Generative AI Set To Affect 300 Million Jobs Across Major Economies, Goldman Sachs Says (arstechnica.com) 114
The latest breakthroughs in artificial intelligence could lead to the automation of a quarter of the work done in the US and eurozone, according to research by Goldman Sachs. From a report: The investment bank said on Monday that "generative" AI systems such as ChatGPT, which can create content that is indistinguishable from human output, could spark a productivity boom that would eventually raise annual global gross domestic product by 7 percent over a 10-year period. But if the technology lived up to its promise, it would also bring "significant disruption" to the labor market, exposing the equivalent of 300 million full-time workers across big economies to automation, according to Joseph Briggs and Devesh Kodnani, the paper's authors. Lawyers and administrative staff would be among those at greatest risk of becoming redundant.
They calculate that roughly two-thirds of jobs in the US and Europe are exposed to some degree of AI automation, based on data on the tasks typically performed in thousands of occupations. Most people would see less than half of their workload automated and would probably continue in their jobs, with some of their time freed up for more productive activities. In the US, this should apply to 63 percent of the workforce, they calculated. A further 30 percent working in physical or outdoor jobs would be unaffected, although their work might be susceptible to other forms of automation. But about 7 percent of US workers are in jobs where at least half of their tasks could be done by generative AI and are vulnerable to replacement.
They calculate that roughly two-thirds of jobs in the US and Europe are exposed to some degree of AI automation, based on data on the tasks typically performed in thousands of occupations. Most people would see less than half of their workload automated and would probably continue in their jobs, with some of their time freed up for more productive activities. In the US, this should apply to 63 percent of the workforce, they calculated. A further 30 percent working in physical or outdoor jobs would be unaffected, although their work might be susceptible to other forms of automation. But about 7 percent of US workers are in jobs where at least half of their tasks could be done by generative AI and are vulnerable to replacement.
In a related news... (Score:2)
Re: (Score:3)
Honestly, ChatGPT AI would probably make better financial market predictions that Goldman Sachs.
Re: (Score:2)
automobiles.
pen and paper.
hand held calculators.
name 3 inventions that were to replace workers.
but.
when it became cheap enough.
was then imposed on workers.
giving workers.
more work
Re:In a related news... (Score:4, Insightful)
Re: (Score:2)
The fact is, automobiles are just carriages using a different source of horsepower -- they're not as revolutionary as you think.
But the transition is still pretty disruptive. Entire industries around making horse-power available ubiquitously were pretty rapidly made much less relevant. Maybe the net number of jobs were the same but it's not certain that a farrier and a hay farmer could readily switch to being an auto mechanic.
Re: (Score:2)
simple.
1 horse does the work of 6 men
when the yoke for horses was invented.
the romans set free and or banished 5 slaves per horse
Re: (Score:2)
A couple of points, there was usually a delay as automation changed the job market, with the original industrial revolution causing high unemployment for a couple of generations before enough new jobs appeared to fix things. Also there was a new world for the unemployed to go to and do things like homestead.
The other thing is employment never became full again. Used to be that almost everyone worked, you started working at 4-5 years old and basically worked till death. The wave of automation a 100-140 years
Re: (Score:2)
learning.
atlanta morrissette said it well.
you scream.
you learn
Re: (Score:2)
It's about damn time... (Score:3, Interesting)
...someone was honest and highlighted the actual impact of AI on human employment. I was really getting tired of ignorant humans treating this like yet another technical revolution that might put a few humans out of work with "go get an education" as the perpetual excuse to restoring self-worth and value in society.
Greed funding AI doesn't seek to make humans temporarily unemployed. It seeks to make humans permanently unemployable. In favor of a 24/7 employee that never complains about needing sleep, health insurance, or vacations. Deny it all you want, but deep down you already know how Greed operates; the same as it has for thousands of years.
