Robots have been about to take all the jobs for 100 years
newsletter.pessimistsarchive.org
newsletter.pessimistsarchive.org
1. We've decimated the bottom end of the skills market in developed countries. Something often touted in the psychometrics literature is that the number of jobs for folks with IQ under 85 are quite rare in the US where once they were common. That's ~15% of the population who are close to un-employable.
2. People who spend years or decades on a skillset see it devalued and their income drops. It is often difficult to get a "second shot" at a new career in your 40s and can be deeply disruptive to you and your family's lives. Growing up I watched this happen to the parents of several friends who'd made careers in things like printing that were obsoleted by digital revolution. None of those people achieved the monetary success they might have in an un-disrupted industry.
You would think that automating your job would make you free to work less. No, of course not, you end up either unemployed or working at least as much as you used to or more.
A lot of the comments here on this thread are already pessimistic in relation to the automation because people low key know that in the last decades almost all societal changes have come to the detriment of general population.
The Fed and the financial class have been printing money relentlessly and have been squeezing all the juice of the economic benefits of automation by taxing it with inflation. The zero rates of the last 20 years have not reflected price stability, they reflected moving goalposts and strong technological deflation (which should've benefited the common people) coupled to high monetary inflation which was only allowed because of the technological deflation.
When you look at where did most of the money gained from technological innovation went - where all the newly found wealth was spent - the answer is obviously houses. The landlord / banking class have been printing money relentlessly inflating a bubble of the assets they hold, and they could only afford to do it because they absorbed the rising economic power of technology. You slave away while they get to print money.
We have effectively two tiered society. For the common people, a dollar is a dollar. For the financial class, a dollar is actually ten dollars, held with leverage and then bailed out by the Fed when it blows up.
In capitalist west, capitalism was also about distribution of goods but literally being nice about it. Working together.
The language is meaningless gibberish. The tone and emotional spin, roleplay are what’s important.
In the US we’re enabling iron fisted oligarchs like the USSR once had. Those who refuse to share without extreme deference and idolatry, not because they’re that important to the well being of billions (they’re not) but because they’re important to the political actors that insulate them from taxation.
The USSR et al had many more problems than most people in the west, at least of my generation, properly appreciate — the Holomodor and the irony of the Berlin Wall's official name (≈ anti-fascist protection barrier) being the most pertinent in my mind.
Conversely, while the USA was a symbol of hope and freedom in my childhood, that childhood was oblivious to the official segregation of the USA even in the 60s, let alone the unofficial segregation whose continued existence was the basis for the term "woke" prior to it being appropriated by all the political talking heads to mean everything and nothing.
The oligarchy in the USA today is not one I favour, but it is kinder than the… I was going to reference the events of the Homestead strike, but it turns out I must remember my own words, for that too is less clear-cut than I had heard: https://en.wikipedia.org/wiki/Homestead_strike
But yeah, reality is a complicated mess.
My opinion is that laissez-faire capitalism is BS, that the Roosevelt New Deal marked an end to that era for the same reason that the equivalent living conditions ended Soviet-style communism; that other forms of communism and capitalism exist; that both over-simplify human nature and that a superior alternative to both can be designed with careful effort; that such a thing probably won't be taken seriously until the current system is utterly broken because of those with power in the current system; and that by default AI and robotics is a lot closer to communism than anything else because "from each according to ability" requires much less personal consideration when everyone has a dozen droid servitors at their disposal.
Yeah, ask my grand grand father how great that worked. Oh sorry, you can’t, he was shot in the gulag in 1937 for anti revolution activities (and rehabilitated in 60s, saying it was a terrible mistake, that’s a relief, right?).
Its intrinsic economic woes were overplayed by western elites who were terrified of communist contagion.
Western ideology is about the primacy of property ownership. It's not economically efficient in the slightest but it's the only game in town these days.
They've been automated for so long and so thoroughly that you didn't even think of them as 'having been automated away'.
Look at any commercial kitchen, any farm, inside any factory. Look at the grocery store, and its self-checkout machines. Look at parking attendants and toll booths. Look at the labour-saving devices employed at any construction site. Look at the demise of cottage industries - which used to be the economic driver of productive output, and are now limited to kitschy crafts on Etsy. Look at the cargo-container shipping yard, where a hundred dock-workers are now do the work that used to take tens of thousands. Look at the answering machine, and the utter insanity that is the modern automated customer support directory.
What hasn't been automated are jobs like janitors - where the work that they do is highly varied, and requires manipulation of different tools, in different spaces. We can automate a floor-mopping machine, or a vaccum cleaner, but we can't (yet) build a robot that will vaccum, and water the plants, and scrub a toilet, and clean up vomit, and refill the paper towel dispenser, all in one package.
Which aren’t automation or even new technology. The stores just cleverly moved the work that must be done to unpaid laborers. The self-checkout machines are an inferior tool with an escape hatch (customer service is on the way) to the real tool which must be operated by an employee.
I find self checkout very beneficial, as I frequently have few items to purchase, and there are more self checkout registers available more consistently than there would be staff.
