Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI Won't Cause Unemployment

pmarca.substack.com

Post reply on HN
Print view View original

351–360 of 362 posts

Re: AI Won't Cause Unemployment

#351

Post by valenterry »

staunton wrote: post #251
valenterry wrote: post #241

Earlier quoted context omitted.

> will be able to write a detailed functional description of a system in natural language, domain models, with data points, and an AI will write, test, integrate, and deploy production level code. Yes, once math is also done in natural language, this will be the next step. I don't believe I will experience this in my lifetime unfortunately, but it will eventually happen.

Math has been done in natural language since the very beginning. Whatever math writing is not in natural language serves a specific purpose, such as being more precise than usual. In time, a jangon develops naturally as people with a lot of shared background knowledge optimize communication within their ingroup.

> In time, a jangon develops naturally as people with a lot of shared background knowledge optimize communication within their ingroup

Exactly. Only once humans completely stop working on trying to understand a system, we will only use natural language as suggested.

Re: AI Won't Cause Unemployment

#352

Post by ChicagoDave »

comfypotato wrote: post #290
ChicagoDave wrote: post #212

This article is pure conservative BS. The stagnation of certain industries is because of tax policy and the insane greed injected into our healthcare systems when the HMO laws passed in the 1980's. I'm not even sure what pmarca is talking about here. It's like he's waving his arms, "look over here at my BS and don't look over there where AI is about to slam information workers." AI will absolutely change the way soft…

I read the article far enough to confirm that it’s too biased to be taken seriously; I’m commenting more from the general high-level discussion of AI and labor. The advances you mention wiping out software divisions: if you apply that to robotics, half of the workers in the world will be out of work. This kind of thing is very far off, yes? (Which, in my opinion, makes your comment a tad bit nonsense in and of itself…

You don’t need a physical robot to replace software engineers. I can envision one person scripting an entire complex application and an AI generating all of the code to perfection. To update the app, that one person adds qualifiers to the original text and the AI will update the code and automatically handle api/schema changes.

I see this happening within five years, maybe sooner.

Re: AI Won't Cause Unemployment

#353

Post by ETH_start »

ETH_start wrote: post #257
TheOtherHobbes wrote: post #246

Earlier quoted context omitted.

The administrators are a direct consequence of insurance-mandated budgeting. You can't seriously argue that they're caused by "regulations" when front-line doctors are having to call insurance companies from their offices to get clearance for life-saving procedures. Or when patients having an emergency have no idea how much treatment will cost because their insurance company has to work out whether treatment will be…

Regulations make the provision of healthcare and health insurance more complex, so seem like the most probable culprit. The decline in the share of healthcare spending constituted by out-of-pocket spending, due to growing insurance-mandated budgeting, could very well have played a role in the growth of healthcare costs, and I believe it did. But two points: 1. Those mandates are regulations, so this development doesn…

To elaborate on 1: the rise in the reliance on third party payers has been mostly due to the steady expansion of Medicare/Medicaid, and the piling on of laws that encourage reliance on health insurance over out-of-pocket healthcare spending, like "Patient Bill of Rights" laws that probibit all but the most comprehensive health insurance plans, and rule out catastophic-only plans that were at one time more common:

https://archive.is/GHjIl

Re: AI Won't Cause Unemployment

#354

Post by comfypotato »

ChicagoDave wrote: post #352
comfypotato wrote: post #290

Earlier quoted context omitted.

I read the article far enough to confirm that it’s too biased to be taken seriously; I’m commenting more from the general high-level discussion of AI and labor. The advances you mention wiping out software divisions: if you apply that to robotics, half of the workers in the world will be out of work. This kind of thing is very far off, yes? (Which, in my opinion, makes your comment a tad bit nonsense in and of itself…

You don’t need a physical robot to replace software engineers. I can envision one person scripting an entire complex application and an AI generating all of the code to perfection. To update the app, that one person adds qualifiers to the original text and the AI will update the code and automatically handle api/schema changes. I see this happening within five years, maybe sooner.

Your prediction here incorporates the paradigm shift I mentioned above. The AI software upon which your prediction relies does not yet exist. There is a huge gap between the current functionality and AI that can produce non-trivial end-to-end programs.

It’s tempting to fall into the trap of “five years away” because things like copilot are so significantly different from what we had before.

As far as I’m concerned, Occam’s Razor tends to suggest much bigger changes in how robotics affects labor. We simply have to build new robots. There’s no paradigm shift necessary to design a new robot; only labor on the part of the engineers. It’s also relevant here that most folks aren’t too keen to be employed as automation engineers. It’s not exactly glamorous to work explicitly towards the purpose of removing others’ jobs.

I personally feel that scientific advances in AI will slow down, not speed up. There’s plenty of research to do in computer science, obviously, but scientific breakthroughs slow down over time.

Re: AI Won't Cause Unemployment

#355

Post by afunctionof »

kragen wrote: post #100
unmole wrote: post #96

Earlier quoted context omitted.

> I can't help but think Andreesen's gone off the deep end a bit The first time I heard of him was in 2016: https://www.fastcompany.com/3056581/marc-andreessen-riles-up... As far as I'm concerned, he's always been kooky.

he led the mosaic project at ncsa, creating the first web browser most people today would recognize as a browser, with things like images mixed in with text then he went off and founded netscape, creating the dot-com boom since then he's mostly been running a successful vc firm so yeah he's always been 'kooky'

Copying comment from earlier thread because I sent it two days after thread:

You should add your Matrix to your contact page.

