Author: David George
Compiled by: Felix, PANews
Editor’s Note: Currently, AI “doomsday” narratives appear to dominate mainstream discourse, with widespread global anxiety over AI “stealing jobs” and mass unemployment. Experts across various fields are offering strategies for the disruptive changes AI is expected to bring. However, in an article, a16z General Partner David George argues that these doomsday views are baseless, lacking evidence and imagination, and failing to understand human nature. Below is the full text.
The argument put forth by AI panic proponents about a “permanent underclass” is unconvincing. This is nothing new—just the old “lump of labor” fallacy repackaged.
The "fixed pie fallacy" claims that the total amount of work needed in the world is fixed. It assumes a zero-sum game between existing workers and anyone or anything else who might do the same work—whether other workers, machines, or now AI. If the total amount of useful work required is fixed, then if AI does more, humans must necessarily do less.
The problem with this premise is that it contradicts everything we know about people, markets, and economics. Human needs and desires are never fixed. Keynes predicted nearly a century ago that automation would lead to a 15-hour workweek, but his prediction proved wrong. While he was correct about automation causing a “labor surplus,” we did not sit idly by—we found new and different productive activities to fill our time.
Of course, AI will eliminate certain jobs and compress others (and there is evidence this may already be happening). The landscape of the labor market will change, as it inevitably does with every transformative technology. However, the notion that AI will lead to widespread, permanent unemployment across the entire economy is poor marketing hype, bad economics, and a disregard for history. On the contrary, increased productivity should raise demand for labor, as labor becomes more valuable.
Here are our reasons.
“Are humans done for?” Don’t make jokes.
We agree with the doomsayers that the cost of cognition is plummeting. AI is becoming increasingly adept at tasks that were until recently considered the exclusive domain of the human brain.
Doomsayers argue: "If AI can replace our thinking, then humanity's 'moat' will disappear, and our ultimate value will drop to zero." Humanity is finished. Clearly, we have already completed all the thinking we needed or wanted to do, and now AI will take on an increasing cognitive load, gradually leading to humanity's obsolescence.
However, the fact is: precedent (and intuition) show that when the cost of a powerful input falls, the economy does not stall. Lower costs, improved quality, faster speeds, and new products become viable, causing demand to expand outward. The Jevons Paradox holds true again. When fossil fuels first made energy cheap and abundant, we did more than put whalers and lumberjacks out of work—we invented plastic.
Contrary to the doomsayers' views, we have every reason to expect AI to generate similar impacts. As AI takes on an increasing cognitive load, humans will be free to explore new frontiers more ambitious than ever before.
History shows that technological advancements will inevitably expand the economic pie.
Each "leading economic sector" is replaced by a larger subsequent sector... which in turn further expands the scale of the economy.
The scale of today’s technology far exceeds that of finance, railroads, or industry, yet its share of the economy or the broader market remains relatively small. Increases in productivity are far from a zero-sum game; instead, they represent a powerful positive-sum force. Delegating so much work to machines ultimately leads to a larger, more diverse, and more complex economy and labor market.
Doomsayers want you to ignore the history of innovation and focus solely on the sharp decline in cognitive costs, treating it as the whole truth. They see task substitution and then stop thinking.
We will increase cognitive output tenfold, but we won’t think more—we’ll just pat our bellies and head early to lunch, and everyone else will do the same. This claim not only reveals a severe lack of imagination but also a failure to observe basic facts. Those who espouse doom scenarios call this “realism,” but it is simply impossible.
The failure of Luddism
(PANews Note: Luddism refers to a social movement initiated by the British working class in the early 19th century that opposed the Industrial Revolution by destroying industrial machinery to protest deteriorating working conditions and unemployment.)
Now let’s see what happens when a massive leap in productivity sweeps through the entire economy.
Agriculture
At the beginning of the 20th century, before agricultural mechanization became widespread, about one-third of the U.S. workforce was employed in agriculture. By 2017, this proportion had declined to approximately 2%.
If automation led to permanent unemployment, tractors should have utterly destroyed the labor market. Yet that’s not what happened—agricultural output nearly tripled, supporting massive population growth, and these workers didn’t become permanently unemployed; instead, they flooded into industries, factories, stores, offices, hospitals, and laboratories that were previously unimaginable, eventually entering the service and software sectors.