For the same ~20 to 30 items I normally purchase, it takes me very little time to self checkout. For people with cartloads of items, they can go to the manned conveyor belts. For everyone else, they can go to one of the many self checkouts.
Except if they have never used self checkout before. Or if they have 15 items in the store's cart but they haven't brought their shopping bags so they take 15 minutes to fill their arms and pockets and hats with their groceries.
Those people are eating time saved by self-checkout !
I tolerate the young ones who comes in pack of 6 individuals, especially if only one of them buy their unique can of coke. It makes the queue looks longer than it is and I am out sooner than the SUCKERS in the other queue.
HA ! (/s ? /s)
Manual jobs that are relatively safe, close to home, steady, and easy to learn will probably have a lot of people vying for them, hence lower pay.
In the U.S. I believe it's through Social Security.
So as more jobs are automated and the base level of ability required to get hired goes up so does the need to expand who qualifies for social programs.
Productivity will be increasing with automation so it's a matter of distributing resources more efficiently and fairly than it is about making sure there is labor for everyone to get a job.
If you don’t live in the U.S. it’s hard to grasp just how hostile the political/economic system is towards the lower and middle classes. The idea of a UBI program to support people whose jobs get automated and lack the skills to work elsewhere is completely out of the question. A trip to the hospital can easily bankrupt the average American family ffs, but we can’t even fix that. There is zero political will to provide robust social programs here, despite any evidence that shows it might improve society.
Sorry for the rant. TL;DR if you look at the last few decades of politics in the U.S. you can see there is close to zero chance of the solutions you’re describing.
Being conscious intelligent creatures, humans ought to be able to learn from their mistakes and implement change accordingly.
This presupposes one refrains from defeatist refusal to engage those problems of course.
> Being conscious intelligent creatures, humans ought to be able to learn from their mistakes and implement change accordingly.
We sure ought to, I agree with you there.
The contemporary habit of envisioning people as some sort of disposable artifice is dehumanising at its core and just serves to rid those in lucky positions of superfluence from responsibility.
Society needs to learn to recognize interpersonal differences in terms other than self-serving judgemental dismissal.
Someone in a wheelchair 50 years ago could not do a lot of jobs vs now that's not even a factor in a software job.
Hopefully, we would reach a Star Trek-like utopia where people aren't required to work in order to live.
More realistically, the middle class just disappears, and we have nothing but the super-rich and the destitute left.
2. Capitalists require consumers to actually buy shit, owning capital doesn't get you anything if you don't have proles to sell to.
The "optimists" might not take notice because they've been in the layers that can keep playing musical chairs and adjust, and things might even seem good - especially when the pool of money available gets more concentrated.
What has been a slow gradual frog boiling process could very well accelerate with AI reaching this level though. I can't think of earlier developments being as widely applicable and moving as fast as this before.
I worry for my kids generation, and personally wish I was 10yrs closer to retirement.
Although I was completely unaware of this literature, I have spent the last few years essentially making the same point with regard to automation: you don't need a super-duper ASI to cause severe employment problems, all you need is something with an IQ of 85 that costs at most minimum wage to run (modulo robotics keeping up and being part of that price), and then 16% of the population are permanently and irreversibly no longer economically employable.
Only if you don't take IQ tests seriously — I think I've heard them dismissed as "only good at measuring your score on an IQ test".
If you do take them seriously, then by definition an AI which genuinely scores 85 (and isn't, say, just programmed to be good at that specific set of puzzles), then it must be able to do what a human of that IQ can do.
In this case I mean the latter, what with it being a thought experiment to demonstrate why we get problems well before having to ask if the AI can take over.
Practical issues like "a native speaker with IQ 85 is almost certainly more fluent than an IQ 160 polyglot on their 53rd, but for an AI they're all equal", or "we can easily hard-code in an algorithm for winning chess and another for doing calculus" likewise don't really matter.
> childcare / elder care, social work, basically anything that requires garnering trust and a lot of social interaction, but not necessarily complex skills
One of the problems right now with ChatGPT is that people trust it waaaay too much, and it's not really even trying.
> that doesn't necessarily make your skills worthless.
worthless != longer economically employable
The arts have significant social value, and I suspect may be to humans what tails are to peacocks; but at the same time, painters and novelists and musicians and actors and playwrights are infamous in their collective struggles to earn enough to get by on those careers alone.
Has anyone ever said this about any test? Are there any tests that are so general that anyone who does better on it is better at every single task than someone who does worse? Is such a test even possible?
I lack the skills to determine that, so I am limited to trusting the people who tell me the tests are "pretty good but not perfect".
For that specific example, athletics, there does appear to be a correlation. It may be any or all of A->B, B->A, and C->(A, B), but "healthy body, healthy mind" is a meme that appears to have a similar age as the concept of writing.
Finally, as I have been specifying IQ 85 as the "huh interesting" threshold rather than "high" IQ, I'll ignore your goalpost shift and say that your citation is me right now saying: I reckon IQ improvements in at least this zone (say +/- 0.5σ) is causal for better outcomes with regard to [avoiding injury, pushing performance limits, maintaining training regimes, learning techniques], all of which improve overall performance.
