AI: Job Creator, Destroyer, or Something in Between?
- Ed Sullivan
- 4 hours ago
- 8 min read
Breaking News: Market Update
Identifying the near-term impacts of artificial intelligence (AI) on employment and the economy is a critical facet of any market forecast. Depending on the assessment, it could impact unemployment, inflation, economic growth, monetary policy, and overall interest rates. No construction forecaster has the luxury of avoiding estimates of its impact. This article outlines potential AI impacts that underlie the basis of The Sullivan Report’s (TSR) forecasts.
The Debate: AI as a Job Destroyer…or Creator?
The debate over artificial intelligence and jobs is often framed in extremes. One view warns that AI will cause massive unemployment, rapidly replacing workers across the economy. The other argues that AI is no different from past technological advances - like the automobile replacing horse-drawn carriages - and that technology has always created more jobs than it destroys.
Both views miss the most important point: timing. AI does not eliminate or create jobs all at once. In its early stages, AI primarily slows hiring by allowing firms to produce more with fewer incremental workers. Entry-level and routine roles are affected first, restraining job growth, and modestly raising unemployment without triggering widespread layoffs. As adoption deepens, some job displacement becomes unavoidable, particularly in clerical and basic knowledge roles.
Only after AI reaches scale - enabling new business models, industries, and demand - does meaningful job creation emerge. Historically, technology creates jobs, but only after a period of disruption and adjustment. Throughout the remainder of this decade, AI is best understood as a near-term drag on employment growth and a longer-term source of productivity-driven job creation. This could deliver important implications for wages, demand, and economic policy – all of which impact the level of construction activity.
Each stage suggests different implications for the economy. During the first stage, unemployment increases modestly. High and sustained productivity gains suggest a further bifurcation of consumers, or the buying power of high-income earners versus low-income earners. Productivity gains and softer labor market imply added factors pushing down inflation. If AI increases unemployment modestly and acts as a force to ease inflation, a more aggressive monetary policy could materialize. In addition, because inflation expectations are lower, all interest rates, on both short and long-term loans, will drift lower. This could have a positive material impact on private construction activity.
To this end, AI’s labor-market impact is best understood as unfolding in distinct stages. Stage I is characterized by hiring friction and is marked by slower hiring - particularly for new labor market entrants. Stage II is characterized by measured displacement as firms restructure their workforce around AI-enabled productivity gains. In this stage job losses become more visible. Eventually, in Stage III, AI-enabled job creation emerges as AI adoption reaches scale and new business models mature.
Stage I: Hiring Friction Has Already Begun
The first labor-market impact of AI is not mass layoffs. It is hiring friction. When firms adopt new technologies that improve productivity, they rarely begin by firing large portions of their workforce. Instead, they reassess future hiring needs. Entry-level roles are delayed, junior positions are consolidated, and replacement hiring becomes more selective. This effect is difficult to observe in headline employment data, but it appears clearly in hiring rates, job postings, and anecdotal reports from recruiters and universities.
Recent U.S. labor market data may signal that Stage I has already begun. Evidence is accumulating that job market momentum is slowing. The Job Openings and Labor Turnover Survey (JOLTS) show a sustained decline in job openings from post-pandemic peaks. Quit rates have normalized. Hiring rates are easing. At the same time, anecdotal and survey-based evidence increasingly suggests that new college graduates are facing a more difficult labor market than would be expected given the aggregate data.
Payroll evidence shows that in occupations most exposed to AI, employment among 22–25-year-olds fell about 6% from late 2022 to mid-to-late 2025, while employment for workers aged 30 and over increased. This pattern points to restrained entry-level hiring rather than cuts to experienced staff. College graduates appear to be on the front line, as many entry-level roles are exactly where AI is delivering the fastest productivity gains.
AI contributes to a slowdown in job creation before it produces visible job losses. By reducing the need for incremental hiring, AI dampens monthly payroll gains even as firms retain existing workers and prompts unemployment to increase 10 to 20 basis points (BP). This increase in unemployment is not from layoffs, but from longer job searches and weaker absorption of new entrants. AI adds to other factors such as cyclical economic conditions, demographics, and policy actions. These are expected to push monthly job creation down from an average 160,000 net new jobs monthly in 2025 to less than 50,000 net new jobs monthly this year.
While the start of this stage is hard to peg, the macroeconomic AI impacts began to clearly materialize in 2023-2024. During this stage, impacts on inflation expectations and employment levels impacts are visible - but not large enough to significantly change monetary policy or interest rates. This stage likely persists for another 18–24 months before clearly transitioning into the next phase of AI-driven labor adjustment.
