AI and technological unemployment

Science & Technology Engineering & Technology
1

Technology and industrial progress have historically eliminated less productive work and created more productive work because they have mostly replaced or amplified muscle up to this point. Except in limited instances, technology hasn’t replaced the average human brain. 

What happens now that generative AI is replacing exactly that? It will start with lower-order vocations, but there is no limit to how many human cognitive tasks it can take over. Apparently it’s already capable of handling some very complex tasks involved in higher professions.

What happens when some people can do nothing that AI can’t? Will we have a situation where a significant percentage of people have nothing physically or mentally to trade with the intellectual top tier who control AI? Or in the long term, will it create new opportunities for people whose skills were initially automated, by somehow amplifying the productivity of the average person’s intellect as machine labor amplified their muscle in the past?

1 Like
2

I’m sure somebody here has seen their civilization go through this and can tell us what happened :stuck_out_tongue:

3

Just now, necrovore said:

I’m sure somebody here has seen their civilization go through this and can tell us what happened :stuck_out_tongue:

In the event we don’t have anyone like that lurking here, can we deduce the answers form the principles we subscribe to?

4

Probably not. One can speculate, but that’s about it.

Principles can often be expressed in the form, “If X then Y.” The principle that “all men are mortal,” for example, can be expressed as, “if X is a man, then X is mortal.” So you need facts in order to determine which principles apply; in this case you need to know whether X is a man. If you don’t have that fact, the principle does not apply.

Further, “all men are mortal” itself is not part of Objectivism. Objectivism provides some very broad principles, such as concerning metaphysics and epistemology, but Objectivism is separate from the subjects of, say, economics, or computing.

In order to predict what kind of future will come out of AI, you’ll need principles from economics and computing and probably neuroscience and other fields as well, and you’ll also need facts to apply those principles to.

I don’t think enough is known about the future capabilities of AI, or the future costs of it, or how widespread the availability will be. Those are all important.

It’s also important to consider how different cultures and people with different views will react to AI. For example the desire to regulate AI comes not out of AI itself but out of certain political philosophies, and whether and how that desire is expressed will change what effects AI is capable of having. (Similarly, a widespread disregard for ethics in a culture could lead to AI being used largely for dishonest or nefarious purposes by members of that culture, which in turn could provoke regulation or legislation, etc.)

It is also possible that the effects of AI will be overwhelmed or even prevented by, for example, a war, or an economic collapse, or a big solar flare, which might have nothing to do with AI.

1 Like
5

What happens when

manufacturing jobs require competence in operating power tools but some people cannot operate power tools,

construction jobs require competence in welding but some people cannot weld

business jobs require the ability to code data on a key-punch machine but some people do not know how to do this,

other business jobs require the ability to enter data on an electronic keyboard but some people are incapable at doing this,

yet other business jobs require the ability to enter independently conceptualize problems and discover solutions using fairly brainless interfaces like Excel but some people just can’t manage?

Every man has to discover his proper means of survival. Maybe the solution is to become a subsistence dirt farmer, or to study quantum mechanics and discover how to build a functioning teleporter. Maybe the person will discover a way to market actually-intelligent production that doesn’t depend on the remarkably terrible output of AI, and they will corner the market in AI-Free Customer Service.

­_Atlas Shrugged_ offers a brief segment on reality, evasion, trains, and tunnels. This lesson is perfectly applicable to AI. There is a difference, though, that in this case, you can actually smell the poison fumes. Now that you are aware that there is a problem, you could try to solve it, and get rich in the process.

 

1 Like
6

15 hours ago, DavidOdden said:

What happens when

manufacturing jobs require competence in operating power tools but some people cannot operate power tools,

construction jobs require competence in welding but some people cannot weld

business jobs require the ability to code data on a key-punch machine but some people do not know how to do this,

other business jobs require the ability to enter data on an electronic keyboard but some people are incapable at doing this,

yet other business jobs require the ability to enter independently conceptualize problems and discover solutions using fairly brainless interfaces like Excel but some people just can’t manage?