Assuming £10/hour and a cashier that can scan an average of 20 items per minute (which is probably 20 items in 20 seconds and then waiting 40 more for the customer to dig out their wallet), that's 20*60 items per hour = 1200 items per hour = 120 items/£1.
I assume it's mainly useful for other reasons?
A healthy society needs to have means to support people of all intelligence levels, we need the equivalent of jobs for ditch diggers, and those jobs need to be respected lines of work for people.
Work has more value than just money, and whether people like it or not, it gives value and meaning to peoples lives. The value we assign (and pay) for work shouldnt just be governed by the economic value it generates, it needs to be enough to meet basic needs, to engender respect.
The basic point I'll make here, using ditch digging is - have you dug a ditch? I have, its hard work, and mine was just a small one in my back yard.
Our default is "lots of babies"; women getting more education and economic liberty decrease that number. Poor places don't generally have significant educational budgets or opportunities for women, but as that specific set of things changes — education and freedom, not even always money — there is a corresponding change to number of births per woman.
It also means that rich couples in societies that lack women's lib still have more kids, even where you can infer high intellect without the convenient proxy of the women being allowed to go to university and giving you a test to score against.
That's not my point, my point is more that it's funny that people do things like going to college when this predictably causes them to have fewer children i.e. college actually hurts their evolutionary fitness.
That line of thinking is the beginning or the road to eugenics, and it goes no place good.
Also, everyone is thinking about evolutionary fitness. Some people just do it explicitly (and others do it by instinct/intuition)
What about Down syndrome adults, who are reliant on state benefits?
In case it's not clear, I don't support the killing of disabled people at any age, but I understand that different people approach these questions from a different set of assumptions.
My point is that adding scare quotes around "that's eugenics" doesn't stop it from being an accurate description, so your question is not the gotcha you might think it is.
But a woman aborting a Downs baby is not the state killing anyone, and most pro choice people would say the woman isn't killing anyone either, since the fetus isn't a person in their view.
Well no, if we still had the same gross economic output as 1000 years ago quality of living would be much lower.
The point of competition is that it drives up total economic output.
Of course it sucks for the people who get outcompeted. Certainly getting them productive again in another capacity is good for society. Letting them simply idle is not.
And on the time scales we're looking at here (millions of people losing jobs within a decade or so), population growth is not a relevant concern, and does not explain why people are worse off despite more automation. Not to mention that population growth is already negative in most societies most affected by automation.
Just wanted to note that I didn’t propose a particular solution—you seem to have responded as if I’ve just said “we should have no automation,” which is not what I was proposing at all—it was just an observation about where things have ended up.
We could imagine lots of ways to not throw people to the wolves when their jobs were automated away, the idea of just continuing to pay them is sort of a lower bound (in the sense that obviously the economy could bear it, because it had just been bearing it, right before the automation was implemented).
My grandfather is about as smart as I am, but he never went to college (nor did well over 90% of his cohort). Now about half the population goes to college, and more than a third graduate.
That's a massive transformation in two generations, and its exacerbated by the degree to which it segregates. I've never eaten lunch with an adult, my age or younger, who didn't attend college.
https://slatestarcodex.com/2014/09/30/i-can-tolerate-anythin...
It is completely wild to me how effectively society has self segregated to the point of just straight up never, not even once, interacting with the other group without even realizing it.
They both have college degrees though. I have a few friends who dropped out of college, but none that never attended.
Remember, even the army during war time, which requires people fulfilling botho f these services won't hire people with such a low IQ.
First of all Peterson’s claim is 83, not 85.
Second, The military does not test IQ. It has a different test with a decent correlation. This test scores in percentiles however. And the bottom 10% are not allowed in. This is not the same thing as 10% of all people cannot get in, as those taking the military test are probably a lower IQ subset on average.
Probably not by much. I would guess the military is current cutting off the bottom 8-10% of the population.
I can’t find much in the way of hard stats but I still feel pretty strongly that a 10th percentile individual is more than well equipped to be a delivery driver of a warehouse worker or whatever.
This is because we decide it's better to pay people to do nothing than to clean streets and medians and countless other things that need done.
We are artificially selecting the worst people to reproduce and dooming ourselves.
The fallacy of eugenics is that 'smart' people have 'smart' children and the same for 'dumb' people.
I think 90% of this forum believes themselves to be smarter than their parents, so maybe pump the brakes.
I don't have the time to work through the math concretely, but intuitively, imagine that the intelligence of parents is normally distributed, and the intelligence of their children is a tighter normal distribution around the parents' intelligence. Set the cutoff for "participating on HN" to to some threshold. Given the right set of parameters, most of the "weight" in the aggregate child distribution above the threshold will come from the long tails of the child distributions of parents in the fat part of the parent distribution.
Again, this is the core fallacy of eugenics.
How many current US PhD's are the children/grandchildren of immigrants from underdeveloped countries who lacked education and worked manual jobs (or maybe even went through extended periods of unemployment)?
