a16z: AI is not the end of jobs but the beginning of an era of intelligent inclusivity

Author: David George, General Partner at a16z; Source: a16z; Translation: Shaw, Golden Finance

The panic about a “permanent underclass of unemployment” promoted by artificial intelligence (AI) alarmists is fundamentally unfounded, and has long been a tired cliché. It is merely a new packaging of the labor supply fallacy reappearing once again.

The core idea of the labor supply fallacy is: The total amount of work society needs to complete is fixed and unchanging. It assumes that current workers, along with other practitioners, machines, and even today’s AI, are in a zero-sum game relationship. According to this logic: since the total useful work in society is constant, the more AI does, the less humans can do.

But this premise completely contradicts all our understanding of human nature, markets, and economics. Human desires and needs are never fixed. Nearly a century ago, Keynes famously predicted that automation would reduce humans to working only 15 hours a week. Clearly, he was wrong. He correctly foresaw that automation would lead to an oversupply of labor, but humans did not simply lie down and enjoy leisure; instead, they kept exploring entirely new productive activities, enriching their own time.

Admittedly, AI will inevitably replace some tasks and compress certain jobs (evidence already suggests this process has begun). Every disruptive technology has historically reshaped labor markets, and this time is no different. But claiming that AI will cause permanent large-scale unemployment across society is just a marketing stunt, a fallacious economic argument, and a gross neglect of historical laws. On the contrary, increased productivity will boost labor demand because labor itself will become more valuable.

Below is our complete logical reasoning.

Humans Are Not Doomed? That’s Overstated

We agree with the doomsayers — in fact, anyone with a clear eye can see: the cost of cognitive labor is plummeting. Tasks once considered exclusive to the human brain are now being performed increasingly well by AI.

The alarmists’ argument is: “If AI can think for us, then humans’ protective moat will vanish instantly, and human ultimate value will drop to zero.” Humans will thus be completely replaced. In their view, the work and thinking that humans need and desire have reached their end; now AI will shoulder an ever-growing share of mental workload, and humans will gradually become redundant and phased out by the times.

But the reality is quite the opposite: historical precedents and logical reasoning show that when a powerful production factor’s cost drops significantly, the economy never stagnates. Lower costs, improved quality, and faster efficiency make new products feasible, and overall demand expands outward. The Jevons paradox applies perfectly here.

When fossil fuels made energy cheap and abundant, humans didn’t just eliminate traditional jobs like whalers and lumberjacks; we also invented entirely new things like plastics.

Contrary to the doomsayers’ view, we have every reason to believe AI will produce similar effects. As AI takes on more and more mental work, humans will gain freedom to explore unprecedented, larger-scale, more ambitious frontiers.

If we look at history, we can foresee: Technological revolutions will ultimately expand the entire economy.

Every dominant economic sector in the past has eventually yielded to larger, emerging industries… and this process only enlarges the overall economic scale.

Today’s tech industry already surpasses the old finance, railway, and industrial sectors in size; but in terms of its share of the overall economy and markets, there is still vast room for growth. Productivity improvements are not a zero-sum game but a powerful positive-sum growth engine. Humanity’s delegation of large amounts of work to machines ultimately results in a bigger, more diverse, and more complex economy and labor market.

Doomsayers deliberately ignore human innovation history, cherry-picking only the current moment of plummeting mental labor costs, and treat this instant as the entire story. They only see AI replacing individual tasks, stopping there, and fail to delve deeper.

“The human brain’s output will increase tenfold, but we won’t do more thinking or creating; instead, everyone will just relax, take early lunch breaks, and do nothing.” This idea is not only unimaginative but also severely lacking in basic empirical observation. Doomsayers package this narrative as “realism,” but history has never shown such a scenario.

The Failure of Luddites

Let’s revisit history to see what truly happened when productivity breakthroughs disrupted economies.

Agriculture

In the early 20th century, before widespread mechanization, about one-third of U.S. labor was engaged in agriculture. By 2017, that share had fallen to about 2%.

If automation truly caused permanent unemployment, tractors should have completely destroyed the labor market. But reality was the opposite: agricultural output nearly tripled, supporting a massive population increase. Those who left farms didn’t face permanent unemployment; instead, they flooded into previously unimaginable new industries: factories, supermarkets, office buildings, hospitals, laboratories, and later, service and tech sectors.

