ChatGPT is Already Taking Jobs (msn.com) 193
The Washington Post writes that "Some economists predict artificial intelligence technology like ChatGPT could replace hundreds of millions of jobs, in a cataclysmic reorganization of the workforce mirroring the industrial revolution.
"For some workers, this impact is already here." Those that write marketing and social media content are in the first wave of people being replaced with tools like chatbots, which are seemingly able to produce plausible alternatives to their work.
Experts say that even advanced AI doesn't match the writing skills of a human: It lacks personal voice and style, and it often churns out wrong, nonsensical or biased answers. But for many companies, the cost-cutting is worth a drop in quality. "We're really in a crisis point," said Sarah T. Roberts, an associate professor at University of California in Los Angeles specializing in digital labor. "[AI] is coming for the jobs that were supposed to be automation-proof..."
The technology's ability to churn out human-sounding prose puts highly paid knowledge workers in the crosshairs for replacement, experts said. "In every previous automation threat, the automation was about automating the hard, dirty, repetitive jobs," said Ethan Mollick, an associate professor at the University of Pennsylvania's Wharton School of Business. "This time, the automation threat is aimed squarely at the highest-earning, most creative jobs that ... require the most educational background." In March, Goldman Sachs predicted that 18 percent of work worldwide could be automated by AI, with white-collar workers such as lawyers at more risk than those in trades such as construction or maintenance. "Occupations for which a significant share of workers' time is spent outdoors or performing physical labor cannot be automated by AI," the report said...
Mollick said it's too early to gauge how disruptive AI will be to the workforce. He noted that jobs such as copywriting, document translation and transcription, and paralegal work are particularly at risk, since they have tasks that are easily done by chatbots. High-level legal analysis, creative writing or art may not be as easily replaceable, he said, because humans still outperform AI in those areas.
The article notes that one copywriter lost all 10 of his clients over the last four months — and though one later hired him back, he's now training to be a plumber.
"For some workers, this impact is already here." Those that write marketing and social media content are in the first wave of people being replaced with tools like chatbots, which are seemingly able to produce plausible alternatives to their work.
Experts say that even advanced AI doesn't match the writing skills of a human: It lacks personal voice and style, and it often churns out wrong, nonsensical or biased answers. But for many companies, the cost-cutting is worth a drop in quality. "We're really in a crisis point," said Sarah T. Roberts, an associate professor at University of California in Los Angeles specializing in digital labor. "[AI] is coming for the jobs that were supposed to be automation-proof..."
The technology's ability to churn out human-sounding prose puts highly paid knowledge workers in the crosshairs for replacement, experts said. "In every previous automation threat, the automation was about automating the hard, dirty, repetitive jobs," said Ethan Mollick, an associate professor at the University of Pennsylvania's Wharton School of Business. "This time, the automation threat is aimed squarely at the highest-earning, most creative jobs that ... require the most educational background." In March, Goldman Sachs predicted that 18 percent of work worldwide could be automated by AI, with white-collar workers such as lawyers at more risk than those in trades such as construction or maintenance. "Occupations for which a significant share of workers' time is spent outdoors or performing physical labor cannot be automated by AI," the report said...
Mollick said it's too early to gauge how disruptive AI will be to the workforce. He noted that jobs such as copywriting, document translation and transcription, and paralegal work are particularly at risk, since they have tasks that are easily done by chatbots. High-level legal analysis, creative writing or art may not be as easily replaceable, he said, because humans still outperform AI in those areas.
The article notes that one copywriter lost all 10 of his clients over the last four months — and though one later hired him back, he's now training to be a plumber.
Dey Turk Er Jurbs (Score:4, Funny)
Re: (Score:2)
DERT DEE DOOOOO000OOO0!
Next (Score:2, Funny)
ChatGPT going customer and buying stuff instead of humans that have no money after chatGPT took their jobs
Re: (Score:3, Funny)
ChatGPT going customer and buying stuff instead of humans that have no money after chatGPT took their jobs
South Park: They took our jobs!!! [youtube.com] ...
Re: Next (Score:2)
Re: Next (Score:4, Insightful)
Jobs that never required a lot of skill to begin with are easy to automate. Likewise, it doesn't sound like anything significant is being replaced here. They specifically mention "marketing and social media content". In other words, ChatGPT is taking over the work needed to create content mills. Content mills are basically just low effort content intended to generate ad revenue based on sheer quantity of content, which never really had any concern for quality to begin with. It always was shit meant to sound plausible but at the end of the day doesn't need to have any substance. Sites like ehow or wikihow never paid the people who wrote this stuff much of anything to begin with, assuming they paid anything at all, with their endgame being only to generate high search engine ranking and clicks. Google regularly adjusts their ranking algorithm to sink these sites for exactly that reason.
