[…]
Addressing Shanghai’s World AI Conference in July, China’s leader, Xi Jinping, said that AI should be “an important driver for shared prosperity and common security”.
But that goal is being pursued amid an increasingly fragile labour market. As well as suffering from high levels of unemployment, China has seen tens of millions of workers drop out of formal employment into the gig economy. According to one thinktank, the number of people in flexible employment will rise to 320 million this year, up from 160 million in 2019. That is about 44% of China’s workforce.
[…]
Taxi drivers are one part of this huge cohort. Across the country, as previously heavyweight industries such as real estate and education have floundered, workers have turned to ride-hailing apps to make ends meet. But now the rollout of robotaxis, one of the most popular applications of AI that is already in use in China, is threatening their livelihoods.
“If the state really took you into consideration, Apollo Go would never have gone on the market,” says one Wuhan driver.
“If things keep going this way in the future, sure, rapid development and scientific progress are great, but ordinary people like us will have no place to survive,” she adds.
[…]
Tian Zemin, a cinematographer who works in Beijing, has worked in the film industry for nearly two decades. But now he’s out of work. “I spent six gruelling years working my way up to become a cinematographer, only to be replaced by a piece of software,” he wrote in a social media post about AI.
“I thought I would spend my whole life in this industry. But the emergence of AI has made me feel a huge shock for the first time,” he tells the Guardian.
Many of Tian’s peers are losing out as AI becomes more widely adopted across the film industry. His own rate as a freelancer is now 40% of what it was in 2019.
[…]
“We could be replaced at any moment,” he says. “Perhaps the only thing we can do is adapt as quickly as possible, and try to find the corner that AI still cannot replace – but that corner is getting smaller and smaller.”
That’s certainly concerning, but such forecasts are problematic in and of themselves, since they assume a status quo that could very well change.
For example, it’s by no means out of the question that tokens could become more expensive rather than cheaper in the future, since the AI giants are currently incurring massive losses across the board and will likely have to raise their prices at some point to become at least somewhat profitable. Either that, or they’ll have to find additional monetization opportunities, such as advertising. Both of these developments would affect the cost of using AI services, which could mean that human labor becomes actually cheaper than AI— there are already examples of this today.
Another point is that a kind of “AI fatigue” could set in—that is, products manufactured using AI for cost reasons could increasingly fall out of favor with consumers due to their poor quality, so that they would rather pay a little more for a higher-quality product made without AI that uses this fact as a selling point.
I consider it fairly certain that AI will have a significant impact on the labor market, but I believe it is uncertain whether its influence will be as severe as is sometimes portrayed at present—especially since there are now various studies concluding that massive investments in AI have not yielded the intended results and that AI is more expensive than human labor.
Therefore, I think there’s still a great deal of hype here, which is being deliberately pushed by AI providers because they rely on it to justify further investments. I consider it out of the question that what the manufacturers promise—a kind of fully automated world in which AI could take over nearly every job—will ever come to pass, because LLMs are simply not capable of this; after all, they are not artificial intelligence, but merely language models.