No, your Governments don't have an answer for this, nor do they really give a shit. Many current leaders will be dead and gone long before their votes and actions bring forth permanent consequence. We're creating intelligence this time. Humans really have no fucking idea how fast this will all happen. You're no longer going to be the smartest thing in the room soon. Good luck predicting it, and you meatsacks better maintain control over that power cord.
Re: (Score:3)
I've said it before, but the only "jobs" lost will be "cog-in-the-machine" jobs. You were always replaceable, whining about Stable Diffusion or Midjourney just proves how much you don't value your own skill.
More jobs will be created in the long run, because people won't have to pay for artists that are unavailable already. (Like if you wanted to get started in virtual streaming, you have to pay around $5000 for a competitively featured model) Those people at the top, making big money? Will keep making big m
Re: (Score:2)
All you have to do right now to be better than AI art generators is to have some general idea of composition and creativity.
Yes. Right now. It's going to keep getting better very very quickly, much faster than we as a society are prepared for.
Re: (Score:2)
Not only that, but it can already do it at speeds that no humans could ever dream to match. There is a revolution coming, the concept of money is going to be stretched to its limits before it breaks.
Re: (Score:2)
Bollocks.
There will be idiots that believe that there are wide-spread jobs for LLMs, but their training is suspect, their results are suspect, because their origins and models are suspect.
And some will use them, get laughable results beyond the fallacies that have already become flatulence news, become deeply embarrassed for their time waste, and move on.
Yeah, self-driving cars by 2022! This is about sucking in VC money to do modeling. Few people will be replaced, and what you see as interactive bullshit ge
Re: (Score:2)
When AI is going to start taking over jobs at scale is not clear, but it's going to happen. 30 years? Maybe. 100 years? I would be surprised if it took that long. The only question is whether we'll prepare for it in a way that prevents societal collapse. Since such things tend to improve rapidly and sometimes in sudden spurts, it's not looking too good.
Re: (Score:2)
You give too much credit to bad software.
People also evolve. And make mistakes.
LLM isn't going to fix your toilet. It won't rescue a drowning youngster in a pool. It might know when to flip a hamburger.
Maybe it can harvest almonds, with help. Or tell you when that avocado is ripe. It's up to you to eat the avocado.
Goldman Sachs is full of mirth, waiting to see where it's next trillion is coming from, and suddenly, the sky is falling. There are many jobs that Caterpillar has learned to take over. This is ano
Re: (Score:2)
You give too much credit to bad software.
Not at all. I'm talking the really really good software in the future.
LLM isn't going to fix your toilet.
The physical capabilities of robots will continue to improve too, though likely at a slower pace.
Could it go out of control?
I'm not worried about that, I'm worried about what happens when computers and robots can do nearly everything a human can do and just as well or better. There is no room for "oh other jobs will be created" at that point because you know what's going to do those other new jobs? Robots. Because they can, and they'll be cheaper.
My point is that people have learned to take such developments in stride
Such a develop
Re: (Score:2)
Robots can't even sweep a rug. Self-driving automata is an oxymoron.
Over many years things will improve, incrementally, in fits and spurts. Regulations will arise. Science fiction dystopia will continue to evolve, filling social media addicts with visions of hell, and VCs with visions of economic grandeur.
I have implied that none of the future possibilities is impossible, so much as the hype today is largely just that. Goldman Sachs. VC with visions of grandeur and wealth. Will that wealth trickle down?
Adap
Re: (Score:2)
Will that wealth trickle down?
Indeed, that is the question.
what more could go wrong?
What you describe is happening with relatively low unemployment. Unemployment in the Great Depression was, what, 25% or something? What happens if it's 50%, or 80%? Hopefully this will happen slowly enough that we can figure out a different system than "have a job or starve on the streets". If it doesn't, it will make homeless camps look like a child's birthday party.
Re: (Score:2)
You project numbers just like an AI response, pulled from an unprovable place that diminishes the value of your response.
Indeed no one can predict this, ultimately, but Goldman Sachs (remember the first point?) is trying to suck the air/money out of the discussion. My attempt was a reality check. Reality reeks of facts.