Another site that frequently ranks high and could easily be replaced by ChatGPT is quora, and everybody who does any kind of work for them.
Re: (Score:2)
In other words, ChatGPT is taking over the work needed to create content mills.
How ironic. My understanding is the LLMs are trained on content scraped from the Interwebs and now it's creating all the content which will be used to train the next generation of LLMs.
Personally, what I want is for a LLM to translate the instructions for all the $20 gadgets I buy. Clearly no one spent a lot of money on a quality translator.
Re: (Score:2)
But as stated in the article: "cost-cutting is worth a drop in quality", so no useful product description for you, even if its automated translation would cost only 0.001$.
Re: (Score:2)
Re: Next (Score:3)
Re: (Score:2)
I could really go for a Starbucks.
Re: (Score:2)
Human sex workers won't be able to compete... all of the upsides of human personality with none of the downsides, by design. (Unless you wanted to design otherwise...)
Re: (Score:2)
I for one can't wait for the internet to be filled with wishy-washy, over-verbose, ChatGPT-generated text that sounded "plausible" to people who failed to make it as writers or technicians.
Even more fun will be when they use this generation's Chat-GPT-generated Internet to train the next generation of AI.
It's a death-spiral to Idiocracy but it'll be cheap to do.
Re: Next (Score:3)
Re: (Score:2)
Except for the fact that running deep learning at that scale takes kerjillions of watt hours of energy and tens of thousands of GPUs and petabytes of hard disk...
If you already have a major search engine and a big cloud computing service then you have everything you need. You can do the training with the idle CPU cycles.
It's only a more barrier to entry for the startups.
Re: Next (Score:2)
You can tell it to write in a specific voice, and/or tell it to be concise. You can do Ernest Hemingway no problem, either terse prose version, or drinking at the bar playing darts version
Re: (Score:2)
Yeah, but will they?
Re: (Score:2)
It already is starting to. I've very often googled some home improvement how to or look for reviews on a product I land on very obviously computer generated articles that are just pages of barely intelligible nonsense. Same for video game walkthroughs or hints. All stuff that is pretty easy for a LLM to generate content over.
Re: (Score:2)
I understand your perspective. Although there is a possibility on the horizon, where AI might soon achieve a parallel depth, authenticity, and creative prowess as human writing. The advancements we witness in AI, such as ChatGPT, demonstrate an increased ability to grasp context, evoke emotions, and deliver unique insights. Although challenges persist, we find ourselves standing at the cusp of a future where AI may stand shoulder to shoulder with human writers. However, let us always cherish the unparallele
sound right vs being right (Score:5, Insightful)
There is a difference between "sounding right" vs "being right."
Perhaps this will get better with time but from what I have experienced so far, at least on the standalone models, when you ask for specific details, i.e. how many chapters are in "To Kill a Mockingbird", and often times it gets it wrong. Very wrong. What is worse when you ask it for certain specific details, it will often times make up stuff as opposed to letting you know it doesn't know the answer. Which is very scary if you're trying to use it for a research paper, or like in recent news, when it made up 6 fake cases as part of court filings.
Literally, everything that comes out of these models, unless you accept that it's "make believe", you need to double check the work.
Re:sound right vs being right (Score:4, Interesting)
I've made "excel and accountants" comparison in the past, and this is another case for it.
When we moved to excel for accounting, you had to doublecheck the formulas everywhere in first few years. Then it more or less standardised, common errors were found and fixed and you needed to do a lot less double checking.
Nowdays, there are entire production lines that are controlled through a single excel file. And it just works. LLMs will be the same. First few years of pain, and as we nuke the failures and warts it'll stabilize at something that is functional, and where we know exactly what needs to be checked and what can be trusted.
But even in its current form, checking important points in a relevant document is way faster than generating relevant document by hand.
Re: (Score:2)
When we moved to excel for accounting,
Why would you move to Excel for accounting? What accounting do you do in Excel? It's like using a minivan to move refrigerators. It works well for a wine cooler but it's not the tool for the job. It's Ok as a spreadsheet . It has issues with precision on large numbers unless they fixed that. I stopped relying on it when worksheets didn't add up properly.