EDIT: To be clear, I know Western culture has pushed a narrative that intelligence is an intrinsic thing that anyone can have, but in practice, it's a reflection of learned skills like reasoning and abstraction. That's the entire point of education, if you don't practice those skills (with some exceptions), you are likely worse at them.
https://www.research.ed.ac.uk/en/publications/how-much-does-...
Why are you conflating immigrants from underdeveloped countries with low intelligence? The parents could be have been (likely were) highly intelligent, just lacking the opportunity or access to education.
The current US PhD's that are the children/grandchildren of immigrants aren't a random selection, they are the product of smart motivated parents working the system (immigration system, education system, admissions system) for and with their children.
The core fallacy of Eugenics is not that intelligence is heritable, it is that you can avoid terrible moral hazards when you try to meddle in this stuff.
I'm trying to say that regardless of who your parents are, your intelligence is primarily driven by your education (and your attitude towards education), with slight variance for genetic factors.
Therefore, we should not be anxious about "dumb people reproducing en masse" (as GP was).
We should be worried about educating everyone equally and fairly.
I don't think 50-80% heritability should be hand-waved away as "slight variance". I think what you're talking about is the manifestations of the intelligence (i.e. you can't be a rocket scientist if you never had the chance to learn about it), but not the intelligence itself.
Also please cite actual studies rather than Wikipedia. You’ll capture more nuance.
We conclude that there is now strong evidence that virtually all individual psychological differences, when reliably measured, are moderately to substantially heritable
For example, citation 6 implies nearly every personality trait is inherited, which seems aggressive to me.
Citation 7 is a book about how the genetic links get stronger with age, which seems also odd to take at face value.
Citation 4 is a study of 3k white (they excluded non-white people intentionally) British people born in the 1920's titled "Genome-wide association studies establish that human intelligence is highly heritable and polygenic," but their results could be restated as 'distant members of the same family clan (in a society famous for inter-generational class structure) score similarly on aptitude tests, but we can't explain it with our data.'
Anyway, as great a source as Wikipedia can be, I wouldn't take its summaries at face value, the three above examples lead me to believe (like in a lot of social sciences) there's still a lot of work to be done to actually get to answers.
But, again, circling back to the problem here - even if 'smart' people lead to 'smart' children, society as it stands is awful at actually identifying and rewarding smart people (i.e. jobs aren't doled out via IQ tests).
I hope not. I saw a pretty funny meme on Reddit today that went something like "50 years ago car manuals explained how to adjust the valves in your engine. Now they ask that you don't drink the battery fluid".
I think another big downside of so much technological progress is that people no longer need to understand, and get the benefit of learning, how things work. Yeah, it was a PITA that 35 years ago I had to muck with config.sys and autoexec.bat files, but it helped me have a better understanding of how things worked under the covers. There was some discussion on HN recently about how many Gen Z folks are woefully unprepared digitally for the workforce because they are so used to iPhones/iPads/apps, etc., that you don't know how thing actually work - like people not understanding what the file system is.
How do you smelt bronze? How do you grow crops? How do you build a lean-to?
Knowing how to do all those things does give you a very intrinsic understanding of a lot of things, but it's probably not wise to force everyone to learn those things.
EDIT: Apparently you don't smelt bronze, you smelt pure metals out of ores. lol.
What makes you say that's a fallacy? Everything I've read has shown intelligence to be highly hereditable. I'm not saying we should make policy changes based on that, but the science behind intelligence being (at least partially) hereditable seems pretty conclusive.
Then you read the paper and you see:
A total of 3511 CAGES (2115 females) participants had DNA extracted...Individuals were excluded from this study based on unresolved gender discrepancy, relatedness, call rate (≤ 0.95), and evidence of non-Caucasian descent...
Analyses of individual SNPs and genes did not result in any replicable genome-wide significant association. A gene-based test for association27 showed one genome-wide significant association (P = 9.2 × 10−7), with formin-binding protein 1-like (FNBP1L) on gf (Supplementary Figure 7). This single genome-wide association result for FNBP1L did not replicate in the independent NCNG sample (P = 0.211, gene-based test)...
Despite the fact that no specific genetic variants have been robustly associated with human intelligence, apart perhaps for APOE at older ages34,35, our results show for the first time that a substantial proportion (approximately 40 to 50%) of variation in human intelligence is associated with common SNPs (Minor allele frequency (MAF) > 0.01) that are in LD with causal variants...
Only 1% (approximately) of the variance was explained in the prediction analysis due to the individual SNP effects being very small and therefore estimated with much error, which detracts to a great extent from the accuracy (8-11%) of the prediction equation30. Our finding that 40-50% of phenotypic variation is explained by all SNPs is fully consistent with the low precision of a predictor based upon a discovery sample of ~3,500 individuals; estimation of the SNPs’ effects is different from prediction accuracy.
So basically, what they did, was take a bunch of British people born in the 1920's (a time/place famous for its rigid inter-generational class history), and found that members of certain family trees (that's more or less what it means to be British and have similar SNP distributions) scored similarly on tests, despite the fact that there's no real model to express why.