So, while it is true that technology has disrupted the career prospects of ordinary farm workers, it has also unleashed a surplus of global labor (and resources) and given rise to an entirely new economic system.
Electrification
The development of electricity has followed a similar path.
Electrification is not merely replacing one energy source with another. It replaces drive shafts and belts with individual motors, forcing factories to reorganize around entirely new workflows and giving rise to entirely new categories of consumer and industrial products.
This is precisely what we expect at different stages of the technological revolution, as documented by Carlota Perez in her book Technological Revolutions and Financial Capital: massive upfront investment and financial gains, significant declines in the cost of durable goods, followed by generational prosperity for manufacturers of durable goods.
The advantages of electricity as a productive force did not emerge overnight. At the beginning of the 20th century, only 5% of U.S. factories used electrically powered machinery, and fewer than 10% of homes were electrified.
By 1930, electricity supplied nearly 80% of the power for manufacturing, and labor productivity doubled over the following decades.
Increased productivity did not reduce the demand for labor; instead, it led to more manufacturing, more salespeople, more credit, and more business activity, not to mention the ripple effects of labor-saving devices like washing machines and cars. These devices enabled more people to engage in higher-value work that was previously inaccessible.
As car prices fell, car production and employment both experienced explosive growth.
This is the power of truly general-purpose technology: it restructures the economy and expands the boundaries of useful work.
We’ve seen this time and again. Did VisiCalc and Excel end the careers of bookkeepers? Absolutely not. More efficient computing technologies actually led to a surge in the number of bookkeepers and gave rise to the entire field of financial planning and analysis (FP&A).
We lost approximately 1 million "bookkeepers" but gained around 1.5 million "financial analysts".
New service sector jobs
Of course, job displacement does not always lead to job growth in related economic sectors. Sometimes, increases in productivity result in new jobs in entirely unrelated industries.
But what if AI means that some people will become extremely wealthy, while others are left far behind?
At the very least, those ultra-wealthy individuals will have to spend their money somewhere, just as they did before, creating entirely new service industries from scratch:
Significant increases in productivity and the resulting wealth creation have given rise to entirely new fields of work that might never have emerged without rising incomes and increased labor supply (even though these fields were technically feasible well before the 1990s). Regardless of opinions about service industries catering to the wealthy, the ultimate outcome has benefited everyone, as increased demand has led to substantial rises in median wages (thereby creating more “affluent” individuals).
Stripe's in-house economist, Ernie Tedeschi, provided a comprehensive case study on how technology has disrupted, transformed, and reshaped the travel agent profession.
Has technology reduced the need for travel agencies? The answer is yes.
The number of employees in travel agencies today is about half of what it was around the year 2000, almost certainly due to technological advancements.
So does this mean technology has killed jobs? The answer is no, because travel agents did not permanently lose their jobs—they found employment in other sectors of the economy, and today’s overall employment-to-population ratio is essentially unchanged since 2000 (after adjusting for population aging).
Meanwhile, for those remaining in today’s technology-empowered travel agency industry, increased productivity means higher wages than ever:
In the peak year of 2000, the average weekly wage of travel agents accounted for 87% of the overall average weekly wage. By 2025, this proportion had reached 99%, indicating that during this period, travel agents' wages grew faster than those in other private sectors.
Therefore, although technology has indeed impacted employment in the travel agency sector, overall, the employment rate among the working-age population remains unchanged, and the situation for remaining travel agency employees is better than ever.
Enhance > Replace (and jobs that have not yet emerged)
The final point is very important and again shows that doomsayers only tell a small part of the story.
For some jobs, AI poses an existential threat. Indeed. But for others, AI is a multiplier: making these jobs more valuable. For every job at risk of being replaced by AI, there are other jobs poised to benefit:
Goldman Sachs estimates that the "AI substitution" effect is far less than the "AI augmentation" effect.
It is worth noting that the management team also appears to focus more on enhancement rather than replacement:
As of now, during earnings calls, "AI as an enhancement" has been mentioned approximately eight times more frequently than "AI as a replacement."