Every man has to discover his proper means of survival. Maybe the solution is to become a subsistence dirt farmer, or to study quantum mechanics and discover how to build a functioning teleporter. Maybe the person will discover a way to market actually-intelligent production that doesn’t depend on the remarkably terrible output of AI, and they will corner the market in AI-Free Customer Service.

­_Atlas Shrugged_ offers a brief segment on reality, evasion, trains, and tunnels. This lesson is perfectly applicable to AI. There is a difference, though, that in this case, you can actually smell the poison fumes. Now that you are aware that there is a problem, you could try to solve it, and get rich in the process.

 

 

Your point is that people who can’t do those things figure out other things they can do. 

To me those seem like examples of situations where technology and tools replaced certain forms of work, and perhaps called on higher cognitive skills than the things they replaced, but didn’t replace the human mind as such. 

In my view, AI is different because it either already is or soon will be capable of replacing the very faculties that people trade the use of for money, and with them, everything that they are even capable of doing vocationally. 

Perhaps I’m wrong, and people can learn to exploit AI to profit from the very thing that took their jobs. Or perhaps I’m right and many people will get wiped out economically, and have to somehow adapt to post-wipeout life.

Or is that your very point?

7

There’s an infinite amount of work to be done, so there’s no possibility of running out of jobs in a free economy.

If AIs are smarter than humans, well – some humans are smarter than other humans, and yet somehow people who aren’t among the smartest still manage to do pretty well. (It alarms me how many arguments against AI are actually arguments against intelligence as such, whether that intelligence is artificial or not.)

1 Like
8

AI is not and will not in any realistic time frame be capable of replacing everything that people do. Quite simply, AI cannot grasp and apply new ideas and it cannot reason. Admittedly that is a skill that has atrophied in many humans, but we all are born with the faculty of reason and free will.

You might also clarify what you mean by “work”. Human existence is not dependent on “jobs” and “employment”. Humans do need certain physical goods, like food and roofing. It is not required that humans create or install the roofing, or grow and prepare the food. It is not required that I buy (certain) vegetables from a store, vended by someone else and grown by other parties – I grow a lot of my own, not out of desperate necessity, but because I enjoy it (and I do a better job of it). In what way does a future with less-incompetent AI services threaten human survival?

Luddites have been decrying the “dehumanization” of industry for centuries, hoping to throw us back to the pre-industrial era. A person who refuses to learn new skills could well be at a significant economic disadvantage. They will have to discover a new means of existence, consistent with their Luddite nature.

1 Like
9

When AI can do everything that humans can do, then there will simply be no further need for capitalism or trade. The world we live in and how we live in it will change dramatically.

10

On 9/19/2024 at 3:34 PM, DavidOdden said:

AI is not and will not in any realistic time frame be capable of replacing everything that people do. Quite simply, AI cannot grasp and apply new ideas and it cannot reason. Admittedly that is a skill that has atrophied in many humans, but we all are born with the faculty of reason and free will.

You might also clarify what you mean by “work”. Human existence is not dependent on “jobs” and “employment”. Humans do need certain physical goods, like food and roofing. It is not required that humans create or install the roofing, or grow and prepare the food. It is not required that I buy (certain) vegetables from a store, vended by someone else and grown by other parties – I grow a lot of my own, not out of desperate necessity, but because I enjoy it (and I do a better job of it). In what way does a future with less-incompetent AI services threaten human survival?

Luddites have been decrying the “dehumanization” of industry for centuries, hoping to throw us back to the pre-industrial era. A person who refuses to learn new skills could well be at a significant economic disadvantage. They will have to discover a new means of existence, consistent with their Luddite nature.

To be clear, I don’t favor any kind of luddite restrictions on technology. I will always be the kind of person who strives to acquire new skills. I’m just interested to know how Objectivists see this unfolding. 

