This article is an original piece by Shangyin Media. For reprints, please contact the background team.

Human fear of AI is likely to remain a significant topic for a long time.

Some time ago, the U.S. research institution Citrini Research released a report titled 'Global Intelligence Crisis 2028,' which caused a stir online. Many people fell into panic over AI, even triggering a sharp decline in the stock prices of AI-related companies.

What exactly does this report say?

It emphasizes that this is merely a 'thought experiment,' depicting an economic collapse triggered by the 'overly successful' development of AI.

For example:

AI is so efficient that knowledge is no longer scarce, making economic activities based on it unsustainable. White-collar workers are replaced on a large scale, lose their income, default on mortgages, and the financial market collapses.

Corporate profits grow due to AI efficiency gains, but with widespread unemployment and reduced consumption, a 'phantom GDP' phenomenon emerges, where 'output grows while the consumption engine stalls.'

Existing business models also face significant challenges. Some models built on human laziness, information asymmetry, and brand dependency collapse due to AI intervention, such as software services, intermediary platforms (food delivery, travel booking), payment processing (credit card interchange fees), and private credit.

Just a few days after the report's release, U.S. fintech company Block announced layoffs of approximately 4,000 employees, reducing its workforce from 10,000 to 6,000. The reason cited was the development of an in-house AI tool named 'Goose,' which significantly boosted efficiency and replaced a substantial portion of the workforce.

In the past, large-scale layoffs by companies almost always occurred during significant survival crises. However, Block reported a gross profit of $10.36 billion for the year, a 17% year-on-year increase, making these 'profitable layoffs.' Wall Street greatly appreciated this move, and Block's stock price surged by over 24% at one point.

The media exclaimed, 'AI has begun to massively replace humans!' Workers fell deeper into anxiety, believing that AI had left humanity with little time.

In fact, as we wrote in our previous article, 'The Truth Behind 'AI Causing 100,000 Layoffs in Silicon Valley,'' while some layoffs by tech giants under the banner of embracing AI may indeed be due to AI replacement, a significant portion cannot be blamed on AI. For example, during 2020-2022, tech companies expanded massively due to a surge in online demand during the pandemic. The Federal Reserve lowered interest rates to near-zero levels, significantly reducing corporate financing costs and directly prompting many tech companies to embark on large-scale expansions. As a result, Amazon's workforce doubled, while Google and Microsoft each expanded by nearly 70,000 employees. During this period, domestic companies like ByteDance, Meituan, and Tencent also surpassed 100,000 employees. Block, too, expanded from nearly 4,000 to 12,000 employees.

The primary reason for the wave of layoffs by tech companies in subsequent years was the short-term absorption of a massive workforce.

According to data from the U.S. job information website Layoffs.fyi, tech companies globally laid off approximately 160,000 employees in 2022, about 260,000 in 2023, nearly 150,000 in 2024, and news of '100,000 layoffs in Silicon Valley' emerged again last year.

'Global Intelligence Crisis 2028' reads like a ' wish-fulfillment fiction ' (a genre of fiction characterized by smooth, rapid, and upgrade-focused storytelling) from start to finish, ignoring the complexity of technological expansion in reality and the extrude (squeezing) and balancing effects of factors such as population, culture, economy, society, and policy on technological expansion.

Currently, the AI industry has not yet escaped the 'Solow paradox'—in 1987, computers were becoming increasingly widespread, but economist Robert Solow observed, 'You can see the computer age everywhere but in the productivity statistics,' meaning that despite widespread computer adoption, no significant productivity gains were evident.

It wasn't until the late 20th century, with transformations like Walmart's 'Retail Link' system boosting inventory turnover by 40% and Dell's revolutionary build-to-order production model, that computer technology truly unleashed productivity.

This demonstrates that new technology alone is insufficient to drive productivity gains; it also depends on a series of factors such as organizational change, business model innovation, and policy environment.

The OECD estimates that over the next decade, AI will only increase labor productivity by 0.4%-0.9%.

Therefore, the impact of the 'employment apocalypse' will be continually diluted by the time lag brought about by the 'Solow paradox' and will not arrive in the short term.

However, it cannot be denied that the impact and disruption of AI on employment will be unprecedented. Where lies the path forward for human employment?

Recently, I read 'China's New Employment Trends: How AI is Reshaping the Labor Market' by economist Cai Fang, which comprehensively and systematically analyzes this topic. It roughly summarizes three possible paths for human employment in the AI era.