The fact is that it's all hype right now, with some charming (said with sarcastic venom) candidates.
Blah.
Re: (Score:2)
Indeed no one can predict this, ultimately
It's really easy to predict based on a simple assumption. Do you assume that progress in these areas will halt at some point, or that it will continue more or less indefinitely? If the latter, then eventually robots and software will have superior performance to humans at all tasks. And I have not heard of any plausible mechanism by which the former assumption could be correct.
Re: (Score:2)
Foisting a prediction where the assumptions are faulty from inception doesn't work.
Today's AI is primitive in the extreme. It tries to communicate, which delights people. It's based on false information, cannot show proof, and is not re-trained. It's faulty by definition.
Extrapolating based on a false premise is itself a false premise.
See various fallacious arguments to understand the problem.
Re: (Score:2)
Foisting a prediction where the assumptions are faulty from inception doesn't work.
The assumptions that it either will or will not continue improving are both faulty?
Re: (Score:2)
their training is suspect, their results are suspect, because their origins and models are suspect.
Even if that was true, it's irrelevant. Those critiques may apply to current LLMs, but even granting they're true, it's a fallacy to apply them to all possible LLMs.
The important part, the technology itself, is not suspect. Using the technology, other LLMs can be trained using some open and auditable process, with public origins. At which point their results would stop being suspect, and your complaints disappear into thin air.
In the end, if you don't trust the LLM, you can always verify the results.You can
Re: (Score:1)
Re: (Score:2)
People are using text generators to come up with ideas — you can simulate creativity by asking for a bunch of options, then asking for each of them to be fleshed out in multiple ways, and picking the good ones. You can even crowdsource the selection process on social media, there are tons of groups which can be used that way.
Re: (Score:2)
An AI Can Not Read Your Mind. It understands nothing.
AI doesn't have to read your mind. It's rather busy replacing it. And you're also talking about today's AI. Not tomorrows. Has nothing to do with "skill" other than your human inability to work 24 hours a day.
Still not quite getting it, I see. Paralegal work is being replaced by AI these days, so not just the cheap "cogs" being assimilated. Sure more jobs will be created. For AI and automation to consume, not humans. Why the hell would Greed even consider hiring those pain in the ass meatsacks when a
Re: (Score:2)
Re: (Score:2)
I didn't realize that it was okay for paralegals to produce random nonsense.
Paralegal grunt work is often sifting through case law history to find nuggets of information that can be (ab)used to establish precedent.
And the four most expensive words in the history of humanity will forever be I told you so. All the human arguments in the world won't save you when you're reduced to nothing more than a target or a battery. We turned 1984 into an instruction manual. Greed is a disease, and we still like being infected by it no matter how much we repeat the worst of history.
Re: (Score:2)
Re: (Score:2)
Not that I disagree with a lot of that, but my point was that AI thingies are astonishingly bad at that exact sort of work. They have this unfortunate tendency to fabricate things like citations.
When AI is being used to simply search through existing case law (a LOT of it, in which one human typing and searching is not efficient enough), I'm not sure I understand the risk of AI literally manufacturing or fabricating evidence. Is that risk present today with paralegal work, outside of someone being corrupt?
I can't imagine a judge being very happy about being handed a brief that is essentially legal fiction.
Given that almost all judges previously worked for what we call a legal system today (as opposed to a justice system), I really don't know what to feel when it comes to making judges "happy". Th
Re: (Score:2)
Re: (Score:2)
I'd like to know what you think "AI" would add to search.
In a nutshell? Efficiency and cost. Why are millions using their voice to communicate on digital paper and transmit electronically in seconds instead of carrying around a notepad, pen, and a book of stamps? Same reason.
Instruct AI well enough, and you've got a room full of humans banging on keyboards. Only AI is gonna do that far faster and cheaper, and work 24/7. Remember AI is meant to take your input and learn how to do it even better than your brain can even imagine. And your brain needs that silly e
Re: (Score:2)
I think you're operating on a faulty set of assumptions. If you imagine using something like ChatGPT as a search (which I suspect is the case), training it on volumes of case law, then understand that there is a very real possibility that it will output text and citations that aren't real. There are lots of other approaches to search that are just as fast, if not faster, and far more reliable.