Although Goldman Sachs did not even list software engineers on its "enhanced" talent roster, they may be the best example of AI-enhanced talent.
AI is a multiplier for coding. Not only are Git push counts surging, but so too is the demand for software engineers:
Since the beginning of 2025, software development roles have continued to grow, both in number and as a percentage of the overall job market.
Is this related to AI? To be honest, it might still be too early to draw conclusions, but AI undoubtedly enhances productivity in software engineering, not to mention that AI has become a top priority for executives at every company.
Given that everyone is striving to explore how to integrate AI into their businesses, it's no surprise that companies are engaging in large-scale hiring, which will undoubtedly enhance the value of certain employees rather than diminish it.
The adoption of AI appears to be driving above-average wage growth, particularly in the field of system design.
These gains may still be limited for now, but we are still in the early stages. As expertise expands, opportunities will grow accordingly. Regardless, this is not the data those “doomsayers” want you to see.
Meanwhile, according to Lenny Rachitsky (founder of Lenny’s Newsletter, a platform for tech industry professionals), the number of open project manager positions continues to rise (after having dropped significantly due to interest rate fluctuations) and is now higher than at any point since 2022:
The growth in hiring for software engineers and product managers precisely demonstrates the validity of the "lump of labor fallacy." If AI fully replaced human thinking, you might assume that "product managers would need fewer engineers," or conversely, that "engineers would need fewer product managers"—but that is not the case. We are seeing sustained demand for both types of talent because the key lies in people becoming more productive.
This is why the doomsayers’ claims are essentially a lack of imagination. They focus only on jobs that will be replaced by automation, while ignoring the emerging fields that will create entirely new roles we haven’t even imagined yet:
Most jobs created since 1940 didn’t even exist in 1940. By 2000, it was easy to imagine travel agents losing their jobs, but it was much harder to envision a mid-market tech services industry built around “cloud migration,” since cloud computing was still at least a decade away from widespread adoption.
What is the current situation?
So far, the discussion has primarily focused on theory and precedent, both of which support the optimists:
That's right. Every increase in productivity leads to growth in demand or the reallocation of surplus resources to other areas of the economy. This means more jobs, many of which will see significantly increased value, and even entirely new types of jobs never before seen. If this time is different, then those “doomsayers” must present stronger arguments—not just empty rhetoric.
The idea that "job displacement" is not the end of civilization (in fact, quite the opposite) makes a lot of sense. Human nature is inherently restless—we complete one job and immediately seek another.
But setting aside theory and precedent, what do the actual data show about AI and employment? Although we are still in the early stages (for better or worse), existing data does not support the doomsday narrative. If anything, the picture has been one of “no significant change,” but emerging data points in the opposite direction: AI is creating more jobs than it displaces.
First, let’s start with some academic research. This is not an exhaustive literature review, but rather a few examples of recent papers:
- AI, Productivity, and Labor: Evidence from Corporate Executives (NBER Working Paper 34984): “In summary, these results suggest that while AI adoption has not yet led to significant changes in overall employment, it has begun to reshape the allocation of tasks and occupations within firms. In particular, routine clerical and administrative activities appear more susceptible to substitution, while analytical, technical, and managerial tasks are more frequently described as being augmented by AI.”
- Enterprise Data on AI (Atlanta Fed Working Paper 2026-3): “Across four surveys, an average of over 90% of firms estimated that AI had no impact over the past three years.”
- The Microstructure of AI Diffusion: Evidence from Firms, Business Functions, and Worker Tasks (U.S. Census Bureau Center for Economic Studies, Working Paper CES 26-25): “The employment effects of AI remain limited, with only about 5% of firms using AI reporting an impact on workforce size: the proportions of firms reporting increases (2.3% weighted by firm, 3.7% weighted by employment) and decreases (2.0% weighted by firm, 2.4% weighted by employment) are nearly equal.”
- 《Tracking the Impact of AI on the Labor Market》 (Yale Budget Lab, April 16, 2026). “Although there is widespread anxiety about AI’s impact on today’s labor market, our data suggests that this remains largely speculative. The picture of AI’s impact on the labor market presented by our data reflects stability rather than significant disruption at the economic level.”