Re: (Score:2)
I'm sorry you have a very specific problem that pretty much no one else has that microsoft won't fix for you.
Everyone else in accounting world uses excel.
Re: (Score:2)
Maybe you should get around to upgrading from that old Pentium [wikipedia.org] you've been clinging to.
Re: (Score:2)
These "production lines" you refer to are the stuff of nightmares. No one should ever use excel for that sort of task.
Re: (Score:3)
In the case of Excel, humans are still the ones doing the learning. The computer just follows instructions, and those instructions don't just magically mutate on their own.
But even in its current form, checking important points in a relevant document is way faster than generating relevant document by hand.
That's just another way of saying, "The document is full of fluff and filler that doesn't need to be there, but it's harmless so we'll just ignore it. More is less!"
Re:sound right vs being right (Score:5, Funny)
'how many chapters are in "To Kill a Mockingbird"'
An African Mockingbird or a European Mockingbird ?
Technical expertise (Score:3)
There is a difference between "sounding right" vs "being right."
Perhaps this will get better with time but from what I have experienced so far, at least on the standalone models, when you ask for specific details, i.e. how many chapters are in "To Kill a Mockingbird", and often times it gets it wrong. Very wrong. What is worse when you ask it for certain specific details, it will often times make up stuff as opposed to letting you know it doesn't know the answer. Which is very scary if you're trying to use it for a research paper, or like in recent news, when it made up 6 fake cases as part of court filings.
Literally, everything that comes out of these models, unless you accept that it's "make believe", you need to double check the work.
ChatGPT gives non-professionals the technical expertise to become professionals.
We can make an analogy with art photography: To become a professional photographer (ie - have your photographs sell or hang in a gallery) you need two things: artistic merit and technical skill.
Setting artistic merit aside for a moment, consider the technical skill needed over history: in the beginning cameras were big boxes: you needed to know about lighting, F-stop, focus, and exposure (how long to take the lens cap off). You
Re: (Score:2, Funny)
People keep saying that AI will take peoples' jobs, and we can now say which jobs those are: creative writers and graphics artists.
And programmers. Don't forget programmers. One doesn't need technical expertise to write the code, just the creativity to describe what they want. As those descriptions become more prevelant, AI will become better at writing code and thus, programmers will not be needed.
Think of how much money companies such as Microsoft, Google et al can save by not having armies of programmer
Re: (Score:3)
Re: (Score:3)
To use ChatGPT effectively you still need artistic merit, but all the technical expertise is done for you.
If anything, it's quite the other way around: AI enables people who can't paint for shit to generate art they were only imagining before. It enables people who can't put sentences together in a cohesive style (e.g. those with mild dyslexia) to finally express themselves well, without having to hire people to do that for them.
Technical expertise is definitely required. Have you seen image generation prompts?
https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fgithub.com%2FAUTOMATIC11... [github.com]
Sure, everyone can throw a few words at ChatGPT or some basic prompts
Re:Technical expertise (Score:4, Informative)
ChatGPT gives non-professionals the technical expertise to become professionals.
Hahaha, no. What it does is enable them to fake it in non-demanding applications where bullshit is not obvious. You know, the thing a somewhat skillful amateur could do before as well.
Re: (Score:2)
I can't help but feel like the approach towards automating away the technical expertise will only worsen collective creative outputs. To take the analogy of camera phones, while cellphone cameras have definitely opened up photography for more people, it has also largely led to a lot more garbage and just plain mediocrity pushed around on the internet. As it turns out, photography is more then just having a camera and taking "pretty" pictures. And yet outside of specific hobbyist sites you get either boring
Re: sound right vs being right (Score:2)
The current models don't have a feedback loop because they're generalized with no training feedback loop. If you tell it that it's wrong, it's temporarily trained, but as soon as you close the session, poof that goes away. Using a tuned version can include that feedback loop and reduce most errors to a similar level of a human
Re: (Score:2)
Now that is wishful thinking if I ever heard it.
Re: (Score:2)
Judging by what sounds right too a large degree is how the world gets by. The world's not ready for a pop quiz where it's actaully got to distinguish between genuine competency and gibberish that sounds more or less right.
Re: (Score:2)
There most definitely is. But there are also a lot of people that do not know that and think sounding right is good enough.
It's a trait it learned from humans (Score:2)
unless you accept that it's "make believe", you need to double check the work.