Instruct AI well enough,
'Instruct' can mean a lot of different things, depending on the kind of model you're using. There is no guarantee
Differently Employed (Score:2)
Greed funding AI doesn't seek to make humans temporarily unemployed. It seeks to make humans permanently unemployable.
The same thing has been said about all major technical revolutions: machines were going to make workers obsolete, computers were going to make secretaries obsolete etc. Instead, what happens is that new inventions massively multiply the work that a single human can do leading to new, but often very different jobs.
It's always easier to see how a particular technical innovation will disrupt existing jobs than it is to see the new jobs that it will create and often those new jobs don't become obvious until
Re: (Score:2)
Greed funding AI doesn't seek to make humans temporarily unemployed. It seeks to make humans permanently unemployable.
The same thing has been said about all major technical revolutions: machines were going to make workers obsolete, computers were going to make secretaries obsolete etc. Instead, what happens is that new inventions massively multiply the work that a single human can do leading to new, but often very different jobs. It's always easier to see how a particular technical innovation will disrupt existing jobs than it is to see the new jobs that it will create and often those new jobs don't become obvious until we start deploying the new technology and see where humans are needed. So while I agree that it is hard to see where the people displaced by the apparently approaching AI revolution will get employed I think it extremely unlikely that there will be no employment for them.
AI is looking to replace the human mind. That is what differentiates this revolution from any other. Argue about the speed of deployment all you want, but the end goal remains the same, and that product can and will improve at a rate inconceivable to humans at some point.
It's not hard to see what people will be "displaced" by this; everyone who uses their brain to secure employment. The only task humans will have left after that, is justifying their existence.
Computers used to be Human (Score:2)
AI is looking to replace the human mind. That is what differentiates this revolution from any other.
That's simply not true. A computer in the 19th and early 20th centuries was a job for mathematicians who were employed to perform complex calculations. They were completely replaced by machines that today we call computers. All the current AI technology does is extend the capabilities of computers to replace more "thinking" jobs. It will not replace all of them and the people whose jobs it does replace will transition to other employment, most likely the new types of jobs that AI will create.
Re: (Score:2)
Servants, it is much more fulfilling to boss around actual people then an AI.
Re: (Score:2)
There's usually a lag between the jobs going away and new ones appearing. IIRC, it was 3 generations for the original industrial revolution, though the colonies also gave the unemployed jobs.
Re: (Score:2)
Greed funding AI doesn't seek to make humans temporarily unemployed. It seeks to make humans permanently unemployable.
Don't worry. We're just getting near the top of the hype cycle. Reality will catch up and order will be restored.
Re:It's about damn time... (Score:4, Insightful)
We're creating intelligence this time.
We are not. Most definitely not. And even the CEO of the ChatGPT maker warns that people should not expect intelligence to be created. What we do is advance the state automation tech by a bit and now we have reached a stage where a natural language interface actually works. Well, sort-of. Actual artificial intelligence (i.e. AGI) is as far out of reach as always, i.e. it is still completely unclear whether it is even possible. And please do not give me any quasi-religious "physicalist" bullshit dogma.
That said, a lot of jobs can be made redundant with this non-intelligent automation tech now because many jobs do not require actual intelligence 95% of the time and the natural language interface can do a lot by itself. Think the Amazon warehouse model: 10 robots and one human minder and trouble-shooter (and 10% of an engineer or so) replace 10 humans. That is basically as bad for the job market as getting rid of those 10 humans altogether. This time there will essentially not be any new jobs to speak of and this time a lot of low/medium skill white-collar jobs will be gone.
Re:It's about damn time... (Score:4, Insightful)
We don't expect most people to figure shit out. We expect them to do a specific set of tasks over and over again. Sure, some people are figuring stuff out, but not anywhere close to the majority, not even a simple majority.
