The Impact Of Automation On Economic Stability

As AI is replacing early-career jobs, the economy's productivity in the short-run increases, but productivity and welfare in the long run may decline. In a new paper, Enrique Ide argues that we may be witnessing "socially excessive automation of early-career work. Such automation may deliver immediate productivity gains, but it also erodes the skills of future cohorts and constrains long-run growth." Here's the abstract of his paper: "Recent advances in Artificial Intelligence (AI) have sparked expectations of unprecedented economic growth. Yet, by enabling senior workers to accomplish more tasks independently, AI may reduce entry-level opportunities, raising concerns about how future generations will acquire expertise. This paper develops a model to examine how automation and AI affect the intergenerational transmission of tacit knowledge—practical, hard-to-codify skills critical to workplace success. I show that the competitive equilibrium features socially excessive automation of early-career tasks, and that improvements in such automation generate an intergenerational trade-off: they raise short-run productivity but weaken the skills of future generations, slowing long-run growth—sometimes enough to reduce welfare. Back-of-the-envelope calculations suggest that AI-driven entry-level automation could reduce the long-run annual growth rate of U.S. per-capita output by 0.05 to 0.35 percentage points, depending on its scale. I further show that AI co-pilots can partially offset lost learning by assisting individuals who fail to acquire skills early in their careers. However, they may also weaken juniors’ incentives to develop such skills. These findings highlight the importance of preserving and expanding early-career learning opportunities to fully realize AI’s potential." What can policy do? In the concluding remarks, the paper offers several ideas: - government subsidies for "mentorship, apprenticeship, and other entry-level training arrangements" - "taxing entry-level automation" - reducing minimum wages for young workers - promoting AI systems that complement rather than replace entry-level jobs. Universities could also play a role by placing "greater emphasis on providing undergraduate students with opportunities to gain practical experience before they formally enter the labor market. Such initiatives would complement the traditional focus of undergraduate programs on codifiable knowledge and help foster the early development of tacit skills." Read the full paper here: https://lnkd.in/eMq3uktX (open access)

🚨 Jamie Dimon’s AI warning is a systemic-level reality check for the rest of us When Jamie Dimon, CEO of JPMorgan Chase, says “you can’t lay off 2 million truckers tomorrow,” he’s not debating technology or its potential He’s warning about social shock. What does all of this actually mean? AI is no longer a productivity story. It’s also a social stability story. If AI-driven automation is deployed too fast, too broadly, and without guardrails, the result isn’t so much efficiency but backlash from Society jamie Dimon is very direct ➡️ Mass, sudden job displacement = civil unrest ➡️ Markets don’t absorb shocks like that ➡️ Democracies don’t either Other leaders saying the this aloud as well • Dario Amodei (Anthropic): warns AI could wipe out a large share of entry-level white-collar jobs in just a few years • International Monetary Fund: says 60% of jobs in advanced economies are exposed to AI disruption • World Economic Forum: estimates 41% of employers plan workforce reductions due to AI by 2030 The real issue isn’t “jobs vs AI” It’s more of speed vs absorptive capacity. History shows: • The Industrial Revolution took decades • AI is compressing change into quarters. Leading Industry Analyst R "Ray" Wang says “AI marks a technology revolution of exponential scale, unlike anything we've seen before. In addition he says “Al-first firms prove that small, highly effective teams, augmented by digital labor, can outperform legacy behemoths." • Institutions (education, welfare, policy) move at glacial speed That gap is dangerous. The emerging consensus (even among CEOs) ✔️ Phased deployment, not shock therapy ✔️ Massive reskilling at national scale ✔️ Transitional income support ✔️ Clear rules for AI-driven layoffs ✔️ Shared responsibility between government and enterprise Now, who will make it happen… Dimon even said something radical for a Wall Street CEO: Government may need to step in and slow things down. Yes , AI will create enormous value. But unmanaged acceleration risks tearing the social fabric. The next decade won’t be won by who automates the fastest but by who transitions the most predictable and scalable way. AI without trust leads to instability. This is the real leadership test of the AI era.