Edit: Also, is there going to be another der* release this year? They and therefore you have taught me more than I'd learned in the last five years otherwise.

Re: AI Won't Cause Unemployment

#356

Post by omgomgomgomg »

rocket_surgeron wrote: post #58

LOL The red lines are lines where (PRIMARILY) skilled labor is required for products and services essential to human life. The blue lines are lines where unskilled labor in Shenzhen factories in pump out unnecessary consumer goods. Medical costs are rising because people like him are pumping money into hedge funds and they are squeezing every red cent out of the systems they have captured-- not because doctors have t…

In his case.....looking at his head, something unique happened.

Re: AI Won't Cause Unemployment

#357

Post by lisasays »

ChicagoDave wrote: post #212

This article is pure conservative BS. The stagnation of certain industries is because of tax policy and the insane greed injected into our healthcare systems when the HMO laws passed in the 1980's. I'm not even sure what pmarca is talking about here. It's like he's waving his arms, "look over here at my BS and don't look over there where AI is about to slam information workers." AI will absolutely change the way soft…

All of those "things" will be plugged into an AI and one senior technologist will be able to write a detailed functional description of a system in natural language, domain models, with data points, and an AI will write, test, integrate, and deploy production level code.

Given that the severe limitations of the current state of the art (as we've seen from from how trivial is to show that these systems are basically spouting nonsense, and certainly have no meaningful understanding of the world we actually live in); and that this 100 percent waterfall model of software creation you describe has never worked in the aforementioned actual world that we live in -

How, exactly, do you "know" that the above will come to pass?

Re: AI Won't Cause Unemployment

#358

Post by kragen »

afunctionof wrote: post #355
kragen wrote: post #100

Earlier quoted context omitted.

he led the mosaic project at ncsa, creating the first web browser most people today would recognize as a browser, with things like images mixed in with text then he went off and founded netscape, creating the dot-com boom since then he's mostly been running a successful vc firm so yeah he's always been 'kooky'

Copying comment from earlier thread because I sent it two days after thread: You should add your Matrix to your contact page. Edit: Also, is there going to be another der* release this year? They and therefore you have taught me more than I'd learned in the last five years otherwise.

wow, that's astonishing and wonderful to hear, thank you

i hope the things you learned from them were mostly true

i was planning to do one in december but i felt like i didn't have much material because i had a pretty unproductive year in 02022, and i'm increasingly dissatisfied with the medium

you're right about matrix

Re: AI Won't Cause Unemployment

#359

Post by PeterisP »

bfdm wrote: post #11

You could also argue that the "red" categories align more closely with "needs" people have little to no choice but to pay, and are ripe for predatory pricing. It's true there are monopoly problems here, but the answer isn't to remove regulation it's to strengthen it. A return to real anti-trust in particular. The blue items lean more towards luxury items and discretionary spending, so pricing can't be so exploitative…

I feel more like the red items are all the ones where the price pressure has been removed for one reason or another, where a new company with a rich investor can't arrive and create a hugely increased supply.

Housing in desirable areas works that way due to zoning/NIMBY restrictions, education works that way because at the point of choosing to pay there's no effective price competition as student loans cover the difference so that potential students don't really change their decisions meaningfully because the price is $10,000 higher than in another place, textbooks have the same but even more - because students paying for textbooks have almost no say in which textbooks will get chosen and can't choose a different book because it's cheaper, and medicine because of both of these factors - artificially limited supply of doctors due to restricted residency&licensing spots, and a price disconnect with insurers paying huge bills according to rules of what needs to be paid or not and lack of price transparency (and price competition!) for procedures; the majority of medical cost is for planned, non-urgent procedures but even for these people aren't making a meaningful choice whether they'll do the same thing in Hospital A for $10,000 or in Hospital B for $7,000, it's either covered/in-network/whatever or not; so there's no free market price setting that would limit these costs.

Re: AI Won't Cause Unemployment

#360

Post by PeterisP »

BurningFrog wrote: post #338
camillomiller wrote: post #116

Earlier quoted context omitted.

You’re applying what I would call the fallacy of scale. Your take is correct as long as these singularity happen at relative scales. We don’t know at all what happens when massive scale becomes a factor. Computers and the internet allowed virtualization and we now base our economies largely on virtualization and abstraction. That also means almost infinite replicability. In turn, infinite replicability allows levels…

My steelman of what you're saying is that "this time it's different" because this will be a much bigger change than the normal yearly loss of 2-4% of all jobs. I'd add that it's different in that it will hit us college educated professionals who are not at all used to this, instead of the usual blue collar industries. We as a group are not ready and will scream very very loudly if/when this happens :) It's important…

I think that the visible impact will also be different because it not only matters how many people suffer but what types of people suffer. When the bad stuff happens to mostly disenfranchised poor people, they can (and will) push for change by e.g. protest votes for populists that promise to break the status quo and turn back time (and, to be clear, I'm not talking about the USA here - this is a global phenomenon), but that won't effectively advance their interests, as that simply can't be done even if the populists both get power and try to keep their promises.

On the other hand, when/if the same suffering happens to "semi-elites" with social capital and connections, they will raise noise in media and try to use the government establishment to "do something", and while it also might be ineffective, at least the appearance of governments needing to take action and help the adjustment will be much more significant than for the blue collar troubles.

Post reply on HN