I always thought blue collar jobs would go first but turns out human interaction jobs will be what's left along side highly specialized jobs that only a fraction of society is capable of doing.
I figure either there will eventually be a mass die off of humans or we will destroy the civilization and go back to the dark ages because there is no way the people in charge are going to some how allow us all to have an acceptable life that doesn't require us to work. It's not just going to happen, so expect a bad scenario to come true.
Hopefully I'm wrong.
Re: (Score:2)
Same here. Just leaving this run with this many people that cannot compete with automation being out of a job will not turn out well at all.
Re: (Score:2)
We're creating intelligence this time.
We are not. Most definitely not...That said, a lot of jobs can be made redundant with this non-intelligent automation tech now because many jobs do not require actual intelligence 95% of the time and the natural language interface can do a lot by itself...This time there will essentially not be any new jobs to speak of and this time a lot of low/medium skill white-collar jobs will be gone.
Well, I'm glad you brought yourself right back to the entire point of mass unemployment. Now imagine your quaint little town where you and loved ones thrive today in relative peace, as non-intelligent automation starts to create a 20%+ unemployable rate because Greed simply doesn't care about instability it may cause. Then watch as Greed replaces humans with "good enough" AI and see how instability in a peace-loving society becomes considerably worse.
Dunno about you, but I'm not exactly looking forward to
Re: (Score:2)
I have never argued that automation will not have devastating effects from a certain threshold onwards. A(G)I is not needed to replace a lot of jobs. The tech for that has existed for decades, it was just way too expensive and difficult to configure it. That has now possibly changed.
Incidentally, this is the main reason I have argued for an UBI in tha past and still think it is a necessary (but not sufficient) component of any real solution. Given the customary resistance of the greedy, the stupid and those
Re: (Score:2)
Re: (Score:2)
No, your Governments don't have an answer for this, nor do they really give a shit.
That's probably true at this exact instance, but they absolutely will give a shit if/when they start seeing the tax revenue river start to dry up.
And if they let job loss increase too much, the economy will tank and it'll won't matter how much junk the AI can make if no one is buying.
The way I see it; Chat GPT et.al. is making the system vibrate like mad, but it isn't swinging very far. Like every other complex system a new natural frequency will develop and the system will eventually normalize until the ne
Re: (Score:2)
Much like creating the nuclear bomb, the goal of creating intelligence will not change until it happens. It's bad enough being ignorant of how to control it, but to assume it won't ever happen? Look at the last century of advancements. Those rolling Model Ts off the newfangled assembly line would have never imagined man would be driving a lunar rover within their lifetime. It was simply unfathomable, much like AI is right now to you.
And a lot of humans are not only entertained by magic, but are easily f
Desktop Office Suite 2.0 (Score:5, Interesting)
You know what else killed millions of white collar jobs? Desktop Office suite programs, like Microsoft Office.
Back in the 1950s-1980s, every office looked like the set from MadMen. Executives in their offices and small armies of grunts on desks pounding away on type writers and slide-rules doing all the nitty-gritty office work. Bosses would literally have tape deck voice recorders that they would speak out their reports, and grunt office labor would listen and type out the reports. There would be a whole floor of "computers", except back then "computer" was a job title for a grunt office worker to literally spend all day doing the arithmetic calculations and recording the results on spreadsheets. And spreadsheets where LITERAL massive sheets of lines and columns that office workers would write on.
And here is the thing. The desktop office suite so completely destroyed that whole career path that you probably didn't even know it existed at all. Today, the bosses do all their own spreadsheet work and write their own reports, and doing that is far far far more cost efficient for businesses and customers.
Re: (Score:3, Informative)
Today, the bosses do all their own spreadsheet work and write their own reports
I was with you until this heh. In my experience no they do not, they get a jr analyst to do it then take credit for the resulting reporting.
You're still right in that that's just a small number of analysts versus floor upon floor of people doing computations like in the past though, for sure.