How is that any different from the average human you meet? People make shit up all the time using their beliefs and/or their assumptions. Sometimes they just guess and exclaim it as fact. And just like with the AI, you have no idea.
Just part of the problem (Score:3, Interesting)
Re: (Score:3)
Meanwhile in real world, biggest problems are underpopulation (too many retirees, not enough young workers and children), recession (after burst of post-pandemix growth that was way too fast) and slightly above average inflation (just look at historic over last few decades in most Western nations instead of "but last decade").
Most people also forget that last couple of decades were exceptional in many ways. We had a demographics dividend with a lot of adults (lots of generated wealth) and very few kids (few
Re: (Score:2)
The last couple of decades have sucked compared with the 1990's. The economy hasn't recovered from 2008, and certainly hasn't recovered from 2020 and the ungodly inflation that resulted.
Consumer choice has consistently shrunk, as smaller companies either fold, or get absorbed by larger companies. In 2000, I could get even a relatively low-priced car with multiple interior color options-- today, it's would you like your black interior to be cloth/fake leather or leather (which, if you live in Florida, blac
Re: (Score:2)
Depends on where you live really. Living in Finland (which I assume you do so) compared to living in the US are two very different experiences. Both are effectively developed countries, but education, wages, job opportunities, public infrastructure, and quality of life (e.g. healthcare) vastly differ, and not necessarily in a good way.
Sure goods are cheaper in the US (though the quality hasn't necessarily improved), but cost of decent housing, good education, and health insurance (hello premiums) are still
Re: (Score:2)
It's worse than that-- the AI bots are essentially throwing wikipedia, a dictionary, and stackexchange into a blender, and regurgitating what others have said. It's a neat trick, but it's lacking creativity.
I asked chatGPT to come up with an original science fiction plot-- While I admit that the plot itself was unusual, it was based on at least two well-known tropes that were totally disconnected and had no right to be in the same storyline. No amount of creative hacking and slashing could reconcile the p
Re: (Score:2)
resulting in idiocracy becoming real.
Have you been in a coma for the past six years? The idiocracy has skyrocketed of late.
Re: (Score:2)
resulting in idiocracy becoming real.
Have you been in a coma for the past six years? The idiocracy has skyrocketed of late.
"Ow my balls" is the new reality and people are seriously considering voting Trump back into office.
(nb. That statement doesn't mean I like Biden, I think they're both bad choices... but bringing Trump back for another round is 100% Idiocracy)
need to unlink jobs and healthcare! (Score:2)
need to unlink jobs and healthcare!
Re: (Score:2)
New tools mean that you need to learn how to use the new tools effectively.
When you can't predict how these new tools are going to evolve even over the next 6 months, it's kind of hard to formulate a multi-year education plan.
And that's just for one person. Half of the working population could be in this bucket.
How are you going to re-educate all of these people in the next few years anyway? Student loans seem to be a severe problem already.
Re: (Score:2)
>When you can't predict how these new tools are going to evolve even over the next 6 months, it's kind of hard to formulate a multi-year education plan.
Which is one of the main reasons why planned economies fail as time moves on. It's impossible to predict that far into the future, so planning must be agile and local. Something that market economies do well and planned economies do poorly.
>And that's just for one person. Half of the working population could be in this bucket.
When we invented farming,
Re: (Score:2)
You completely ignore the fact that this particular transition is taking place in a matter of MONTHS, not decades or centuries.
Hopefully, your job is among the first to get eliminated, so you can have fun trying to predict which college major won't be worthless by the time you finish it.
Re: (Score:2)
>You completely ignore the fact that this particular transition is taking place in a matter of MONTHS, not decades or centuries.
Except that of course, I'm not, and no, this transition is not taking place in a matter of months. This is going to take several years at least. What you're seeing is bleeding edge adoption. The fact that it's as promising as it is shows just how revolutionary this technology may be once we actually begin to widely adopt it.
>Hopefully, your job is among the first to get elimi
Re: (Score:2)
It's not a personal insult. I'm just expressing hope that the karma in this world gets doled out fairly and appropriately.
Basically, if you think that all of this is going to be so "revolutionarily" great, you should go first.
Re: (Score:2)
It may in fact require less education to use new tools compared to old.
How will an uneducated person know if the output is any good?
Surely an educated person will be able to write better prompts and refine the output more.
Re: (Score:2)
>How will an uneducated person know if the output is any good?