That and twin studies (which also have their obvious limits) are about the extent of current "evidence" for inherited intellect. I'm willing to be proven wrong, but 'conclusive' is hardly how I would characterize the state of true evidence-based research.
Seems to suggest that intelligence has a lot to do with genes. Why do you think that part supports your view that intelligence doesn't? Do you have any evidence that kids doesn't inherit any intelligence?
I'd be impressed if they saw the same thing across two populations where we know people aren't related, but I'd argue they just back-solved Britain's class structure.
One has to think critically in science.
>Do you have any evidence that kids doesn't inherit any intelligence
Well, if intelligence is inherited, then why are IQ tests going up over time (and therefore being recentered on 100 constantly)? Doesn't that imply learning is the driver?
You're not owed that your profession stay relevant, and so you always have to keep an eye on trends, and if needed, adjust course.
I think Jordan Peterson quoted the stats as people with IQ under 83 are completely unemployable i.e. even the army can't get a net positive out of them. I found that deeply concerning in several ways.
IMHO we need to work on reducing the cost of living so people can mostly opt-out without losing their home and dying. Instead it's always push harder, more growth, etc...
If your programming job consists of simply receiving a greenfield spec and banging out the code for it over and over, LLMs might be an issue. But most jobs are not like this. The bulk of the work, as always, is in dealing with the people at either end of the process. And knowing what the computer is doing and how it works will continue to be invaluable when an LLM writes logically perfect code that happens to run slowly or do the wrong thing once it makes contact with the real world.
"There is always a global shortage of developers who know what they are doing"
At the time, I wasn't sure if this was the case but two things have helped me believe that this is true:
1. a few years ago there was a stat going around on HN that something like less than 5% of the US population could even write a Hello World application in ANY language
2. 15+ years of working in tech and 5+ years of being a hiring manager have shown me that even people with great looking resumes often fail the Fizz Buzz test
I agree though, off-shoring could have been the cheap labour which displaced a lot of devs, but I actually found a lot of work cleaning up off-shored applications and made great money doing it. It's perfectly possible that some horrible amalgamations of software will be puked out by AIs for a few years and ultimately create as much work as it displaced.
Considering how important it is to not have bugs - we're extremely unlikely to have an LLM writing code at an acceptable level of quality, and anyone being qualified to verify that quality beside an eng, before I'm retired.
But I'm confident many 22 year-old new grads will reply to tell me how wrong I am, and how I'll be starving to death under the burden of unemployment by next week.
What would be great is for senior ICs to push into non-IC roles (ideally: ICs can use AI, so you need less ICs but this is unproven). Technical skills and knowledge applied to writing requirements cuts down on the number of conversations around those requirements. Then you get closer to a "greenfield spec -> bang out AI-generated code" situation because you're not slogging and negotiating through requirements written by non-technical people.
I think LLMs are going to be very powerful for brownfield coding too.
So here's the thing; what in that list of non-codegen tasks is more complicated than that which GPT can already automate?
I mean... we did put nearly all the horses out of a job. Ditto oxen and other beasts of burden.
Recording and mechanical reproduction did cause the social(!) and economic value of small-time talent in the arts to plummet from "pretty high, actually" to "basically zero". That never recovered.
To no small degree, the US has coped with the effects of automation with huge public spending on the military (a massive blue-collar jobs program, effectively) and a painfully-inefficient healthcare system (a massive white-collar jobs program). Both are post-war (as in, WWII) developments. I have a feeling our optimism about the effects of further automation on unemployment would be rather bleaker without the action of those two very-expensive systems in mitigating the effects of past automation.
[EDIT] Oh, you can add the effects of pushing a whole lot more people into college to the list of expensive things we do that mitigate the effects of automation on employment. Keeping millions more people out of the workforce for an extra four years or more is a pretty big deal. Also a post-war development (the GI Bill was what kicked off that trend)
Exactly. The buggy drivers became tax drivers, the horses became glue.
The automobile was a great invention for the participants on the right side of that bell-curve, and absolutely horrible for the participants on the left side of the bell-curve.
There's no guarantee, universal law, or dogmatic hit-the-brakes switch which will ensure that (without a major economic adjustment) further technological development won't put an ever-growing fraction of people to the left of the 'useful enough to be employable for a living wage' side of the bellcurve.
Don't forget farmers. Most americans were farmers in the early 1900s. Today it's about 1%.
Sure, jobs were eventually created for the unemployed farmers but "robots taking all the jobs" was a cataclysmic event in the early 1900s. You could argue it was the driving force for much of the 20th century. WW1 was primarily a result of european and american societies having excess population to "thin out". The excess population was also a reason for the anti-immigration movement that dominated the first half of the 1900s.
"Robots taking our jobs" isn't the end of the world. But it can be very painful.
No. Some actual reasons for WW1: Franco-German animosity, the famous web of alliances, advances in weapon technology and accompanying arms races, nationalism, and cold geopolitical calculus (Germany thought their power relative to the other Great Powers would decline in the future).
Source?
This argument about automation is no different. We don't know what is going to happen because of AI. We don't even know where AI is going to go as a product/feature/whatever. It could totally make obsolete an entire class of lower-earning white collar workers whose main contribution is writing things. On the other hand, it could fizzle out and we could be comparing Bing Chat to Clippy a few years from now and laughing at all the investors who wasted their money on the next big thing after crypto.