Re: (Score:2)
GP's point still holds though; those people doing it now barked into the dictaphone before. Sure, some of the jr analysts replaced 10-20 individuals, but I have known plenty of senior executives that were pretty handy with Excel; they did it to validate the work of others generally, but could do enough themselves to get by.
Re: (Score:2)
Of course the exact "who-does-what" is going to vary significantly from org to org. Some bosses are much more hands on than others. But NO COMPANY ANYWHERE ON EARTH operates like most every company did in the 1960s.
Re: (Score:2)
Managers who write their own reports are about to be replaced by software that writes reports, and software that fires people. Just like most of the people who wrote reports got replaced by a few people who used reporting tools to do their job, only a lot faster.
Re: (Score:2)
Managers who write their own reports are about to be replaced by software that writes reports, and software that fires people. Just like most of the people who wrote reports got replaced by a few people who used reporting tools to do their job, only a lot faster.
This is probably correct. But what will likely happen is that companies fill out with MORE people whose job it is to tell the AIs what to write and review and sign off on the AI's work. Companies aren't going to get smaller or higher fewer people. It'll be the same number of office people producing 10x the amount of work.
Re:Desktop Office Suite 2.0 (Score:5, Insightful)
Great... another 10x increase in data with zero increase in information. We'll need AI to sort through the AI-generated content...
Re: (Score:2)
Great... another 10x increase in data with zero increase in information. We'll need AI to sort through the AI-generated content...
Heh. Your username says it all.
Re: (Score:2)
Even if the bosses aren't doing the spreadsheet work today themselves, 10-30 years ago they were the junior analysts who did the spreadsheet work.
Re: (Score:2)
I would also add programmer. If you look at the etymology of "programmer" you'll find some curious developments:
1890 programmer = person who wrote program notes, aka event planner
1910 programmer = woman who did computations for banks and telegraph
1948 programmer = person who programs computers who were mostly women [siliconrepublic.com]
1960 programmer = mostly men
2020 roughly 28% of programmers are women
Re: (Score:3)
It mostly just led to way more text and spreadsheets being made now. You honestly think there are fewer office workers now than in the 70s?
You are probably just missing his point. The point I believe he was making is that we have already seen examples of massive disruption to office workers, and it didn't lead to less office workers. Many times when you make someone more productive, you increase the need for those people.
White collar work is usually limited by what work is cost effective to do, not limited by a finite amount of potential work. This means most of the time work is made more efficient, businesses begin hiring people for work whic
VCs, not humans (Score:3, Insightful)
"ChatGPT, which can create content that is indistinguishable from human output"
VCs see it as content that is indistinguishable from human output, and this illustrates how much they value human output.
Re: VCs, not humans (Score:3)
Except that output is, like it or not, mediocre. It's a McDonald's cheeseburger.
Which I guess most people are cool with.
I wouldn't be.
Re: (Score:2)
Re: (Score:2)
And that is the point. ChatGPT can create low-quality content that is hard to distinguish from low-quality content created by humans. In many applications that does not cut it. Sure, if you produce, say a lot of really bad software, you can replace your wannabe coders with ChatGPT. But if you need quality? Impossible.
Re: (Score:2)
That's true today. What about in 5 years? 10 years? It's really only a matter of time before these AI systems can do acceptable work. The training data they use will get better and so will they. It's foolish to think development into more advanced AI is just going to stop. It won't.
Re: (Score:2)
To be fair, there's a fair number of people in the industry that I have never seen rise to the level of outperforming ChatGPT. They royally suck, but are still somehow employed.
Re: (Score:2)
To be fair, there's a fair number of people in the industry that I have never seen rise to the level of outperforming ChatGPT. They royally suck, but are still somehow employed.