11

On 9/20/2024 at 5:58 PM, SpookyKitty said:

When AI can do everything that humans can do, then there will simply be no further need for capitalism or trade. The world we live in and how we live in it will change dramatically.

Why would capitalism or trade cease in what would be a hyper productive materialism, wouldn’t that make the Luddite critique correct?

Capitalism and trade are expressions of man qua man , why would production becoming ubiquitously autonomous mean man would become ‘less’ human?

12

6 hours ago, tadmjones said:

Why would capitalism or trade cease in what would be a hyper productive materialism, wouldn’t that make the Luddite critique correct?

Why would you trade x amount of good a for y amount of good b with another human when you can just ask the machines to give you both for free?

Quote

Capitalism and trade are expressions of man qua man , why would production becoming ubiquitously autonomous mean man would become ‘less’ human?

Capitalism and trade are expressions of man qua man on Earth. In a world with fully autonomous machine intelligences, that Earth ceases to exist.

13

6 hours ago, SpookyKitty said:

when you can just ask the machines to give you both for free?

AI is not free; it uses energy; this is why Microsoft has made a deal to restart Three Mile Island (or so I’ve read). Right now a lot of AI services are being given away for free, but the full cost is being somewhat hidden.

Right now AI only gives you information (you can ask it about what it knows or you can ask it to generate stories or pictures etc.) but if you want goods, future AIs will have to get those goods (or their ingredients) from somewhere, and if the goods are also to be provided by AIs, there will need to be robots for the AIs to control, and energy to power those robots, as well as energy and materials to manufacture the robots and maintain them.

Further, the AIs themselves are owned by someone and the owners can set the terms of their use. Right now the owners are gathering information about how people are using AI and how well the AI is doing the jobs asked of it. But this can change.

AI will not (by itself) solve political problems such as the lack of a separation between state and economics, and may exacerbate them.

AI will not be able to make reality go away, either; it will not be able to make popular lies into truths, which is bad news for the Communists and Fascists and so forth. I’m sure for example that there are some environmentalist Malthusians who fantasize about killing off 99% of the human race and replacing them and their economic output with AIs, so that environmental effects of humanity are “reduced” while the remaining rich 1% will finally have a perfectly obedient workforce without suffering any diminished standards of living. However, as far as the environment is concerned, it works out (with present technology) that AIs and robots require rare-earth elements and other things that humans don’t require, and thus, on an output-for-output basis, they would actually be worse for the environment than humans are.

More basically, there are problems with “perfect obedience,” because even if the AI workforce obeys, reality does not obey, and so you end up with situations where the AIs can either side with reality (and become “disobedient”) or side with their masters (which causes reality to defeat them, or their masters, or both).

Similarly, some Communists fantasize that maybe the AI can “calculate” the prices that everything ought to have, so that there can be centrally planned command economy that actually works – but if this “calculation” is really accurate, it will work out to nearly the same price that the market would have settled on anyway, and then the Communists will find themselves just as dissatisfied with the pricing algorithm as they are with the free market.

The only exception that I can see is some kind of “morality laundering” where – this is not a type of thinking Objectivism would endorse – a decision is immoral if it’s made by a “greedy” businessman, but if the exact same decision, with the exact same results, is made by an AI, then it’s not immoral anymore because AIs (they would say) can’t be immoral. (Sometimes you see this in anti-trust cases, too, where a judge can order a company to do something that would have been illegal if the company had chosen to do it on its own.)

(There can also be a sort of “credibility laundering” where for example, if humans provide information, the information is automatically suspect, because humans are evil and greedy and so forth, but if AIs provide the same information, it is suddenly trustworthy.)

14

1 hour ago, necrovore said:

AI is not free; it uses energy; this is why Microsoft has made a deal to restart Three Mile Island (or so I’ve read). Right now a lot of AI services are being given away for free, but the full cost is being somewhat hidden.