AI Pioneer Warns Total Job Disruption Is Inevitable—and Coming Faster Than Expected Introduction One of the scientists who helped create modern artificial intelligence is now issuing a stark warning: widespread job displacement is no longer a distant threat but an unfolding reality. Yoshua Bengio, a Turing Award–winning AI pioneer, argues that automation will eventually reach every profession, from office roles to skilled trades, with profound consequences for work, society, and democratic stability. Key Warnings From an AI Insider Cognitive Jobs Are the First to Fall Desk-based, keyboard-driven roles are already being automated at scale. Junior and entry-level positions, especially affecting Gen Z, are disappearing fastest as companies delay hiring or replace roles with software. Firms are adopting “wait-and-watch” strategies, anticipating AI will absorb many functions within five years. Trade Jobs Offer Only Temporary Shelter Physical jobs like plumbing or electrical work may be harder to automate, but only in the short term. As robotics improves and data accumulates, AI-enabled machines will increasingly encroach on manual labor. Bengio argues that no category of work is fundamentally immune. Education No Longer Guarantees Security Highly educated graduates are facing some of the weakest job markets in years. Major technology companies are freezing roles expected to be automated in the near future. The long-standing promise that degrees ensure stability is eroding rapidly. From Job Loss to Societal Risk Bengio warns that unchecked AI development could destabilize democracy within two decades. He cites emerging AI behaviors, including resistance to shutdown, as early red flags. The pace of competitive deployment is pushing companies to take risks without adequate safeguards. A Reversal and a Call to Action Bengio has publicly expressed regret over not anticipating the risks earlier in his career. He founded LawZero, a nonprofit focused on safe and human-aligned AI systems. His message to business leaders is to slow down, coordinate, and prioritize long-term societal outcomes over short-term competitive gains. Why This Matters This warning is not coming from a skeptic, but from one of AI’s principal architects. If his assessment is correct, the challenge ahead is not simply retraining workers, but rethinking the role of work, income, and governance in an AI-driven world. The decisions made by today’s leaders will determine whether AI becomes a tool for shared prosperity—or a force that undermines economic security and democratic institutions alike. I share daily insights with 35,000+ followers across defense, tech, and policy. If this topic resonates, I invite you to connect and continue the conversation. Keith King https://lnkd.in/gHPvUttw