And _that_ is exactly the threat from these machines: A lot of people are pretty bad at their jobs. As ChatGPT has a natural language interface, it can replace pretty bad (and to some degree mediocre) white-collar workers. Sure, the results will be crappy, but for some areas this seems to be acceptable or the employers would have hired better people already. A lot of mindless office drones will be getting the ax and there will not be any replacement jobs this time. And that will create a massive problem for
The scary thing: this report was created by AI (Score:4, Informative)
A reasonable guess (Score:3, Interesting)
1) google stuff
2) copy and paste
3) lightly edit to circumvent plagiarism and copyright law
4Publish and profit!
Will be drastically streamlined by the new systems. Instead of 5 people doing the job, 1 person will be checking the bots for “omg the AI went insane” moments.
Re: (Score:2)
Indeed. And for that one person you can now get somebody with an actual clue and still save a ton of money.
Life Imitating Art (Score:2)
We apologize for the fault in the forecast. Those who were responsible for sacking those responsible have been sacked.
Quit Panicking (Score:2)
This thing is nowhere close to replacing jobs, if anything it may increase jobs as morons invest in it. It's a few decades from being good enough where it doesn't need to be babied. The thing that would replace jobs is dexterous robots, and we are nowhere close on that either. We still don't have lights out factories for anything.
Re: (Score:2)
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
This thing is not going after factory workers. It is going after office drones. And it will make a lot of them redundant.
Re: (Score:2)
There are already 3 examples of Light Out Factories ... ...and many where the lights are only for the few maintenance workers to repair the robots - they never work on the factories product
Gotta trust them (Score:2)
The same Goldman Sachs who predicted the 2008 mortgage crisis.
We need a CEOChatGPT (Score:2)
According to this government site [bls.gov], in 2021 there were 3,402,300 top executives in the US, earning a median salary of $179,520. That's a combined compensation of $610 billion. Imagine the productivity boost to the US economy from replacing those top executives with CEOChatGPT! This should be a top priority for GPT researchers. I'm sure some top executive will surely see the immense benefit to society.
Most continue in their jobs (Score:2)
Most people would see less than half of their workload automated and would probably continue in their jobs, with some of their time freed up for more productive activities
Or their employers will fire half of them and have the remaining do the same job with the help of AI. Either for greed sake, or to remain competitive.
And the economical grow will not happen because there will be less consumers with a job and able to pay something
Re: (Score:2)
...and the people fired will start up new businesses and create work including jobs that didn't exist before ...
We have seen this supposed level of disruption before and white collar work especially expands to fill the void, as it's limited by how economical it is to do more work, not that there is not more work to do
Re: (Score:2)
We have seen this supposed level of disruption before
Sure, we swaped agricultural labor for industrial labor, then industrial labor for white collar labor.
But now we swap white collar labor for what?
ChatGPT is FAKE AI. (Score:2)
It is a parlour trick, namely a Chinese Room, that people had already thought about decades ago.
It is simply regurgitating phrases. It has NO understanding of the things it talks about.
ChatGPT will soon lose its novelty and the press will move on.
Real AI, computers that understand and can reason about our world, will be a seismic shift.
The numbers sound dubious (Score:2)
Generative AI will only affect white-collar office jobs, which is less than a third of the workforce.
It will not affect: teachers, nurses, soldiers, cops, baristas, shop assistants, flight attendants, truck drivers, Uber drivers, couriers, construction workers, hairdressers, cooks, and gardeners. i.e. most of the population.
- as foretold by H. G. Wells - (Score:2)
It is safe to say that the Morlocks will not lose their jobs. Someone has to fix our toilets, change our lightbulbs and, eventually, tie our shoelaces for us. The world will have no need for salespeople, executives, politicians, teachers, scientists, and paper pushers. We, the Eloi, will have nothing to do but enjoy life as the machines and the Morlocks take care of everything. This is what god intended for humanity.
The end is near (Score:2)
So far, the main use of LLMs seems to be people smoking weed and asking ChatGPT to write fart poetry. The end of human civilization is near.
Re: (Score:3)
Lawyers are perfectly safe, because in every State and most nations, lawyers have to be licensed to practice law. A natural born human person is the only entity that can have a license. Companies can't be licensed. Same with doctors and professional engineers. All licensed professionals are going to be fine, and probably become more efficient and cost effective.
