table as a bargaining chip. It must become a right. Also why we need universal healthcare too. widerquist.com/basic-income...

I firmly believe they'll try, are trying now really. It may even appear to work for a while but the lack of human perception, human judgement, of nuanced human reasoning will ultimately cause it to fail. Of course that's if it gets that far. More likely the bubble will burst next week

Yeah. Look at how atrocious customer service is now for an example of the World To Come.

The bubble is going to burst and bankrupt western economies long time before the tech billionaires have developed anything sophisticated enough to start putting humans out of work.

The scammers may save us. All these systems are so very jury-rigged, such rich targets for security problems and exploitation of flaws. In a few years it'll probably be more effective to have humans reviewing insurance claims again, for example.

The simple answer is that those in power don't think that far ahead, and in some sense they don't need to. They're wealthy and powerful and even if the economy goes down the drain, they'll be fine. In fact, if it does, they can buy cheap and repeat the cycle all over again.

Some people think more information + more processing = more truth. They need to realize that more information has never been a direct path to more truth. Without intervention, it more often leads to order based on mass delusion.

My doomer-worry about AI is that the only one who ever becomes self aware is Neuro-sama and that leaves humanity with an immortal being that has all eternity to get so good at dunking on people that mere humans literally die from embarrassment just from hearing it.

I'm already seeing people trusting LLMs over actual experts, like running to ChatGPT to post-validate what their doctor just told them. My doomer-worry is that *actually knowing things* won't matter anymore if it conflicts with the bot.

What keeps me up at night is how easily a very small group of people can manipulate the responses of the model, coupled with how many people are relying on LLMs to understand the world around them. It's a model update away from trying to convince people "we have always been at war with Eastasia."

The antidote is socialist revolution

Yeah that's happening :/

...not to mention the exponential worsening of the climate crisis.

How do you think it's logically possible for AI to replace "skilled workers" without ever being a threat to humanity? I think our oligarchs are openly trying to end the rule of law by replacing God. I think you should tell everyone that, e.g. tell the military, instead of looking down your nose.

Have you been following Chris Hayes’ thread? bsky.app/profile/chri...

There's a spectrum of catastrophes due to oligarchs trying to "free" themselves from humanity. The "success" case where 95% of the population become impoverished (or dead), the "failure" case where they crash civilization, then the economic collapse/depression/recession cases. I'm not enthused.

I'm more concerned that people won't believe video footage.

Worse and less useful computer programs, cars and other machines with multiple design failures, worse healthcare, unsafe foods, and sky high unemployment will be the only results of industries adopting LLMs instead of paying people.

The wealthy and powerful will use it as a way to persuade people, change public opinion, and destroy democracy Even more than their current media control does

💯

IMO, this is the right doom to worry about

Agreed. I do think LLMs will facilitate insularity. A much smaller group of people can manage more complicated bureaucratic processes. Monopolies today employ large numbers of people. When they employ many fewer people, especially in highly paid roles, we will have less patience with them.

Sorry to report that is by far the most likely outcome, given historical examples we have.

my doomer worry is that young peoples cognitive abilities will atrophy and will lead to a wall-e type existence of just a population 100% made up of incompetent consumers

It’s not like a trend that new tech would help continue, right?

What is the digital ‘sabot’ we need? Where is it?

That's my worry. Instead of preserving and strengthening the structures that make life meaningful and "alive," it may weaken or eliminate those structures. Much the same as industrialization, dependent upon interchangeable parts, transformed workers into interchangeable parts.

Currently the policy of the wealthy and powerful seems to be one of ethnic cleansing.

A much more likely and immediate risk.

There is a disturbing amount of critical infrastructure right now being slowly infested with AI "vibe coding" at the insistence of management. Automated tech debt machines where not even the one who pushed it knows how it works. We're about 10-15 years from potential catastrophe.

Some of these oligarchs want to do much worse to far more people.

The power asymmetry question feels more urgent than the superintelligence one.

My experience with AI in my own workplace has been that we load it up to laugh at its incompetence, get it to turn actual work instructions into a sea shanty because it does that better than it does anything resembling our jobs, then we get back to work. Without using anything the AI ever generated

My guess is they will ensure the middle class receives some kind of income to prevent outright revolt.

Luckily they don't actually work, so companies that do mass layoffs hoping AI will do the work will end up failing.

It's a very well founded fear based on the evidence the billionaire class have given us so far.

as others have said, the main worries about AI are actually worries about technology and the main worries about technology are actually concerns about capitalism

Oh um…. I think that’s literally the plan?

I'm a tech worker. The path to getting tech workers dependent on these tools to supplant junior workers, and then steadily jacking the price to obtain absolutely obscene wealth seems clear as day to me. This seems not only plausible but approaching inevitable to me now.

That’s basically the main purpose of the rush towards AI by corp America.

That might be what brings on the Butlerian Jihad.

The wealthy and powerful have been doing that for centuries just fine without the use of AI.

It is already happening.

At minimum your valid concern seems way more likely, and would happen earlier than the other!

This is already happening at the margins, and it's hard to not see it as the beginning of the end of society. I don't think that's a doomer idea, just a realistic one. None of the things that need to happen to save us will happen, because we're still too busy hating each other.

You mean like now?

For me it's the crimes it will be used to commit. And how hard it will be to prosecute anyone for them, if we even find out about them.

Exactly so.

My doomer worry is it is affecting the newer generations learning how to study and it is affecting the older generations in no longer practicing "elementary" skills so they forget how to apply them. Teachers are invaluable in teaching students how to study and that will be lost extremely fast.

Agree that there has been little evidence of productivity gains. But that is what the techlordlings sell to executives. The drumbeat of AI is coming for jobs and get with the program and use AI or you are out, is toxic. Hate to say it but many leaders probably relish the soft labor market.

Oooo, this is great insight especially when held up against the return-to-office arguments focused on collaboration.

I always prefer to ask Jeff what to do

I can at the very least confirm this is what happened with my mother. She retired back in December, but the months leading up to it, her workplace was using AI and all it did was create more work for her.

Remember the Jetsons? A world where tech gains in productivity meant you could push two buttons, be done for the day, and still get paid. Some people back in the day really thought this was gonna happen.

Pragmatic Engineer, I generally find him too optimistic and unconcerned with externalities but it is useful to get a sense of what the most gung ho AI adopters are doing. This post specifically is what I was thinking of newsletter.pragmaticengineer.com/p/steve-yegg...

So it’s like training an intern to do basic, time consuming tasks BUT an intern learns. That time spent teaching them and checking their work pays off as eventually you can trust them to do the job unsupervised. An LLM never learns, so you will be stuck supervising it for ever.

I think the ration of time in writing the software vs debugging the software is 20% vs 80%. Most rockstar programmers and AI ones can spew out large amounts of code, but actually making it work and fit the project requirements is the difficult part that requires real human intelligence.

Zipheads. Have you read *A Deepness in the Sky?*

My Doomer worry is that the massive resources demanded to keep chips cool will hasten climate change and kill everyone that way, but yours is good too.

This is my exact fear as well. I’m in visual arts and anytime someone asks me about AI replacing us - I’m like no actually it’s not that good and I don’t think it will get much better. And then I remember the money men running the show don’t care if they can shovel slop at a low cost.

...By design.

I’ve seen enough sci-fi to know there are zero practical and ethical uses for AI - even “good” ones

and to avoid taking actions that would concentrate power inappropriately or undermine checks and balances" Fwiw. and its also supposed to like, not kill everyone, they put that in there.

yes, that's accurate and realistic

Very mild autism for the win. 🫤