Pointing to times in the past when people feared automation, and showing how those fears did not pan out, is not a good argument. Everything in history is contingent, everything could have gone a different way if a few people made different decisions at a few key times, and fear of automation may have motivated some of those people to make the decisions that prevented automation from destroying all the jobs.
To put it another way: If you play Russian Roulette with a six-shot revolver and you survive five rounds, should you be more or less likely to believe that you will survive the sixth round? People who look to the past and say "bad thing that was predicted did not happen then so it won't happen now" are guilty of thinking that if they survived five rounds then they will definitely survive the next one.
The primary issue for me is the rate at which technological progress is moving is making it ever more difficult for humans to adjust. I mean, suppose that in 10 years we have fully capable self driving cars. I'm not necessarily saying that's going to happen (in fact I actually think it won't), but I do think it will happen at some point. Right now, truck driver is the most common job in that vast majority of US states. When self driving cars do eventually come, all of those jobs will disappear pretty quickly. Do you honestly think someone who has been driving a truck for 20 years will be able to "retrain" quickly?
And it's certainly not just truck drivers. Count me as one of the people who are astonished by ChatGPT's programming abilities. While I don't believe it's "there yet", I was pretty amazed that I was able to ask ChatGPT to program something for me, and it did quite well but the code had a subtle bug, and then when I pointed out the bug ChatGPT was able to fix it itself. And this wasn't even on the GPT-4 version. This stuff only gets better with time.
There is no guarantee that the overall rate of technological progress will continue to increase. The opposite could well happen. Right now we are seeing the effects of multiple S-curves stacked on top of each other but progress is already slowing in many fields because the low-hanging fruit has been picked. For example, civil aviation has only eked out a few percentage points of improvement in efficiency and safety over the past couple decades; the latest airliners are actually slower than their predecessors. Or in medicine we seem to be running out of new small-molecule drugs to find and thus on average with all the failures it now takes more work than ever before to bring a new drug to market.
Imagine an LLM running locally on your car, refined to pick your lane based on the current traffic conditions and upcoming map data.
Many of the problems with self driving come for a lack of basic reasoning ability, an ability LLMs seem to emulate quite well. For example you can lay out driving scenarios to ChatGPT and get reasonable responses back.
It’s not as simple as running a Llama instance on your Tesla, but I expect language models will have had a big part to play when we finally get self driving to work.
Some of the underlying core technology could theoretically be used in that way by taking input from multiple videocameras and other sensors. But that approach is a dead end for consumer product development until someone solves the testing and explainability problem. Current limited AI systems are too unpredictable for safety critical applications. Just because AI software seems to work in testing doesn't mean you can have any confidence that it won't do something wildly dangerous in a novel real world situation. Vehicle manufacturers and their insurers will not take on the high potential liability even if autonomous systems are on average safer than human drivers.
Natural language is useful for systems with discrete elements and complex rules, which seems like a good fit for parsing intersections, construction zones, things like that.
I also disagree that the explainability problem needs to be solved before self driving cars can be implemented. The effects of drugs are inherently non-deterministic, but the FDA still lets companies develop drugs and ship them to customers. You test self driving cars the same way we test drugs: by collecting data and calculating the probability of harm.
Trains long predate the truck, and allow a single driver to drive hundreds of cars. But the major advantage of trucks is last-mile logistics - going from the depot where large volumes of cargo pass through to the specific place the cargo is needed. The number of truck drivers, and of future truck attendants, is proportional to the number of stops, not the amount of material being delivered. If anything, removing the bottleneck of requiring a person with a CDL to make a pickup or delivery will cause the number of truck attendants to be much larger than the current number of truck drivers (as already seen for example with Amazon delivery services, UPS trucks, etc).
I don’t have strong opinions on the ratio of attendants to trucks. You may be correct.
The people on this forum who go on and on about "we have gone through this before and society kept going, people were fine in the end" are half right. Society as a whole did keep going and people outside of the Rust Belt weren't directly affected.
But what is different this time is that this is going to annihilate 70-90% of jobs in the "knowledge sector" and it is going to do so very, very quickly. This will knock out most of us comfortable engineers, but also middle management, consultants, journalists, writers, artists, paralegals, influencers, wall street traders, maybe even teachers, lawyers, and doctors to an extent.
I don't know what the US will do with the middle/top of the income brackets all across the country suddenly facing mass unemployment and then the subsequent race to fill service or trade jobs. How is a software engineer who got a 30 year mortgage that costs $3k a month going to survive going from a $180k a year salary to being an electrician apprentice at $50k a year? And that is in an environment where this hypothetical former engineer was the top candidate out of 100 other former engineers vying for a spot. GPT-4 will kill many people via unemployment unless it is paired with a UBI, or at least that is my pessimistic view.
And it isn't going to be a high tech utopia, because a small group at the top is going to capture 95% of the wealth generated.