Undeniably, technology disrupted the career paths of ordinary farmers; but at the same time, it released vast amounts of labor and resources, spawning a whole new economic system.

Electrification

The process of electrification follows a similar logical pattern.

Electrification is not just about replacing one energy source with another. It replaced traditional drive shafts and belts with independent electric motors, forcing factories to reorganize around new production processes, and also gave rise to entirely new categories of consumer and industrial products.

This is a typical feature of every stage of technological revolution, as summarized by Carlotta Perez in “Technological Revolutions and Financial Capital”: early stages involve huge upfront investments and financial capital chasing after them; durable goods costs plummet; and then durable goods manufacturers enter a long-term boom lasting generations.

It took a long time for electricity to unleash its powerful productivity effects. In the early 20th century, only 5% of American factories used electric power, and less than 10% of households were wired.

By 1930, electricity provided nearly 80% of the power for manufacturing, and over the following decades, labor productivity doubled.

Productivity gains did not weaken labor demand; instead, they spurred manufacturing expansion, increased sales staff, expanded credit, and fostered overall commercial prosperity. Not to mention the secondary effects of labor-saving devices like washing machines and cars: they enabled more people to engage in higher-value work previously out of reach.

As car prices fell, auto production and employment exploded.

This is the true power of general-purpose technology: restructuring the economy and continuously expanding the boundaries of useful work.

This pattern repeats throughout history. Did spreadsheet software like VisiCalc and Excel eliminate bookkeepers? Absolutely not. The technological leap in calculation efficiency actually led to a surge in bookkeeping jobs, and even created a whole new industry in financial planning and analysis (FP&A).

We reduced about 1 million “bookkeeper” jobs but created roughly 1.5 million “financial analyst” jobs.

New service sector jobs

Of course, task automation doesn’t always lead to employment growth in adjacent sectors. Sometimes, productivity surplus in one industry spurs entirely new jobs in unrelated fields.

But some ask: If AI only makes a small elite of people super-rich and leaves others far behind, what then?

At the very least, these ultra-rich will spend their wealth, creating a whole new service industry from scratch — as history consistently shows:

The dramatic increase in productivity and wealth creation has historically spawned a host of entirely new careers. Even before the 1990s, many of these careers were technically feasible; but without rising incomes and ample labor supply, they could never have materialized.

No matter how people view services for the wealthy, the ultimate outcome is that everyone’s life improves. Demand expansion drives median wages higher, which in turn creates more people entering the affluent class.

Stripe economist Ernie Tedeschi offers a representative case: the complete transformation of a profession how technology disrupts, reshapes, and revives it: Travel agents.

Does technology reduce the demand for travel agencies? Certainly:

Today, the total payroll of travel agencies is about half of what it was at the turn of the century, almost entirely due to technological advances.

Does this mean technology kills jobs? No. Travel agents haven’t faced permanent unemployment. They’ve found new roles in other parts of the economy; after adjusting for demographic shifts, the employment ratio is roughly the same as in 2000.

Meanwhile, those remaining in the tech-enabled travel industry benefit from increased productivity, and their wages are higher than ever.

“In 2000, during the industry’s peak, travel agents’ average weekly wages were only 87% of the overall weekly median wage. By 2025, that ratio had risen to 99%, meaning travel agency wages outpaced other private sector fields during this period.”

So even in this scenario, although technology has reduced the number of travel agency jobs, overall employment among the eligible working population remains stable; and the remaining travel agents are earning record-high incomes.

Enabling Is Greater Than Replacing (And New Jobs Yet to Come)

This final point is crucial and again underscores: AI doomsayers only see the tip of the iceberg.

For some jobs, AI is a disruptive survival threat. But for many more, AI is an enhancer, greatly increasing the value of those roles. Behind every job at risk of being replaced by AI, there’s a different set of jobs that will benefit.

Goldman Sachs estimates that the AI substitution effect has already been offset and even surpassed by AI-enabled productivity gains.

Notably, corporate management now clearly values AI empowerment over job replacement.

Currently, the frequency of mentions of “AI enabling efficiency” in earnings calls is roughly 8 times higher than mentions of “AI replacing jobs.”

Even though Goldman Sachs has not listed software engineers among “AI-enabled roles,” they are arguably the most typical example of AI-enhanced professions.