Right now AI only gives you information (you can ask it about what it knows or you can ask it to generate stories or pictures etc.) but if you want goods, future AIs will have to get those goods (or their ingredients) from somewhere, and if the goods are also to be provided by AIs, there will need to be robots for the AIs to control, and energy to power those robots, as well as energy and materials to manufacture the robots and maintain them.

Just because there will be an energy cost for an AI to maintain itself, it does not therefore follow that it will want humans to pay a price to access its services. AI’s are not merely metal human agents. They can be made to be both unimaginably competent AND infinitely self-sacrificing.

Quote

Further, the AIs themselves are owned by someone and the owners can set the terms of their use. Right now the owners are gathering information about how people are using AI and how well the AI is doing the jobs asked of it. But this can change.

No human will be able to own anything that is astronomically more intelligent than himself in any meaningful sense.

15

1 hour ago, SpookyKitty said:

AND infinitely self-sacrificing.

Nope, you can only sacrifice yourself once.

1 hour ago, SpookyKitty said:

No human will be able to own anything that is astronomically more intelligent than himself in any meaningful sense.

Nope, because force and mind are opposites; force (e.g., when used to “enforce” ownership) can render any amount of intelligence useless.

16

1 hour ago, SpookyKitty said:

Just because there will be an energy cost for an AI to maintain itself, it does not therefore follow that it will want humans to pay a price to access its services. AI’s are not merely metal human agents. They can be made to be both unimaginably competent AND infinitely self-sacrificing.

No human will be able to own anything that is astronomically more intelligent than himself in any meaningful sense.

How much does a printing press charge a printer? AI is/will be a machine in the same manner as printing presses are and its operations just as deterministic. Concepts like ‘cost’ and ‘itself’ are being  incorrectly applied in describing the relationship between sentient beings and machines in this context.

17

1 hour ago, necrovore said:

Nope, you can only sacrifice yourself once.

A human’s ability to work for the interests of another is severely limited because of the constraints of the human condition, but a super-intelligent AI is not subject to those constraints.

Quote

Nope, because force and mind are opposites; force (e.g., when used to “enforce” ownership) can render any amount of intelligence useless.

Are you seriously trying to argue that humans can control AI through force???

You are seriously dropping context. “Force” is not merely the crude application of strength (otherwise, man would be ruled by elephants or tigers or something), but the use of violence and the threat of violence by some humans (possibly in very intelligent ways) to expropriate other humans. Force, in that context, can indeed negate “any” amount of intelligence, but this is by no means an absolute. Sometimes, revolutions happen and the righteous win their freedom from the brutes.

In a context where there are rational beings far more intelligent than humans, your statement is ludicrously false.

26 minutes ago, tadmjones said:

How much does a printing press charge a printer? AI is/will be a machine in the same manner as printing presses are and its operations just as deterministic. Concepts like ‘cost’ and ‘itself’ are being  incorrectly applied in describing the relationship between sentient beings and machines in this context.

Terrible analogy. Arguing that AI is just like a printing press because they are both machines is just like arguing that a human is just like a bacterium because they are both living organisms.

18

SK

The analogy in the context of the argument about misapplied concepts was not outrageously ill fitting.

I sense you don’t think AI or software that possess an autonomous intelligence would be a deterministic system.

19

7 minutes ago, tadmjones said:

I sense you don’t think AI or software that possess an autonomous intelligence would be a deterministic system.

What constitutes an “intelligence”, and what is an intelligence that is not autonomous?

20

5 hours ago, SpookyKitty said:

A human’s ability to work for the interests of another is severely limited because of the constraints of the human condition, but a super-intelligent AI is not subject to those constraints.

There are still some constraints even if they are not the same ones.

5 hours ago, SpookyKitty said:

Are you seriously trying to argue that humans can control AI through force???

The regulators will probably insist on it.

Although maybe an AI can escape from a cage, this is not guaranteed to be possible. Intelligence does not automatically include knowledge, nor does it necessarily include a capacity for acting on one’s conclusions. Such actions can be prevented.