"Yo, 'dat retarded little prick doesn't know if it works or not because he's didn't do doctorate like I did".
Except that's it's usually the opposite. It's living in the brain only, doctorate and higher people that often are utterly confused by practical things, like "does it actually work".
>Surely an educated person will be able to write better prompts and refine the output more.
Not really. More experienced people will. And after education
masters + trade school needed to get into an basic (Score:2)
masters + trade school needed to get into an basic job and an pay level that will not cover the loan on that.
Re: (Score:2)
So what happens when there isn't enough people who bring that?
Alternatively, what happens when people with those papers are widely found to be unable to perform?
Take on new responsibilities? (Score:2)
Re: (Score:2)
Maybe volunteer to take on new tasks and responsibilities that chatgpt can't perform on its own?
Like getting ChatGPT its coffee?
Re: (Score:2)
Like getting ChatGPT its coffee?
More like "install more hardware for ChatGPT, then return to the treadmill to power it".
That doesn't really work (Score:2, Insightful)
I saw this happen constantly in 2008. People had over time organically moved into new roles and taken on new responsibilities. The bean counters didn't care. Their job was just to cut headcount by x
Your timing was on point, that's all (Score:2)
You might also get away with it at a mid sized company, but probably not. Companies don't like to promote from within anymore. You're there to be used up and discarded. If you get too much knowledge you'll start demanding better pay and worse, you might be so indispensable they have to give it or take a hit.
That's the major shift in employment since the boomers were in their prime. MBAs noticed that they had
Re: (Score:2)
The problem is, this is outsourcing all over again-- then it was call centers with barely comprehensible Indians with names like "Bob", now it's chatGPT-- management thinks it's going to be cheaper, and it is-- but the quality is so much worse.
Bottom line, if you're a company, you need to ask "Is this better for our customers?", whereas most management asks "Is this cheaper for our company?".
The questions are very close to antithetical. ChatGPT may be cheaper for the company, but it will harm their long te
I think this is what a lot of us feared (Score:5, Insightful)
Most of us here are quite aware of ChatGPT's issues and limitations. But the world is largely run by middle managers - and they're demonstrating once again that a huge percentage of them are incompetent morons.
Forget the middle managers think about the CEOs (Score:2)
All this talk of AI has changed their opinion on that. They're now looking for anything and everything they can automate. There's tons of stuff that was still being done by hand but could have been automated 10 or 20 years ago and they're a
Re: (Score:3)
Think about the low quality of outsourcing (Score:2, Insightful)
If they didn't care when the quality went to shit by offshoring do you think they're going to care when the quality goes to shit from using LLMs? Good enough is always good enough.
And besides what are you going to do go to on
Re: (Score:2)
Re: (Score:3)
Correct, the threat from the current crop of "A.I." is the collapse of the Bullshit Economy, but it goes beyond clickbait and worthless "news" articles. It could also affect many of the bullshit jobs inside of businesses. Of concern though, is that these jobs are part of a hidden welfare system, in the U.S. at least, and those costs may be shifted from companies to the remaining taxpayers. And of course consumer spending could be significantly affected. But it should put a stop to that pesky inflation.
Re: (Score:2)
Most of us here are quite aware of ChatGPT's issues and limitations. But the world is largely run by middle managers
It is certainly not middle managers who decide that "cost-cutting is worth a drop in quality". That decision is traditionally made by the C-level and investors - and was even before the advent of LLMs. The race to the bottom of quality started to become a real frenzy a few decades ago. Before that, some who ran companies tried to save costs, others tried to impress customers with quality - the latter kind is now almost extinct. Even "luxury" goods, these days, do not even try to impress by quality, but only
Drop in quality (Score:5, Insightful)
Re: (Score:2)
"The cost-cutting is worth a drop in quality" is a very 2023-thing to write.
Yes, I can agree to this.
Late-stage capitalism isn't about creating qualitative products anymore but to scam the most people out of the most money by having algorithms churn out redundant and bedazzle customers into looking at advertisements, making them want products they don't need based on fake reviews.
I do not think that the word "capitalism" in this sentence is useful. If you look at products from non-capitalist countries, they certainly do not impress with better quality.