I would predict serious social unrest. The problem is people won't understand why they don't have jobs, so they will blame x country or z group for their problems which will lead to extremism.
I would argue many of the major historical events in the US, such as civil war, great depression, WW2, great recession were actually connected to automation requiring a reset of society.
And a lot of the remaining jobs went to Mexico.
The man is going to ask it for explanations for days, throw something together that kinda works and hire an expert 'consultant' that should have been an employee, when it inevitably breaks or requirements change.
Who would bet the legal success of their company on the 'truthiness' of ChatGPT?
The skilled knowledge workers just got upgraded to a better set of tools and the layman is as equally clueless as ever. C++ didn't get easier to read, we just got a robot teacher.
But the paradigm that websites, iphone apps and windows programs are the breadwinners, can very well change.
If we're really getting on the 'gpt engineering revolution' bandwagon, then it must logically bring quantum computing, metaverse, more cybersecurity vulnerabilities, more sophisticated internet backbone and better computing hardware with it. (And fusion power and flying pigs, ect idealism).
A revolution for one, will bring a revolution for everything and that involves an awful lot of developer-hours, once we have changed the target market from "http-websites and apps" to the next market.
When I started working in FinTech at a big bank, our Managing Director mentioned that he was told:
"In 2 years, there will no longer be the job of 'trader'"
This quote was from 2005 and he was sharing it with us in 2007.
While, yes, there are a LOT less traders than there used to be, it is certainly not zero.
That is the core of the problem, which has been iterated several times in this thread.
Where do those traders that used to be there go? Historically, switching careers or positions has not been insanely difficult. But as there are fewer and fewer jobs, it becomes more and more difficult for people to retrain.
Obviously there are tons of practical considerations to deal with, but in a theoretical, general sense, is “left behind” a problem? Shouldn’t lifelong learning be a thing everyone is indeed capable of?
This is only an issue if either there are genuinely not enough jobs in the field lots of people are trying to go into, in which case people should not be encouraged to go into it; or if older people demand higher compensation for the same skillset as younger people based purely on age, in which case that doesn't seem fair anyways.
And realistically, a mid-life learner is generally the better option to begin with - it's nearly impossible that someone after years of working has picked up no transferrable hard or soft skills and made no network connections of any value.
That's just a polite way of saying "willing to work for less" which again means either somebody's asking for too much or someone's being exploited, which is the real issue.
Obviously its mere anecdote, but every company I have ever worked for has gladly hired middle aged entry level employees. If anything they have been preferred.
Why is the burden placed on the employer to grow their employee and not the human being to grow themselves? In a generic sense, it could be outside of the purview of the employer to consider the employee's future in that specific way (in other ways it makes more sense, like health and happiness while employed).
It seems oddly paternalistic, to me, and once you remove that moral presumption doesn't the conversation become more complex?
Even if we did presume that jobs should be required in society, how would it even work for a job creator to preemptively know what skills were needed with enough time to ensure at your current job you were trained for future jobs? Why would that fall to your current employer?
Leaving people behind is problematic because not everyone can keep up or has the time and means to do so. Should a single mother be expected to put in hours every week learning some crap on top of working and raising kids or can that person just breathe and deal with their other problems? Full time house parent that loses a spouse and has to start working. Someone in a low income house that can’t afford professional courses/books or even has access to a computer.
You ever see those commercials about kids that can’t focus on school because they’re hungry? Yeah it’s not just them. When people are deprived of life essentials or dealing with personal problems studying is the last thing on their minds.
Here we have people in tech burning themselves out and we want to export that mentality to everywhere else instead of seeing it as a problem.
Ray is a trans humanist who believes we’re on the cusp of eternal life via the singularity.
One of the things he is relying on is that people will keep up with technology through computer brain interfaces like Neural Link but faster.
We should be seeing this come to market sometime around now.
It’s not here so I guess that’s not good.
https://www.youtube.com/watch?v=ykY69lSpDdo
Worth a watch…
https://www.smbc-comics.com/index.php?db=comics&id=1968#comi...
We "ought" to share the rewards and efficiencies of automation with the people that it displaces, rather than solely with the owners of capital.
For example you place the burden of “learning new things” at the feet of the employer, but you don’t really justify that. Why, in a general sense, can people not learn on the remaining hours of the day?
Your comment is riddled with moral presumption and I think it’s worth figuring out what those are.
But yeah, I think there are many books on spaced repetition and deliberate practice such that it's hardly a secret.
Besides, does everyone have a right to every single one of those things you've listed all the time? I wouldn't argue so, considering we're talking about people who are providing less value over time to their fellow human being.
When time conflicts exist, you must prioritize, and I would imagine "remain able to contribute meaningfully to society" (with caveats around capability) ends up being more important than, "maintain multiple hobbies".
And no, I don't think people are, generally, perfectly optimizing their time.
If you did indeed live in such a world, would your opinion change? If people did indeed have ~30 minutes a few times a week to keep up with their field's growth (and that's all it took), would that matter to you?
Trying to find the theoretical line for you.
I'm not sure where you're getting those things, these aren't "either/or" suggestions, but "both/and".