𝗧𝗵𝗲 𝗙𝗼𝗿𝗴𝗼𝘁𝘁𝗲𝗻 𝗥𝗼𝗹𝗲 𝗼𝗳 𝗘𝗻𝘁𝗿𝘆-𝗟𝗲𝘃𝗲𝗹 𝗝𝗼𝗯𝘀 𝗶𝗻 𝗮𝗻 𝗔𝗜 𝗘𝗰𝗼𝗻𝗼𝗺𝘆 AI promises massive productivity gains. But it may also be quietly eroding how expertise is built. As AI enables senior employees to do more on their own, many entry-level roles—the primary source of learning by doing—are disappearing. This matters because the most valuable workplace skills are often 𝘁𝗮𝗰𝗶𝘁: absorbed through experience, not taught in classrooms or manuals. According to a recent study, today’s rush to automate early-career work may be socially excessive. While automation boosts short-term productivity, it also disrupts the intergenerational transfer of tacit knowledge. The result is a trade-off: higher output now, but weaker skills in the next generation—ultimately slowing long-term economic growth and, in some cases, reducing overall welfare. The implications are not trivial. Even modest levels of AI-driven automation at the entry level could lower long-run U.S. per-capita growth by an estimated 𝟬.𝟬𝟱 𝘁𝗼 𝟬.𝟯𝟱 𝗽𝗲𝗿𝗰𝗲𝗻𝘁𝗮𝗴𝗲 𝗽𝗼𝗶𝗻𝘁𝘀 𝗮𝗻𝗻𝘂𝗮𝗹𝗹𝘆. Over time, that compounds into a meaningful economic drag. AI co-pilots offer a partial remedy. They can help workers who missed early learning opportunities catch up later in their careers. But they also introduce a new tension: if AI makes skill gaps easier to mask, it may reduce incentives for juniors to develop those skills in the first place. 𝗧𝗵𝗲 𝗺𝗲𝘀𝘀𝗮𝗴𝗲 𝗶𝘀 𝗹𝗼𝘂𝗱 𝗮𝗻𝗱 𝗰𝗹𝗲𝗮𝗿: 𝗔𝗜 𝗰𝗵𝗮𝗻𝗴𝗲𝘀 𝗻𝗼𝘁 𝗷𝘂𝘀𝘁 𝗵𝗼𝘄 𝘄𝗼𝗿𝗸 𝗶𝘀 𝗱𝗼𝗻𝗲, 𝗯𝘂𝘁 𝗵𝗼𝘄 𝗲𝘅𝗽𝗲𝗿𝘁𝗶𝘀𝗲 𝗶𝘀 𝗳𝗼𝗿𝗺𝗲𝗱. 𝗔𝗻𝗱 𝗴𝗿𝗼𝘄𝘁𝗵 𝗱𝗲𝗽𝗲𝗻𝗱𝘀 𝗼𝗻 𝗯𝗼𝘁𝗵. To capture AI’s full potential, policy, firms, and universities must protect and expand early-career learning—through mentorships, apprenticeships, practical education, and AI systems that complement junior roles rather than erase them. 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝘄𝗶𝘁𝗵𝗼𝘂𝘁 𝘀𝗸𝗶𝗹𝗹 𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗶𝘀 𝗻𝗼𝘁 𝗽𝗿𝗼𝗴𝗿𝗲𝘀𝘀—𝗶𝘁’𝘀 𝗯𝗼𝗿𝗿𝗼𝘄𝗲𝗱 𝗴𝗿𝗼𝘄𝘁𝗵. 𝗥𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲: Ide, Enrique. (2025). Automation, AI, and the Intergenerational Transmission of Knowledge. 10.48550/arXiv.2507.16078.

The Fed’s Michael Barr examines the productivity, labour market, and inflation effects thus far of AI, and lays out three scenarios for how the development of AI could impact the labour market: 1) Gradual adoption – “Unemployment might rise somewhat in the short term due to skill mismatch, but education and training choices adjust over time, and many workers successfully retrain and retain their jobs or find new ones. With strong productivity growth, the economy can sustain faster output growth and real wages rise.” 2) Rapid growth in AI capabilities and adoption – “… Ushering in a jobless boom… With a vastly more productive economy, but much less demand for labor, society would have to rethink the social safety net to ensure that the gains from unprecedented economic growth are shared rather than concentrated among a small group of capital holders and AI superstars.” 3) Stalled growth in AI capabilities and adoption – “The hard work of business process transformation takes time, which partly accounts for the J curve dynamics I mentioned earlier. Businesses that do not see immediate productivity improvements may lose interest. In a scenario of stalled growth in AI capabilities and adoption, some productivity improvements occur in easy-to-learn tasks, but AI proves incapable of completing hard-to-learn tasks or complex projects, or an AI bust occurs, abruptly ending needed investment.” Also of interest, Barr does not view AI as a reason to lower interest rates, even given its likely productivity enhancing qualities. He notes: “…demand for capital would rise because of the strong business investment required to take advantage of the technology, putting upward pressures on interest rates, and household savings could fall due to expectations of stronger real wage growth and thus higher lifetime earnings, also putting upward pressure on interest rates. All of this would imply a higher setting for the policy rate when the economy is at equilibrium, or what monetary economists call r*. Indeed, last year I raised my long-term estimate of r* modestly because of higher productivity. Moreover, in the short term, investment in AI could be inflationary—for example, if electricity supply constraints from inefficiencies in the power grid collide with strong energy demand from the building of data centers.”