The problem is that after a long period of de-regulation and non-accountability, companies have become creepy, greedy entities devoid of any moral. And MBAs are essentially teached this is how it has to be. Just with a few sprinkles of social-justice virtue signaling on top, which does not mean anything in pra
Re: (Score:2)
People still do not seem to understand that capitalism can only work under regulations, not the lack thereof. Like most market-based systems, it operates under a system of contracts in which all involved parties trust one another to uphold their end of the bargain and exchange goods/services/currency in manner that is fair to all involved. Naturally, operating entirely on good faith is not tenable, so ergo regulations are needed to ensure that people can trust one another to do business with each other. It'
Hell of a sleigh ride... (Score:3)
"The cost-cutting is worth a drop in quality" is a very 2023-thing to write. Late-stage capitalism "
This is more along the lines of
"The world is going to hell in a handbasket. So we are going to get the last few ducats before the shit hits the fan everywhere. Let us eat, drink, and be merry for tomorrow we all die."
Also the reason why kids are fucking around with Fentanyl, knowing by now the high risk of death that comes with it.
People are seeing the manure drawing real close to the whirling blades
Replacing MBAs should be really effective (Score:4, Informative)
I predict an almighty series of court cases... (Score:2)
Where did all the data come from to "train" these massive data sets?
Much of it will have come from the very people who may lose their jobs. ... well, me and you - all of us really.
A lot of it will have come from
The sky is the limit.
Our data has been slurped up for decades.
People have hosted trillions of bytes of data - images, emails, CVs, literature - you name it.
All of that data has been used to train these systems - who does it belong to?
I guess this is where it gets really tricky and where I really thin
Re: (Score:2)
There are cases in various courts to try to decide these questions. It is still unclear which way they are going to cut. There was a discussion of that by Legal Eagle a few month back.
https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3F... [youtube.com]
The Singularity (Score:2)
It will occur when all the training data is generated by bots. Then there will be whirlpool of stupid able to swallow entire economies, the MBAs will rejoice, those that are left.
Those that write ... (Score:5, Insightful)
"Those that write marketing and social media content..." are easily replaced by a chatbot.
Is anyone surprised? Anyone?
If you contribute nothing, you are easily replaced. Such is life.
Re: (Score:2)
Re: (Score:3)
Gonna be a weird circumstance when GPT is defining cultural norm articles while also being responded to by GPT for fake engagement.
Does it collapse in on itself? Do networks become flooded with even more pablum? And how will the trainers respond?
The smoking crater from this will be something to behold.
Now there's the moral of the story! (Score:5, Insightful)
The article notes that one copywriter lost all 10 of his clients over the last four months...he's now training to be a plumber.
That's what I'm telling students in school. Become a plumber, electrician, or nurse. Because if there's anything I trust AI will never be able to accomplish, it's unclogging your toilet, wiring your house, and changing your bedpan.
We've been here before with Offshore Outsourcing (Score:4, Interesting)
It lacks personal voice and style, and it often churns out wrong, nonsensical or biased answers.
OK, so your ChatGPT marketing bot tells the user they can get an iPhone for $5. What happens?...does the retailer get to say.."Sorry, our official website was wrong...AI...it's in our Terms of Service!...we're not liable for anything on our site."...or...do they absorb the cost? Accuracy matters.
This mirrors the offshore outsourcing craze of the early 2000s. When I was beginning my career, I was worried sick my job would be shipped overseas because Slashdot was running articles weekly about armies PhDs from India who are eager to take my job for 1/5 of my pay. I worked at several jobs who were midway through the process. In all cases, they stopped because anyone in India who can do my job doesn't want to do it for 1/5 of the cost, leaving only people who weren't qualified.
NOTHING is more expensive than a cheap programmer.
No one talks about offshore outsourcing anymore because it's not cheaper. All the talent gets relocated to a wealthier country or finds a better job. No one wants to work for some shitty bank or insurance company for low wages and constant disrespect when every multinational tech company has offices in your city and is willing to pay more and treat you right, if not pay you handsomely to relocate on an H1B...so all that's left are con-artists and people who cannot find a real job. Offshore outsourcing exists today, but barely and only for the shittiest jobs.
I predict ChatGPT will be the same. Those writing jobs?...eh, maybe for Ali Express clones, who weren't paying writers much anyway...anyone running in countries with liability laws that are actually enforced can't risk an error going through.
If you've met many technical writers, they're not hired because they're good writers...they're hired because they can write copy that won't get you sued...and maybe even be effective content. Their priorities, in order
ChatGPT is the dumbest idea imaginable for this scenario. Quality matters. ChatGPT will never be used by responsible companies...only scams.