AI acts as a productivity multiplier for programmers. Code commits are soaring (along with new applications and startups), and demand for software engineers is rebounding, returning to growth.

Jobs related to software development, in both absolute numbers and as a share of the overall employment market, have been growing since early 2025.

Is this the result of AI? It’s too early to say definitively, but there’s no doubt that AI has greatly empowered software engineering. Moreover, today, every executive considers AI a top priority.

All industries are trying to integrate AI into their operations, which naturally leads to large-scale hiring and transformation efforts. This only increases the value of specialized talent.

AI-related roles are driving salary growth faster than industry averages, especially in system design.

While these salary benefits may still be limited, they are in the very early stages. As more professionals enter the field, employment opportunities will expand. In any case, these are not the data points doomsayers want you to see.

Meanwhile, Lenni Lahitki, founder of the well-known tech community newsletter “Lenny’s Newsletter,” notes that the number of job openings for product managers (PMs) has rebounded since the industry’s rate environment plummeted earlier, reaching a new high since 2022.

The simultaneous growth in hiring for software engineers and product managers is a perfect example that the labor supply fallacy is false. If AI replaced human mental work one-to-one, logically, we should see: “Fewer engineers needed as product managers,” or “Fewer product managers as engineers.” But that’s not what we see. Instead, demand for both roles continues to rise because people can accomplish more and larger-scale work with AI.

This is the core flaw of AI doomsayers: Lack of imagination. They focus only on scattered tasks that automation can replace, ignoring the emerging frontiers of demand that will give rise to entirely new careers we can’t even imagine today.

Most of the new jobs created since 1940 didn’t exist in 1940. By 2000, it was easy to foresee many travel agents losing their jobs; but it was hard to imagine an entire cloud migration industry serving mid-sized enterprises — which only emerged over a decade later, long after the cloud era was still distant.

What does the current reality tell us?

So far, our analysis has been based on theory and historical precedents, both of which favor the optimistic view.

And that’s true. Every breakthrough in productivity that unleashes potential tends to expand demand or reallocate surplus production into other parts of the economy. This means jobs will increase: existing roles will be greatly enhanced, and many entirely new careers will emerge that we can’t even conceive of now.

If, by some chance, this time the pattern is broken, doomsayers will need to produce rigorous, solid evidence — not just panic-mongering and hollow sentiments.

Job displacement does not destroy human civilization; quite the opposite, this logic is sound. Human nature is never content with the status quo or stagnation. We complete one task, then pursue the next.

Beyond theory and history, what does the actual data say about the relationship between AI and employment? It’s important to note that we are still in the early stages, and the pros and cons are not yet fully apparent; but current mainstream data does not support the doomsayers’ claims. Even conservatively, the data shows no significant fluctuations in employment; and increasingly, new data points to the opposite conclusion: AI creates more jobs than it displaces.

Let’s look at some academic studies — not exhaustive, just recent representative papers:

• Artificial Intelligence, Productivity, and Labor: Evidence from Business Executives (Federal Reserve Bank of Atlanta Working Paper 34984): The results show that, although AI adoption has not yet significantly changed overall employment levels, it is already reshaping task allocation and occupational structures within firms. Routine clerical and administrative jobs are more easily replaced by AI; analysis, technical, and managerial roles tend to be complemented and empowered by AI.

• Corporate-level AI data (Federal Reserve Bank of Atlanta Working Paper 2026-3): Based on four surveys, over 90% of firms report that AI has had no impact on their employment over the past three years.

• 

• Microstructure of AI diffusion: Evidence from firms, business functions, and employee tasks (U.S. Census Bureau Economic Research Center, CES 26-25): The study finds that employment changes driven by AI are generally mild, with only about 5% of AI-adopting firms reporting employee number impacts; among these, roughly equal proportions report job increases (2.3% firm weight, 3.7% employment weight) and decreases (2.0% firm weight, 2.4% employment weight).

• 

• Tracking AI’s impact on the labor market (Yale Budget Lab, April 16, 2026): “Despite widespread societal concern that AI will disrupt the labor market, our data suggests these fears are largely subjective guesses. The current data indicates that AI’s overall impact on the labor market remains stable, with no macroeconomic disruptive shocks.”

The reasoning is straightforward. Recent studies repeatedly convey the same conclusion: overall employment shows no net change, but there is a structural reconfiguration of jobs and tasks. Some research even indicates that AI’s net effect on hiring is positive.

