> The question for doomers is why is this time different?

Difference 1: The tools themselves only exist due to rampant exploitation of vast intellectual resources that were created, catalogued, and published with the intent to help people learn, not machines. And this was frequently done against the explicit expressed wishes of the holders and creators of those resources.

Difference 2: The tools do not work deterministically. Every run of a prompt is effectively pulling a lever on a slot machine. Sometimes you win, sometimes you don't. However just like a good slot machine, the developers behind them have put a number of little touches and "features" that change nothing about the underlying machinery and instead explicitly target psychological weak points to build trust and confidence in said tools.

Difference 3: The tools do not work reliably. Agents are a complete shitshow but even if you strictly work in the confines of a chat bot, it will occasionally just make shit up, in our industry that means: code that does not compile and run. Hallucination has been a consistent issue since day 1 of this tech and probably the most impactful to user experience overall, and despite these facts it hasn't been solved in the subsequent 3 years whatsoever. It is by every account I've read, an intractable problem because these are machines that are fundamentally designed to make approximate representations of final products that look plausible, not that function correctly. That is a big distinction and it matters.

Difference 4: Saturation. Despite your assertion above, there was virtually no drive to prevent the ascent of tractors. Tractors were in high demand basically as soon as they were available, barring some ups and downs as the technology was fully built out and proven. The advent of motor vehicles as a whole completely redesigned our cities and to a great degree, society itself. Horses are still around today but as a hobby for monied people. Despite all of the above, Silicon Valley is simply tripping over itself, pouring literally billions of dollars into these things, and putting them in every product category whether it makes sense or not. This engenders a lack of trust, not just for the ethics of the tool itself, but whether OpenAI and company actually think this is good, or weather it simply must be good or the entire show is going to come crashing down around them. Instead of a product vision, where a company is setting out to solve specific problems with well made solutions, we have a few companies who feel distinctly desperate, trying to get their product in front of as many eyes as humanly possible, irrespective of it's ability to add any value at all to what those eyes are looking at.

>Why aren't LLM's just the tractors of the Information Age?

We didn't have every company and every government in the world spending trillions of dollars to completely reorganize society and technology around tractors, and to force them into every possible aspect of our lives whether people want them or not, to the point that tractors are all but existentially necessary to maintain our technological civilization.

We didn't have billionaires with deeply entrenched political interests believing in a cult of neo-fascist "tractor accelerationism."

We didn't have people forming parasocial relationships with tractors so deep and profound that they superceded their relationships with real people, and subsumed their own identities. Tractors didn't gaslight people into insanity and suicide. There's no such thing as "tractor psychosis." No one committed murder to pacify the Tractor Basilisk.

Tractors didn't flood every form of media with misinformation and slop that in a decade's time will make subjective reality and history utterly indecipherable from its own gibberish.

No one believed that tractors would lead to a perfect post-scarcity utopia... somehow. But only when everyone submits to tractors and the new world order that tractors will create and accepts that tractors are the new reality and that they just have to learn to live with that, regardless of any harm they might do.

No one forgot how to farm because they used tractors. There was no mass loss of farming knowledge or "deskilling," or degradation of cognitive or social abilities that came with using tractors.

There was no one complaining about how utterly meaningless farming had become now that tractors came along, but they themselves had become so addicted to tractor-maxxing that they couldn't simply stop.

And on top of that, no one wanted to replace farmers entirely with tractors, just because tractors didn't need to be paid. And then the plumbers. And the architects. And the cooks. And the doctors. And eventually the judges, police and politicians.

LLMs are very obviously not the tractors of the information age. If they were, Hacker News wouldn't be consumed by them. People wouldn't talk about them with the same awe and passion usually reserved for religion.

right... AI will magically fix greed and we'll all live happily ever after

No, it's very much "if". AI can't even do the things it is claimed to be good at today. It is a wash, or a small gain, for engineers of any reasonable competence. The idea that it's going to replace people is not remotely guaranteed based on current capabilities.

You're describing an economic situation that has never happened. Electricity and vehicles didn't make income dissapear.

So in the short term, it seems likely to deflate the cost of software and other products/services produced by professional workers. If it becomes good at other forms of production (i.e. physical things, physical labor, etc.), perhaps it causes currency exchange to become pointless.

There are a few caveats to this that everyone seems to leap over, one very important one is that the a priori control of production, land, and natural resources is not egalitarian, and there's no reason to really believe it would become egalitarian even if people could in theory produce everything they wanted/needed for themselves.

But even barring that, we live in a world with finite natural resources. And that by itself should be a serious concern for anyone who sees this as a possibility. I get that optimists also seem to believe that technology we don't have like dyson spheres, asteroid mining, intergalactic travel, etc. will solve any resource scarcity concerns, but that is not a given, and developing any of those things would still rely on our planets natural resources to even bootstrap.

We live in a world that can't even coordinate the safe development of models or nuclear anti-proliferation, and already experiences rampant injustice surrounding use of our planets natural resources, which does not bode well for the future.

And even if we did end up in a world where these problems were solved, advanced models posed no threat to our existence, etc. they would not be solved instantaneously, and would likely cause catastrophic changes to society and massive amounts of suffering. We can solve scarcity of basic needs and access to medical care, decarbonization, etc. today. Without the use of AI. We do not need to go through years of suffering because tech execs want their play things.

By putting almost everyone else on basic income because they're jobs were automated? How much would that need to be? Even if it's enough for basic needs, you're relegating most of the population to a low class status with limited options for improving their situation.

Food insecurity could be fixed now if it wasn't a politically divisive issue. There's no reason to believe UBI won't be even more divisive.

Basic income sounds like welfare dressed up, where is the dignity of work, where is the independence from dependence? You shouldn't bite the hand that feeds you, sounds like less freedom to me, which is why I think the oligarchy is the one the earnest proposers of the idea