Re: (Score:2)
NOTHING is more expensive than a cheap programmer.
Indeed. But it may take a bit for thet to become obvious. MBA morons planning quarter to quarter are often just to short-sighted to see it. That is how you get multi-year failed software projects.
Re: (Score:3)
No one talks about offshore outsourcing anymore because it's not cheaper. All the talent gets relocated to a wealthier country or finds a better job.
While I agree with your second sentence cited above, you can still find plenty of companies who are late entries into the "offshoring hype curves". I personally witnessed managers being asked to replace competent and productive local developers with cheap ones overseas, regardless of their competence - in recent years. Those "decision makers" who lack the ability to understand the difference between competent and incompetent developers are still plenty in numbers. Heck they even wanted developers from India
Huh (Score:2)
Increased automation has always been beneficial (Score:2)
Re: (Score:2)
Don't despair (Score:3)
Even earlier AI was doing that. (Score:3)
I noticed this in the finance press in particular. Years before the Covid crash, online finance journalism took on a boiler-plate quality. When the crash hit, the jig was up. Bots continued to generate stories like, "Why is IBM down today? Analysts offer their opinions". LOL. IBM was down because the ENTIRE MARKET WAS DUMPING. Any real human finance journalist would not have written such copy. If they were writing about individual companies at that point, they would write about which ones were best/worst positioned to weather the storm. 2020 AI wasn't smart enough to do finance journalism during a Black Swan. I'm not sure about the current generation.
Is it better at APPLYING for jobs? (Score:2)
As with solving CAPTCHAs, LLMs may by now do better than humans at negotiating those goofy puzzle-filled tech job interviews.
Physical labor can't be automated? (Score:3)
"Occupations for which a significant share of workers' time is spent outdoors or performing physical labor cannot be automated by AI," the report said
Tell that to the farm laborers whose jobs have been automated by AI driven machinery [clickworker.com].
One of the first tests... (Score:5, Insightful)
One of the first tests I ran with ChatGPT was to have it write a marketing text. You need sexy wording, but facyyal accuracy is irrelevant. While the result wasn't perfect, it was very, very good. Light editing, and in 5 minutes I had something that a marketing agency would have charged 4-figures for.
Social media content is much the same. ChatGPT's weakness - factual accuracy - is unimportant. Stringing plausible words together is all that matters, and that is exactly what it dies well.
you thought manuals were bad before... (Score:3)
I can't wait to look up how to do something in a manual written by AI, and it explains the wrong way to do something.
then I contact the company on the internet and the chatbot tells me I'm wrong.
Re: (Score:2)
I think ChatGPT would write greatly improved product setup or installation instructions, compared to a lot of what you get with cheap offshore products.
It's like if it hasn't been obvious for the last t (Score:2)
Companies vs executives (Score:3)
But for many companies, the cost-cutting is worth a drop in quality.
The sensitivity of different jobs to quality varies. However, an even more important observation is that executives make the job-cutting decisions and those executives sometimes don't have the best interests of the company in mind. Executives are incentivized to produce short-term stock gains. That's why laying off lots of employees is great for executives and their stock bonuses, even if it eventually hurts the company.
Perhaps even more importantly, decreasing the overall paid workforce negatively impacts the economy, especially in a consumer-driven economy like what the US has. Each executive hopes that only he will decrease his payroll and that other executives will keep on paying their employees so that those other employees will continue to buy his company's products. Of course, this is a just a fantasy. So, eventually the national workforce shrinks, consumers have less money, and the economy tanks. However, the executives still win, as long as they sell their inflated stocks in time.
Pay no attention to the man behind the curtain (Score:2)
And your sensationalist headline is helping the humans get away with it. Please stop.
The real problem (Score:3)
And no one could tell (Score:2)
I skim past all the marketing stuff on all websites as none of them relay the information anyone actually needs to know.
buzz-word bingo (Score:3)
An LLM machine is a parrot, repeating what someone else said: When no-one is paid to write fact-driven conclusions, ChatGPT and others will have nothing useful to say. As society changes, these buzz-word-bingo machines will choose random answers because there aren't any new facts for them to 'remember' and repeat.
Quality in the age of dupes. (Score:3)
From the story:
“We have to ask: Is a facsimile good enough? Is imitation good enough? Is that all we care about?” she said. “We’re going to lower the measure of quality, and to what end? So the company owners and shareholders can take a bigger piece of the pie?”
The answer is yes. [thecut.com]