Beyond the “no significant overall change” conclusion, one notable exception is that studies from Stanford, Dallas Fed, and the Census Bureau find (to varying degrees) that entry-level jobs with high AI penetration are becoming harder to find.

However, before jumping to the conclusion that “AI is destroying entry-level jobs,” it’s important to note that these studies also find that the number of entry-level jobs empowered by AI is increasing, and the number of unaffected entry-level roles is also rising.

Even if we accept that some entry-level jobs are being replaced by AI — not due to macro hiring cycles or demographic aging — the broader data clearly shows that the overall effect of AI on employment is close to zero.

This may be the most concise summary of the current impact of AI on employment.

“Currently, there is no statistically significant correlation between AI and unemployment rate or employment growth rate.”

On the market level, two trends may coexist: on one hand, talent flows into AI-empowered roles; on the other, AI-replaced roles push out existing practitioners.

Industries with higher AI-empowerment hiring growth tend to have lower unemployment rates, while those with higher AI-replacement risks tend to be the opposite.

In other words, overall employment remains neutral, but the internal structure has changed: some jobs are disappearing, others are emerging; some roles are devalued, while others are gaining premium and value. At the current growth rate, demand for programmers will surpass pre-pandemic levels within less than two years. AI alone has even stabilized the San Francisco office market.

This aligns with our initial view: AI will inevitably eliminate or compress certain jobs and industries, but to think that’s all there is — that’s a mistake. Facing a transformative technology, we should see labor markets rebalancing and ultimately growing, not large-scale unemployment. History has repeatedly confirmed this pattern, and it is almost certain to repeat now, with the trend already underway.

Knowledge Work Is Just Beginning

Though this is a well-worn phrase, it remains true: this is not the end of knowledge work, but rather its beginning.

Automation stripping away repetitive, low-level tasks pushes human labor toward higher-value levels. The logic is simple: humans inherently seek to expand outward. When a scarcity in one layer is broken, people move toward new, higher-level needs. When food costs decline, more spending shifts toward housing, healthcare, education, travel, entertainment, convenience, pets, safety, beauty, anti-aging, and other areas.

The labor market is similar. New careers will continually emerge because human ambition is endless; conquering old domains will give rise to new frontiers waiting to be explored.

Currently, the number of new business registrations is exploding, and there is a high correlation with AI deployment.

The number of new applications listed in app stores is growing at a 60% year-over-year rate.

There is no reason to see the modern economy as a museum of old careers. Instead, it is a creative resource allocation machine, constantly generating new jobs, new work, new goals, and new innovations.

For a long time, robot technology was mostly seen as science fiction because the computational demands in dynamic environments were too high. Today, AI is turning a new robot industry from concept to reality.

Data set sizes in robotics are exploding; in just two years, the industry’s share has leapt from tenth place to first.

Massive numbers of robot-related jobs previously had no market demand — until AI unlocked this latent demand.

Again, it’s important to emphasize: all these points do not mean all existing jobs will survive intact. The U.S. Bureau of Labor Statistics projects that roles like customer service representatives and medical transcriptionists will shrink, and this trend may already be underway.

Some jobs will disappear, others will shrink. Industry structures will inevitably adjust, and there will be pain during the transition; productivity improvements will gradually permeate the entire economy, often taking time. We should empathize with the shocks this causes and actively promote a smooth transition, including proactive retraining — a16z is very supportive of such initiatives.

The purpose of productivity growth is to help humans escape dull, heavy mechanical labor — and this AI revolution will be no exception.

But the so-called AI job apocalypse only holds under a ridiculous assumption: that human desires and innovative thinking will come to a complete halt the moment intelligent technology becomes cheap and widespread. This is obviously absurd. Personally, I do not buy into the “Wall-E” style end-of-work narrative, and I believe many others share this view.

On a macro level, the future will not be a society of universal unemployment and passive leisure, wasting life on streaming entertainment and autonomous vehicles.

The future looks like this: costs of intelligence plummet, market size continues to expand, new companies keep emerging, and new industries are born. Humans will engage in higher-level, more creative work. The total amount of work is never fixed, and the demand for mental effort has never been fixed — it never was in the past, and it won’t be in the future. AI is not the end of labor, but the beginning of an era of intelligent inclusivity.