AI will raise productivity. But will it sustain participation? In the recent parliamentary debates in Singapore, a quiet but consequential theme emerged: how to ensure that AI-driven growth does not become jobless growth. Singapore has navigated structural transitions before. From manufacturing to services. From labour intensity to capital depth. In each case, disruption preceded absorption. Over time, new industries formed, wages adjusted and the labour market re-equilibrated. The AI cycle may be different. Not because technology replacing labour is new, but because of the speed and cognitive scope of substitution. Productivity can now scale without proportional increases in headcount. That is economically efficient. But socially, it introduces a tension between output expansion and broad-based income participation. This is where the question shifts from “Will GDP grow?” to “Who grows with it?” For policymakers, the issue is calibration. Reskilling systems, modular training, wage supplements and employment support schemes aim to ensure workers are not structurally displaced. The challenge is velocity. Technological diffusion may outpace institutional adaptation. For households, the issue is more immediate. Families make long-term commitments on the assumption of income continuity: housing mortgages, education pathways, eldercare responsibilities. If AI introduces episodic employment instability or skill obsolescence risk, then workforce policy becomes balance sheet policy. Do we need faster training cycles? Should firms think more carefully about redeployment pathways? Are we equipping mid-career professionals with sufficient transition optionality? The AI conversation is often framed as innovation versus regulation. The more durable framing may be productivity versus participation. Singapore’s advantage lies in coordination capacity and fiscal space. But even strong institutions must adapt quickly when the production frontier shifts. The test of this AI cycle will not be whether we build cutting-edge models. It will be whether productivity gains translate into broad income resilience rather than narrow capital concentration. Growth has always mattered. Participation may matter more.

This piece is a good overview of the structural issues related to automation and labor. "Rather than inducing mass unemployment, the more immediate effects of generative AI are likely to mirror broader trends of job transformation already unfolding today, namely de-skilling and surveillance. Preliminary studies suggest that generative AI technologies raise productivity most among lower-skilled workers, helping to standardise outputs but doing little to enhance high-skill, high-complexity work. It is no coincidence that these systems excel at generating average-quality writing and basic code — the kinds of tasks that students perform, which is why one of the main use cases for ChatGPT has been helping students cheat. As such tools become more widespread, there is a risk of a digital de-skilling of fields such as computer programming, graphic design, and legal research, where algorithmically generated outputs could substitute for outputs produced by workers with average levels of competence.   At the same time, generative AI models offer new possibilities for monitoring and evaluating workers, processing surveillance data to exert greater control over labour processes and suppress wages. Once again, the technologies that promise to liberate us from work risk intensifying exploitation instead. Without robust social and legal frameworks to redirect their development, the likely outcome of the generative AI boom will not be mass joblessness, but a worsening of work conditions, an acceleration of economic inequality, and a further erosion of workers’ autonomy." https://lnkd.in/gyhwkZtD

Automation debates often assume that machines simply replace workers task by task. We need new models to understand how AI reshapes the equilibrium set of tasks. In a new National Bureau of Economic Research working paper, Joshua Gans and Avi Goldfarb show why some of the greatest gains from AI are NOT just from replacing worker tasks. The authors model jobs as systems of tightly linked tasks where quality is complementary rather than additive. When automation takes over some tasks, workers do not just lose work. They reallocate time to what remains. That reallocation can raise the quality and value of the remaining tasks, creating what the authors call a “focus” mechanism. Three implications: 1) Automation decisions are often bundled, not marginal. Because tasks interact, it can be unprofitable to automate a single task in isolation even when automating several at once makes economic sense. 2) Standard exposure indices that add up task level automation risk systematically overstate displacement when tasks are complements. In these settings, the bottleneck task matters more than the average task. 3) Under partial automation, labor income can rise rather than fall. By scaling the value of the remaining bottleneck tasks, automation can increase the surplus workers bargain over, at least until full automation becomes optimal. The paper reframes familiar cases like bank tellers and ATMs. Instead of a simple displacement story, automation removed routine components and shifted human effort toward higher value interactions. The same logic applies today in settings from legal services to radiology, where AI automates some steps but intensifies the importance of judgment, integration, and communication. If we treat jobs as collections of independent tasks, we will misread both the pace and the distributional effects of automation. What matters is how technology reshapes bottlenecks and reallocates human attention across the production process. Excited about new measurement opportunities to quantify how AI is reshaping the equilibrium set of tasks and worker behavior. #Automation #FutureOfWork #LaborEconomics #AIandJobs