Nothing about that is new. It has been a common fear and very real issue since the industrial revolution took hold.
Programmers aren’t going anywhere. ChatGPT doesn’t “understand” things. It optimizes its output for some set of scoring metrics. It sees from it’s training data that tokens score better (on average) in a particular order or that some tokens are associated with some sentiment. It doesn’t see english words.
With code generation the biggest benefit is generating code scaffolding. That’s nice and all, but certainly NOT where the heavy lifting happens during system development. The scaffolding it shits out usually needs to be heavily modified by a human. The amount of capital, compute, memory, and energy it requires is insanely inefficient for generating boilerplate code. Don’t even get me started on the privacy implications…
OpenAI will make programmers more productive by reducing time spent on trivial/mundane tasks. That’s awesome! However, it’s not going to replace the SWE industry.
"Tell me about the structure of the solar system. I want you to only use English words of Germanic origin; avoid using words derived from Latin or Greek. If there's no alternative, make up a new word by calquing the original into Germanic roots."
Just to give an idea of the output GPT-4 can produce from this, here's a snippet:
"... beyond the worldrocks, rockbits, and icebits, the sunstead also has spacegrit, small bits of rock and metal that fly through space. When spacegrit enters Earth's air and burns up, it makes a lightstreak that we can see in the night sky. If a bit of spacegrit reaches Earth's ground, it is called a starstone ... "
Rockbits are asteroids (and it does consistently use this term, like all others, throughout the composition). Can you explain how this invented word, or even its individual components, would be "associated with some sentiment" in the training data that leads to this output, if the model doesn't actually understand, on some level, what it's asked to do?
It doesn't have to be perfect to replace developers. If every developer in a company becomes 5x more productive but you can only sell 3x harder, something might still have to give.
Even if GPT-4 doesn't quite cut it, what about GPT-5? What about the crazy, off-the-wall concept that comes out of some university and completely revolutionizes the whole thing again by 2024? Stanford is already demonstrating an inclination for disruption. There is an incredible amount of energy, attention and capital being poured into this space right now. If all acceleration forces were removed, I think we are still going to wind up somewhere crazy.
We certainly aren't ever going to go backwards in capabilities. It will only improve - even if just marginally. I've got GPT3.5 suggesting very high quality methods for my codebases. You could take 20% of the capability away right now and I'd still consider it a revolution in productivity.
> It makes me believe most of the people haven’t taken the time to evaluate the output of these LLMs.
How much experience have you had with GPT3.5/4?
People seemingly like to fall prey to this particluar incarnation of over-generalization. For obvious reasons, as they simply don't know what to do about the impending demise of the established economic order.
Where are the essays proposing solutions?
People will continue to do things for so long as people prefer receiving money to doing nothing, but people won't be doing what we're doing now for long.
See who's still laughing in a couple more years.
I will indeed laugh when the lawyer market collapses shockingly soon.
One issue here is the resulting fragility of the economy. If crucial nodes of competence were to be wiped out by disaster, domino effects could have system-level implications as recovery might take too long.
On a more personal level, people more and more lack the competence to understand non-local connections of the world they live in. Resulting in them not being able to judge consequences and making according mistakes.
A whole generation thinks it's normal that every thought is logged, stored, censored, sold and supervised by a third party.
Get out pencil and paper!
1. Money encourages people to help each other by facilitating reciprocation: I do something for you and you pay me, someone else does something for me and I pay them, etc.
2. Private ownership fosters accountability: People maintain their own houses because it's theirs. See also tragedy of the commons.
If these purposes are not being served for a majority of people, they will reject capitalism. But the right regulatory environment can ensure that they are. In a distant future with extreme levels of automation, people will still need help from one another and people will still want to own their own things. We just have to make sure it's possible. Progressive tax schemes and UBI, possibly in the form of equity, could accomplish that.
"It is much easier to imagine someone losing their job to a new technology, than it is to imagine many people gaining jobs that haven’t been invented yet."
When I was in school in the pre-Internet era, I had one very insightful teacher comment that many of the jobs we may end up working in hadn't even been invented yet. Turned out to be true in my case.
So, we change that by requiring corporations that want to continue indemnifying their owners from damanges become workfare programs and less profit oriented. The current regime sure aint working, and I think it is fair to expect businesses that benefit from public infrastructure have some skin in the societal game as well.
I love the huge robot in the shadows that just says AUTOMATION across its chest.
What new things will humans do in the face of automated intelligence and creativity?
Though ignore the 'upload nirvana' scenario, which is perhaps even more scary, given what we now know about big tech, capitalism, government control, etc. Yep, we really could give criminals 2000-year sentences to purgatory, if we uploaded them.
Because the population is so large and the planet so far into ecological overshoot that there won't be any jobs to take... in 100 years? :)
Correlation does not equal causation but you cannot deny the possibility of causal connections here.
Generally the lower tech the area than the more children you have the better off you are economically. Mainly because children can function as manual labor. With much of manual labor replaced with automation, nowadays having more children is more of a burden.
Robots have not only been about to take our jobs. They have been taking our jobs. It's just it's so subtle it's hard to see it.