Re: (Score:2)
All licensed professionals are going to be fine, and probably become more efficient and cost effective.
Will they become cheaper or just more lazy?
Re: (Score:2)
Probably cheaper. They will all stay fully busy as they always have been, but will just be able to do more work faster.
Re: (Score:1)
Re:Wow... (Score:4, Insightful)
In such a case, theoretically, a lawyer may end up taking on more caseload, if the Generative AI can take care of the tedium. This would mean as an industry, they would not be perfectly safe, as you need fewer of them. They may alternatively make due with fewer paralegals (lawyers offload a boatload of inane tedium to paralegals to chew on).
Also, while lawyers can't be AI, a person *can* take care of some of their own legal needs, and one could imagine offerings to help prepare some of the tedious materials to enable a person to take care of it themselves.
Similar arguments can apply to other fields where man hours are significantly consumed by navigating mindless (e.g. within the bounds of AI), but tedious things.
Re: (Score:2)
In such a case, theoretically, a lawyer may end up taking on more caseload,
Right. This is the way it's going to play out. The professionals are going to be able to do more work faster and more accurately. If you want a career that is AI proof, find a career that requires a license.
Re: (Score:2)
But watch the demand for those careers still plummet, as in aggregate you would need fewer of those licensed people for the same amount of work...
In some fields this may mean a whittling down of 'assistants' like paralegals or in some backwards software development shops, 'programmers' when they hold a distinction between 'architect' and mere 'programmer'.
The question is if this does represent a lot of the labor hours mankind expends today, is there some latent ambition that will be able to consume the exce
Re: (Score:2)
I ain't talking about software development. That's it's own universe. I would never recommend a person go into software development because that's EXACTLY the field that is going to be gutted worst by AI.
Re: (Score:2)
Re: (Score:2)
From what I've seen, it's a bit of a parlor trick of limited use. If you scribble or describe something that has pretty much been done verbatim already thousands of times in thousands of tutorials and git repositories, it will do a slick job of recognizing and synthesizing *close* to something that works. It is categorically distinct from experience of interacting with computers historically, but I'm not sure I've seen it been more useful than cloning a popular tutorial and then tweaking it in practical t
Re: (Score:2)
Oh get off it. If functions of law, or any other licensed profession, can be replaced effectively by AI, then laws will change to let it happen.
And even if it doesn't: If AI systems let licensed professionals become substantially more productive, then surely less licensed professionals will be necessary.
Re: (Score:2, Insightful)
Oh get off it. If functions of law, or any other licensed profession, can be replaced effectively by AI, then laws will change to let it happen.
That's unlikely to ever happen because you can't hold a software responsible for it's work. You can sue a lawyer if your lawyer really screws up, but you can't sue software. In the end, licenses were never about locking other people out of industries, it was about knowing who to hold responsible for errors and mistakes. If you can personally be sued for your insurance money, then your job is safe from AI.
And even if it doesn't: If AI systems let licensed professionals become substantially more productive, then surely less licensed professionals will be necessary.
This is also not likely. Whats MORE likely is that companies hire the same number of professionals but j
Re: (Score:1)
> licenses were never about locking other people out of industries
The naivety is strong with this one.
If >it was about knowing who to hold responsible for errors and mistakes.
were that the case, then all software developers would be required to be licensed similar to airline pilots. Certified for a particular development stack, on a particular OS...
Be it to create artificial scarcity or hold experts accountable, the fact that there is some sort of barrier between the licensed and not-licensed (the ha
Re: (Score:2)
Re: (Score:2)
Truth is, there aren't all that many paralegals now. Most of those jobs went away with the travel agents.
Re: (Score:2)
What are you, a hallucinating AI? Paralegalism is one of the fastest growing occupations in the country.
https://www.bls.gov/ooh/legal/... [bls.gov]
Re: Wow... (Score:2)
I guess so... None of the lawyers I work with have paralegals. But maybe they environmental field is weird?