The tidal wave of artificial intelligence (AI) barrelling toward many professions has generated deep anxiety among workers fearful that their jobs will be swept away -- and the mental health impact is rising.
The launch in November 2022 of ChatGPT, the generative AI platform capable of handling complex tasks on command, marked a tech landmark as AI started to transform the workplace.
"Anything new and unknown is anxiety-producing," Clare Gustavsson, a New York therapist whose patients have shared concerns about AI, told AFP. "The technology is growing so fast, it is hard to gain sure footing."
Legal assistants, programmers, accountants and financial advisors are among those professions feeling threatened by generative AI that can quickly create human-like prose, computer code, articles or expert insight.
Goldman Sachs analysts see generative AI impacting, if not eliminating, some 300 million jobs, according to a study published in March.
"I anticipate that my job will become obsolete within the next 10 years," Eric, a bank teller, told AFP, declining to give his second name. "I plan to change careers. The bank I work for is expanding AI research."
New York therapist Meris Powell told AFP of an entertainment professional worried about AI being used in film and television production -- a threat to actors and screenwriters that is a flashpoint in strikes currently gripping Hollywood.
"It's mainly people who are in creative fields who are at the forefront of that concern," Gustavsson said.
AI is bringing with it a level of apprehension matched by climate change and the Covid-19 pandemic, she contended.
But she said that she tries to get patients to "embrace the unknown" and find ways to use new technology to their advantage.
For one graphic animator in New York, the career-threatening shock came from seeing images generated by AI-infused software such as Midjourney and Stable Diffusion that rivaled the quality of those created by humans.
"People started to realize that some of the skills they had developed and specialized in could possibly be replaced by AI," she told AFP, adding she had honed her coding skills, but now feels even that has scant promise in an AI world.
"I'll probably lean into more of a management-level role," she said. "It's just hard because there are a lot less of those positions.
"Before I would just pursue things that interested me and skills that I enjoy. Now I feel more inclined to think about what's actually going to be useful and marketable in the future."
Peter Vukovic, who has been chief technology officer at several startups, expects just one percent or less of the population to benefit from AI.
"For the rest, it's a gray area," Vukovic, who lives in Bosnia, said. "There is a lot of reason for 99 percent of people to be concerned."
AI is focused on efficiency and making money, but it could be channeled to serve other purposes, Vukovic said.
"What's the best way for us to use this?" he asked. "Is it really just to automate a bunch of jobs?"
© 2023 AFP
10 Comments
Login to comment
Mat
Yet more FUD and hyperbole. Artists complaining about AI image generators may as well be portrait painters in the time when cameras were invented. They Took Our Jobs! Cameras are evil! Nah, it's just progress, move with the times.
Do we want the world to have banned the camera, so painters can keep their jobs?
Prevent the development of the car, so horses can keep their jobs?
Ban the electric lightbulb because it reduced sales of candles?
Outlaw medical research so witchdoctors and scammers can remain employed?
That's just not how the world works. In a short time, "is AI taking our Jobs" style articles with be viewed with the ridicule of the flag-man of history, who had to walk in front of a car so people knew the car was there.
dagon
Goldman Sachs and the centraliztion of hyper-financialized capital is at the heart of the problem, not just LLM.
And this is patently different from other waves of automation and comparisons to Luddism, steam engines and factory automation.
When the majority of intellectual, "white collar" work is easily automated, as is happening now and would be happening faster without bureaucratic and capital inertia, there is no other place to go.
Human "labor' and ever decreasing compensation is at a dead end but there may still be creative outlets.
But the obstacle is late stage capitalism.
https://moores.samaltman.com/
https://www.theguardian.com/sustainable-business/2015/mar/18/fully-automated-luxury-communism-robots-employment
dagon
Quite old analogies and a clear and present danger.
AI can tweak code better than most humans so....? Physical capacities and intellectual capabilities is the clear difference.
https://www.theguardian.com/commentisfree/2023/apr/01/chatgpt-write-code-computer-programmer-software
Mat
Write code better than most humans? No, not true.
The humans has understanding of what the code does, the LLM does not have this understanding at all. That makes a large difference.
The LLM can indeed create code quickly, and really quite good, but as the requirements get complicated, it starts to make dumb mistakes really easily (so do humans) and often completely fails to step toward a solution and instead will bounce between the current wrong answer and the previous wrong answer. Although, to be fair, I've seen similar among human coders!
LLM can code, yes, but it cannot code better than a (skilled) human. It can absolutely be used to make that smart human faster, but (competent) coders are not worrying about their jobs.
dagon
True at this level but the difference is human coders are not exponentially iterating their coding skills like LLM.
https://blog.tryamigo.com/gpt-4-coding/
The old paradigms are gone
theFu
I've been replacing departments of people with little Unix scripts for decades. This isn't anything new. I try to automate tedious aspects of my job away too. If it take more than 5 minutes a month, I want to automate it.
Computers writing code isn't anything new. We've called that "meta-programming" for decades. Some of my very first code, created code to make life easier. Earlier today, I created a 5 line script to create a list of files to be sent to another program for processing. Those inputs would have taken me hours to find manually. The script took about 3 minutes. At some point, I'll have a local AI coder who could create the core of the script in less than 1 second, but I'd still need to validate that the script generated the outputs with every edge condition handled. When AI fails, it can fail massively or slightly. For example, some AI created images seem to think a human with 3 hands is ok. At first glance, I didn't notice the issue, but once I saw the 3rd hand, my trust was gone. There's a famous group of photos - is it a muffin or a dog? https://www.freecodecamp.org/news/chihuahua-or-muffin-my-search-for-the-best-computer-vision-api-cbda4d6b425d/ Some of the best AI image recognition engines aren't 100% accurate. Would you like a dog delivered when you expected a stuffed animal or muffin?
Sven Asai
Don’t worry so much. This over dimensioned hype is sooner over than you might think or expect, and then a lot or most of the people, who now might be fired or ‘replaced’, will be urgently needed again, to clean up the AI mess and to return back to normal.
opheliajadefeldt
Where AI could make a real difference is in the boardrooms and government offices, they cannot be corrupted and one does not have to pay them grossly inflated bonuses or take bribes from corporations.
justsomeguy8008
One percent of the entire world's population? That is approximately 78 million people. Now, depending on what jobs those people do, such as healthcare or education, that number grows. I have heard of moonshots that are trying to make personalized AI for children that will grow with the child, be tailored to that child's learning style, and help that child learn in the best way for that child—basically, a private tutor for every child. Now they want to give children in the 3rd world these AI tutors so that every child gets the best education they can. Also, if they start mass producing humanoid robots with next-level AI, then blue-collar job industries will also change. So I'm not sure exactly what Mr. Peter Vukovic is referring to by saying "one percent or less of the population to benefit from AI." I myself am already benefiting from AI in everything from coming up with recipes for what to make for dinner to solving error messages that pop up on my computer to using AI like PI AI to bounce ideas off of or just to have a general conversation with when I am home relaxing on my own.
Roy Sophveason
Who would have thought we'd see a propaedeutic enchiridion straight out of Neil Stephenson's "The Diamond Age" come true within our lifetimes.