The latest research consistently concludes that “overall, there is no change, but there is evidence of reallocation between jobs and tasks.” In some cases, the implementation of AI has had an even positive net impact on hiring.
But there is a notable exception to the claim of “no change.” Researchers from Stanford University, the Federal Reserve Bank of Dallas, and the U.S. Census Bureau have all found (to varying degrees) that entry-level positions with “high AI exposure” are becoming harder to find. However, before anyone concludes that “AI is eliminating entry-level jobs,” it’s worth noting that these researchers also found an increase in entry-level positions where AI plays an assisting role (as well as in roles where AI has no impact at all).
However, even if we temporarily assume that AI is “eliminating” certain entry-level positions (rather than being influenced by broader cyclical hiring trends and “aging in place”), the data at the broader macro level clearly shows that AI’s overall impact on employment is essentially zero.
This may be the most concise summary of AI's impact on employment:
There is still no statistically significant relationship between AI and unemployment rates or employment growth.
Perhaps people have a certain preference for AI-augmented roles, and also a certain impetus toward AI-replaced roles:
For industries described as "AI-enhanced," hiring growth appears stronger (with lower unemployment rates), while the opposite is true for industries at higher risk of "AI replacement."
In other words, the overall situation is neutral but not static: some jobs disappear, some emerge, some depreciate, while others appreciate. At this rate, job postings for developers will surpass pre-pandemic levels in less than two years. AI may have single-handedly saved San Francisco’s job market.
This was our starting point: AI will undoubtedly eliminate or compress certain jobs (and businesses), but it is wrong to think this is the end of the story. The real expectation we should have for this transformative technology is the realignment of the labor market—ultimately leading to growth, not widespread unemployment. This has happened before, and it is almost certain to happen again (and it already appears to be underway).
Knowledge work has only just begun
This may sound cliché, but it’s true: this is not the end of knowledge work—it’s just the beginning.
Automation has removed repetitive tasks and elevated human work to a higher level. The reason is simple: humans crave expansion. When one scarcity disappears, people move toward higher-order needs. As food prices fall, we spend more on housing, healthcare, education, travel, entertainment, amenities, pets, security, beauty, and longevity.
The labor market is no different. New jobs continually emerge because human ambition never ceases, and conquering old frontiers reveals new ones that need to be conquered.
The emergence of new businesses has exploded in growth and is highly correlated with the application of AI:
The speed at which new apps are launched on app stores has increased by 60% year-over-year:
We should not view the modern economy as a museum of yesterday’s jobs. Instead, it is a creative machine for allocating resources, constantly generating new jobs, new tasks, new goals, and new inventions.
Robotics was long regarded as science fiction due to the excessive computational demands in dynamic environments. But AI is bringing a brand-new robotics industry into view:
Robot-related datasets have experienced explosive growth, rising from tenth place to first within just two years.
Before AI truly comes into play, there are numerous jobs in the robotics field that remain unclaimed.
Again, this does not mean that all jobs will be spared. The U.S. Bureau of Labor Statistics (BLS) projects a decline in jobs for customer service representatives and medical transcriptionists, and this reduction may have already begun:
Some jobs will disappear, others will shrink. The economy will undergo adjustment and a painful transition period, and gains in productivity may take time to gradually benefit the entire economy (with ups and downs). We should understand these changes and work to make them as smooth as possible, including actively promoting vocational retraining.
The goal of increased productivity is to eliminate arduous labor, and this time is no different. However, the claim that AI will lead to mass job loss only holds if one assumes that human demand and desires abruptly cease the moment AI becomes cheap. This is absurd. Personally, I do not subscribe to the “WALL·E” narrative, and I believe I am not alone in this view:
From a macro perspective, the future is not an era of mass unemployment; rather, after retirement, we’ll be enjoying Netflix with leisurely ease, riding electric scooters.
The future is cheaper intelligence, larger markets, new companies, new industries, and higher-level human work. The amount of work is not fixed, nor is cognitive capacity—neither has ever been. AI is not the end of work, but the beginning of an era enriched by intelligence.
Related reading: Transcript of Jensen Huang’s latest podcast: NVIDIA’s future, the “AI doomsday” theory, corporate moats...