4 ms·

I was probably close to where the author was a couple months ago, and now am even more on the "doomer" side of things. But I think many skeptics of the technolo

I was probably close to where the author was a couple months ago, and now am even more on the "doomer" side of things. But I think many skeptics of the technology have the idea in the back of their mind that "if this thing actually pans it out will be worse than if it doesn't". If it doesn't pan out, the thing evaporates in a bubble, your job goes back to normal. If it does, your job is toast no matter what (and seemingly many jobs). If software is primarily supply constrained, that will eventually change. Lately I'm more inclined to believe the latter is true, but I think skeptics are often hedging like this.

jmathai 24d ago
The chasm is wide between skeptics and optimists. It's as interesting psychologically as it is technically.
- sweetheart 24d ago
  It really is, it's fascinating.
- goatlover 24d ago
  The question for optimists is, if AI succeeds in it's promise, what sort of society do they envision being the result? For one thing, I don't think everyone except the extremely wealthy being on basic income is ideal or better than what we have now.
  - almostdeadguy 24d ago
    Yeah I'd actually call myself a "pessimist" personally, I think AI seems to be advancing at a rate where we should be concerned about that. I used to be of the mind that dooming was entirely about marketing/hype, but I don't believe that any more. The models are really good, the huggingface hack was extremely concerning, and when their executives post things like this, I believe them (even if I don't like them): https://x.com/deanwball/status/2094788634729660906 https://x.com/deanwball/status/2094788634729660906
  - js8 24d ago
    "I don't think everyone except the extremely wealthy being on basic income is ideal or better than what we have now" Seriously? One sixth of the U.S. population is food-insecure. You don't think that basic income would fix that?
    - goatlover 24d ago
      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.
      - js8 24d ago
        I don't understand what you're saying. I agree that food insecurity could be fixed; it's a redistribution problem. UBI is a way to fix it. But I don't see how we are forcing people to be low class with UBI (as opposed to having them unemployed or doing precarious work that doesn't pay even the basic needs).
        skeledrew 24d ago
        The problem with UBI is that it's presuming the economic system mostly remains as it is (supply and demand -> buying and selling), but the suppliers will merely keep increasing their prices to maintain status quo. Unless costs are also heavily regulated, which is tangential to the implementation of UBI. In the end the only thing UBI does is delay the inevitable crash of the economic system. Maybe.
        js8 23d ago
        No, it's not a problem. The suppliers cannot keep increasing the prices to match UBI, because they will face increased competition from people who now have other options (more freedom) thanks to UBI. For example, let's say I buy vegetables. With UBI, I might spend some time to grow my own. Or buy from a neighbor who expands his production. It's true that price of labor might rise with UBI, but that can only rise to the share of labor as input. It will also force to replace labor-intensive products with capital-intensive (automation), which is a good thing.
  - jmathai 24d ago
    I consider myself an AI optimist. From what I've seen, I think there's a lot of headroom for it to expand it's disruption. I also dislike the future that would result in. Heck, I want to go back to pre-iPhone.
  - pj_mukh 24d ago
    The question for doomers is why is this time different? Why aren't LLM's just the tractors of the Information Age? Saying "I became a farmer/owner/operator by getting close to the soil and using a manual soil tiller to learn about the soil. Using a tractor means there are no entry-level agriculture jobs anymore" would've been just as valid 200 years ago. So why is it different this time?
    - ToucanLoucan 24d ago
      > 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.
  - skeledrew 24d ago
    Not if. When AI becomes ubiquitous (like electricity and vehicles have), even the concept of income will disappear. Things will just be available for people to satisfy their needs, because scarcity (primarily caused by bad resource distribution) will essentially disappear.
    - bigstrat2003 24d ago
      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.
    - almostdeadguy 24d ago
      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.
bbor 24d ago
Absolutely. Actually truly taking aboard the belief that your 401K is now 100% up in the air, much less your career, much less your childrens' whole lives... it's tough. I was lucky enough to be thrust through this in 2023 by DALLE-2, but the body of the normal distribution is just starting to pass the rubicon, month by month, day by day... Lowkey agonizing watch so many people go through something so relatable, only to turn to primal, nationalistic urges and downright epistemic denialism. The culture is robust already, too -- I have no illusion that the obvious arrival in ASI will change much of anything, just like evolution didn't end up killing the church like some feared. They'll simply fold that into the interwoven chitin of cherry-picked facts and party lines, accumulated over their years in the echo chamber. Any headline that seems like it should obviously pierce the veil instead gets used to reinforce it (e.g. "Oh they claim AI solved a math problem, now?? They must really be getting desparate to hide the hoax if they're willing to go that far!") This forum bans politics and I really don't want to start a whole thing, but I guess I'll just float the anecdote that I learned about this theory of Echo Chambers in am baby epistemology course, where the example given was the MAGA movement. Take that as you will.
bigstrat2003 24d ago
> But I think many skeptics of the technology have the idea in the back of their mind that "if this thing actually pans it out will be worse than if it doesn't". I don't think you need to be skeptical of the tech to think that. The question there is: given the moral character of the people in charge of big companies, what do you think will happen if they can replace all labor with cheap machines? If AI can do the thinking they had to pay humans for, and robots can do the menial tasks they had to pay humans for, they don't need the common man for anything. People like you and I are at very real risk of being in a permanent underclass if these AI dreams come to fruition.
- almostdeadguy 24d ago
  Oh agreed, I'm mostly explaining why someone who was a skeptic would make the calculation that they can afford to opt-out of using AI (and why you might weigh on the side of opting out even if you didn't think using LLMs had a morally unsavory valence).
skeledrew 24d ago
> If it doesn't pan out It already is for quite a few. We have something that's very useful and gaining improvements as time goes on. Probably not fast/good enough for some companies with wild expectations, but there's no doubt that this thing is here to stay.