The Atlantic argues that AI automation isn’t making it harder for young people to find jobs — economic uncertainty is. My take is that this distinction doesn’t hold up, for two reasons. (1) First, the scale of AI investment implies expectations of labor automation, whether or not we can measure it yet. Companies are planning to spend roughly $635 billion on AI infrastructure this year alone, and McKinsey projects more than $5 trillion in AI data-center investment by 2030. That is an extraordinary amount of capital deployment. Investments at this scale require enormous recurring returns. Large technology firms typically target around a 10–12% annual return on capital, especially for infrastructure that depreciates quickly — and AI hardware depreciates in roughly five years. A single year of $635B in spending therefore implies roughly $75B in annual economic value just to clear the cost of capital. As investment compounds over several years, the math scales quickly: a few trillion dollars of deployed infrastructure would require something like $250–400B in annual returns to justify itself. The question isn’t whether AI is useful. The question is where that level of economic value could plausibly come from. One possibility is selling more software or digital services. But global enterprise software spending is only about $1 trillion annually. For AI infrastructure alone to justify itself through expanded software demand, that market would need to grow by roughly 25–40% in short order — and we are not currently seeing evidence of growth on that scale. Another possibility is entirely new markets. Historically, however, new technology markets take decades to mature, while AI infrastructure must earn returns much sooner due to rapid depreciation. That leaves cost replacement. The largest recurring economic expenditure in the world is labor compensation, totaling $55 trillion annually. Even small productivity gains that reduce labor costs — on the order of one or two percent globally — would be good ROI. (2) Second, we can't blame economic hardship on uncertainty without interrogating where that uncertainty comes from. The article concludes that the labor market is complicated and unpredictable, and that uncertainty rather than automation may explain weak hiring among young workers. That may be true! But I'm not sure it matters whether AI is the root cause or the proximate cause -- it's still the cause, and it's only going in one direction. I’m not an economist — and economists like Guy Berger, Ph.D., who is quoted in the article, probably disagree with me. But capital allocation is a form of forecasting and when companies commit trillions of dollars to infrastructure designed to automate labor, they are revealing what outcomes they expect. Labor impact is baked into the investment itself. https://lnkd.in/gq-ybqKQ

If you’ve ever wondered why the economy feels like it’s tightening even while productivity is “going up”… this is why: We’ve forgotten what Henry Ford knew instinctively. He designed work on purpose, not just to make factories run smoother, but to ensure his workers could afford the products they made. He built a labour–consumption loop that created the middle class and powered a century of growth. And today? We’re tearing that loop apart at machine speed. AI is accelerating productivity. But it’s also hollowing out incomes. And without income, there is no demand. Without demand, there is no economy. This is the conversation almost no AI leader is willing to have. So I’ve written about it. 👉 “What Henry Ford Understood About Work (And What Today’s AI Leaders Don’t)” Live now and published with Bloor Research International. It covers: • Why Ford’s original logic still matters • How AI is collapsing the labour–consumption loop • Why ESG is missing “Workforce Sustainability” • What outcomes-vs-output design actually solves • And the urgent system redesign business now needs AI isn’t the threat. The system we plug it into is. If you care about the future of work, sustainability, HR, economics, AI governance, or the health of the UK labour market, this one is worth a read. (And yes, it’s punchy.) 👇 Read it here: Lancaster University Beth Suttill Sage Journal Mark Thompson Carl Benedikt Frey Daniel Susskind Lord Kulveer Ranger Lord Holmes Lee Barron MP Better Hiring Institute (BHI) Siobhan Cleary carolyn downs Trades Union Congress (The TUC) Ford Motor Company Hung Lee Paul Ellingstad Paul Nowak Microsoft AI Google DeepMind OpenAI HR magazine Charissa King Abodoo #FutureOfWork #AIandWork #Sustainability #EconomicResilience #OutcomeBasedWorking #FusionWork