First, the ideal scenario is achieving 'human-machine collaboration.'

AI is merely a technological platform. While it may destroy jobs through automation, it can also create more productive roles by transforming production processes. The direction AI takes depends on policy choices and institutional arrangements.

Human capabilities include cognitive and non-cognitive abilities. The former, such as arithmetic reasoning, vocabulary, text comprehension, mathematical ability, and coding speed, are easily measurable and can be imitated by AI. The latter, including motivation, self-control, adaptability, social skills, empathy, and compassion, are tacit knowledge unique to humans—things that are done but not spoken.

Thus, humans and machines are complementary in capabilities. AI does not need to be motivated by replacing jobs; machines can enhance human capabilities, with humans instructing machines on what to do, thereby increasing creativity and experience in services at a higher level and improving consumption quality and consumer surplus.

If robots merely replace humans and mass-produce what humans previously made, it only replicates industrial-era products faster, leading to severe oversupply and ignoring changes in consumer markets and product innovation.

Second, transitioning to roles characterized by 'Baumol's cost disease.'

What is 'Baumol's cost disease'?

Certain industries in the economy experience particularly slow productivity growth, with operating costs tending to rise over the long term and exhibiting significant income elasticity of demand.

Economist William Baumol initially used the performing arts industry as an example in his research, later expanding it to many fields, such as healthcare, education, social work, culture, sports, entertainment, public administration, social security, and social organizations.

Productivity in these areas is difficult to improve significantly, unlike in the digital or manufacturing industries, and many even rely on subsidies to survive. However, as society develops, demand in these areas generally increases.

For example, the recent surges in theater and concert popularity; in healthcare, which is a matter of life and death, no one dares to fully entrust it to AI, so demand will only rise; education in the AI era is not a competition between humans but between humans and machines, necessitating personalized instruction and holding significant development potential.

Generally, the higher the overall social productivity, the more it can support roles characterized by 'Baumol's cost disease,' which also absorb a substantial number of jobs.

Third, roles exhibiting 'reverse Kuznetsification.'

In economics, the 'Kuznets process' refers to the transfer of labor from low to high labor productivity sectors, driving industrial structure upgrading.

'Reverse Kuznetsification' is the opposite—a shift from high to low productivity sectors, such as white-collar workers losing their jobs and becoming ride-hailing drivers or delivery personnel.

If a large number of 'reverse Kuznetsification' roles exist, it means the response to technological employment shocks has failed, with many unemployed individuals competing for low-productivity, low-paying jobs, offsetting the overall productivity gains brought by technological progress. This is also an important reason for the 'Solow paradox' mentioned earlier.

Equally important is the productivity-sharing mechanism in the AI era.

For example, introducing an 'AI tax' to fund retraining and skill development for workers affected by AI.

Expanding the supply of public goods is also an important means of income redistribution. Once public goods account for a higher proportion of total social supply and demand than private goods, it alters the market price of labor factors.

For instance, if housing, education, and healthcare—areas where people spend significantly—become public goods, even switching to a lower-paying job would not reduce living standards but could improve them.

With a significant increase in social productivity, implementing universal social welfare may also become feasible. Current social security systems often target 'eligible' groups, providing only basic living guarantees with significant urban-rural disparities. Universal social welfare should be distributed without distinction, offering higher protection levels that rise with productivity gains.

In the AI era, workers' employment difficulties stem not from a lack of effort or poor choices, making it unnecessary to distinguish social welfare recipients.

It is also worth noting that Cai Fang mentions in his book that, facing the AI onslaught, the government should establish a public interest bottom line (bottom line) to protect workers and jobs, rather than merely acting as a balancing third party or a neutral adjudicator. Since an equilibrium of interests does not naturally exist, and workers and jobs are inherently the weaker party during major upheavals, tilting protection in their favor aligns with social interests.

In our article 'Why Can't Hard Work Alone Lead to Wage Increases?' we analyzed how U.S. workers gradually lost their bargaining power for wage increases. During major upheavals like the oil crisis, globalization, and technological progress, employers often used efficiency improvements and natural market and technological changes as excuses to systematically transfer economic risks and uncertainties to ordinary workers, while gaining more bargaining power.

In the current AI technological revolution, companies like Block attribute mass layoffs entirely to technological change, heralding not an AI era but an 'AI apocalypse.'

Thus, 'human fear outweighs AI.' Whether the negative effects of new technologies manifest or are contained depends on human actions.

Welcome to exchange and discuss with the author~