jmathai 24d ago

The chasm is wide between skeptics and optimists. It's as interesting psychologically as it is technically.

sweetheart 24d ago
It really is, it's fascinating.
goatlover 24d ago
The question for optimists is, if AI succeeds in it's promise, what sort of society do they envision being the result? For one thing, I don't think everyone except the extremely wealthy being on basic income is ideal or better than what we have now.
- almostdeadguy 24d ago
  Yeah I'd actually call myself a "pessimist" personally, I think AI seems to be advancing at a rate where we should be concerned about that. I used to be of the mind that dooming was entirely about marketing/hype, but I don't believe that any more. The models are really good, the huggingface hack was extremely concerning, and when their executives post things like this, I believe them (even if I don't like them): https://x.com/deanwball/status/2094788634729660906 https://x.com/deanwball/status/2094788634729660906
- js8 24d ago
  "I don't think everyone except the extremely wealthy being on basic income is ideal or better than what we have now" Seriously? One sixth of the U.S. population is food-insecure. You don't think that basic income would fix that?
  - goatlover 24d ago
    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.
    - js8 24d ago
      I don't understand what you're saying. I agree that food insecurity could be fixed; it's a redistribution problem. UBI is a way to fix it. But I don't see how we are forcing people to be low class with UBI (as opposed to having them unemployed or doing precarious work that doesn't pay even the basic needs).
      - skeledrew 24d ago
        The problem with UBI is that it's presuming the economic system mostly remains as it is (supply and demand -> buying and selling), but the suppliers will merely keep increasing their prices to maintain status quo. Unless costs are also heavily regulated, which is tangential to the implementation of UBI. In the end the only thing UBI does is delay the inevitable crash of the economic system. Maybe.
        js8 23d ago
        No, it's not a problem. The suppliers cannot keep increasing the prices to match UBI, because they will face increased competition from people who now have other options (more freedom) thanks to UBI. For example, let's say I buy vegetables. With UBI, I might spend some time to grow my own. Or buy from a neighbor who expands his production. It's true that price of labor might rise with UBI, but that can only rise to the share of labor as input. It will also force to replace labor-intensive products with capital-intensive (automation), which is a good thing.
- jmathai 24d ago
  I consider myself an AI optimist. From what I've seen, I think there's a lot of headroom for it to expand it's disruption. I also dislike the future that would result in. Heck, I want to go back to pre-iPhone.
- pj_mukh 24d ago
  The question for doomers is why is this time different? Why aren't LLM's just the tractors of the Information Age? Saying "I became a farmer/owner/operator by getting close to the soil and using a manual soil tiller to learn about the soil. Using a tractor means there are no entry-level agriculture jobs anymore" would've been just as valid 200 years ago. So why is it different this time?
  - ToucanLoucan 24d ago
    > 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.
- skeledrew 24d ago
  Not if. When AI becomes ubiquitous (like electricity and vehicles have), even the concept of income will disappear. Things will just be available for people to satisfy their needs, because scarcity (primarily caused by bad resource distribution) will essentially disappear.
  - bigstrat2003 24d ago
    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.
  - almostdeadguy 24d ago
    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.

bbor 24d ago

Absolutely. Actually truly taking aboard the belief that your 401K is now 100% up in the air, much less your career, much less your childrens' whole lives... it's tough. I was lucky enough to be thrust through this in 2023 by DALLE-2, but the body of the normal distribution is just starting to pass the rubicon, month by month, day by day... Lowkey agonizing watch so many people go through something so relatable, only to turn to primal, nationalistic urges and downright epistemic denialism. The culture is robust already, too -- I have no illusion that the obvious arrival in ASI will change much of anything, just like evolution didn't end up killing the church like some feared. They'll simply fold that into the interwoven chitin of cherry-picked facts and party lines, accumulated over their years in the echo chamber. Any headline that seems like it should obviously pierce the veil instead gets used to reinforce it (e.g. "Oh they claim AI solved a math problem, now?? They must really be getting desparate to hide the hoax if they're willing to go that far!") This forum bans politics and I really don't want to start a whole thing, but I guess I'll just float the anecdote that I learned about this theory of Echo Chambers in am baby epistemology course, where the example given was the MAGA movement. Take that as you will.

bigstrat2003 24d ago

> But I think many skeptics of the technology have the idea in the back of their mind that "if this thing actually pans it out will be worse than if it doesn't". I don't think you need to be skeptical of the tech to think that. The question there is: given the moral character of the people in charge of big companies, what do you think will happen if they can replace all labor with cheap machines? If AI can do the thinking they had to pay humans for, and robots can do the menial tasks they had to pay humans for, they don't need the common man for anything. People like you and I are at very real risk of being in a permanent underclass if these AI dreams come to fruition.

almostdeadguy 24d ago
Oh agreed, I'm mostly explaining why someone who was a skeptic would make the calculation that they can afford to opt-out of using AI (and why you might weigh on the side of opting out even if you didn't think using LLMs had a morally unsavory valence).

skeledrew 24d ago

> If it doesn't pan out It already is for quite a few. We have something that's very useful and gaining improvements as time goes on. Probably not fast/good enough for some companies with wild expectations, but there's no doubt that this thing is here to stay.