Stage Two: Measured Job Displacement
Eventually, AI adoption will move beyond the initial stage and prompt companies to redesign workflows around AI rather than merely layering AI on top of existing processes. At this point, job losses become visible—not as a sudden shock, but as a steady erosion of certain occupations. Unlike cyclical layoffs, these losses are structural. They reflect permanent reductions in labor demand for specific tasks.
It will take time for these processes to unfold. Budgeting and organizational redesigns must occur before headcount reductions materialize. In the meantime, companies are likely to rely on retirements, attrition and hiring freezes. In addition, firms need time to build confidence in AI systems, resolve legal and regulatory uncertainties, retrain existing staff, and reengineer processes.
Those processes and assessments are already underway and may overlap stage one. The AI impact on jobs is expected to be a “white collar” phenomenon. More specifically, it is likely to impact routine office work first - such as basic coding, data processing, clerical and administrative roles, customer support and call center functions. It can expand as well into certain types of content production and marketing, paralegal, compliance, and entry-level legal research as well as back-office finance and accounting tasks.
Initially, it does not seem to threaten white collar workers who are not engaged in these areas. The daily routine of every white-collar job, to a greater or lesser effect, will likely be impacted by AI. “Blue Collar” jobs will also be impacted, but probably later and to a smaller degree than white collar workers.
The forecast baseline expects job growth to be reduced by 0.25- 0.50 percentage points per year between 2026 and 2030, concentrated in white-collar and early-career roles. This does not imply net job losses economy-wide, but it does imply slower labor-force absorption, especially for new entrants. In total, AI could reduce job creation by at least 300,000 workers per year and add at least 25 basis points to the unemployment rate. Over several years, this can translate into millions of “missing jobs” relative to a no-AI baseline.
If this stage unfolds as expected, workforce participation may soften, particularly among older workers. Since job creation from AI remains limited at this point in time, and AI is still primarily being used to replace labor, political and policy responses will likely accelerate. This may lead to new regulations and laws which could diminish the adverse impact on job destruction. At the same time, it prolongs the timeline before the job-creation phase takes hold.
During this stage, inflation expectations and employment levels impacts heighten and become significant enough to change monetary policy and interest rates. These changes may usher in more aggressive monetary policy easing - even though AI-driven unemployment is structural rather than cyclical. The Fed typically reacts only to cyclical pressures, not structural ones.
Once displacement becomes visible and politically salient, pressure will build on the Fed to respond through easier monetary policy, treating higher unemployment as labor-market slack rather than parsing its cause. The combination of easier monetary policy and lower inflation expectations results in lower long-term loan rates – to the benefit of construction activity. This stage is assumed to run from early 2027 through 2028. Keep in mind a strong construction recovery.
Stage Three: AI-Enabled Job Creation
Historically, technology creates jobs, but only after a period of disruption and adjustment. Eventually, AI will create jobs. The third stage, often cited by AI optimists, does not arrive quickly. Historically, general-purpose technologies generate net job growth only after they enable new products, services, and industries that did not previously exist. Electricity did not create jobs simply by replacing steam engines. It did so by enabling factories, appliances, suburbs, and entirely new consumption patterns. The internet followed a similar trajectory.
AI must reach a scale large enough to enable the creation of new business models, industries, and demand. By the close of the forecast horizon, AI is expected to slowly progress to a net job creator. For the entirety of the forecast horizon, this job creation is unlikely to fully offset the initial job displacement.
It is important to note that many AI-enabled jobs require higher skill levels than the jobs being displaced. This creates a mismatch problem, not just a quantity problem. Labor markets may experience simultaneous job openings and joblessness - not because jobs are unavailable, but because skills are misaligned. Training and reskilling programs are likely to materialize internally within companies, by trade and colleges, and by industry trade associations.
Implications for the Construction Industry
Long term, this is a story of adjustment. An adjustment that after some disruption could ultimately prove beneficial to the economy and construction activity. Throughout the forecast horizon, AI acts as a deflationary agent. Heightened productivity lowers production costs and hence prices. Initially, AI will also slow or weaken job growth. This slows the near-term increase in wages and as a result places less demand pressure on product markets – also lowering inflationary pressures.
Over the longer term, AI could contribute to lower interest rates by reducing inflation premiums and the “neutral rate.” Lower inflation translates into lower inflation expectations and reduced inflation premiums on long-term loans. AI will help drive lower Interest rates. Arguably, higher productivity also suggests an even lower “neutral rate.”
The neutral rate is the interest rate that neither stimulates nor restrains the economy, keeping inflation stable and output near its long-run potential. That means the Federal Reserve can ease interest rates more than if AI was not a factor. Lower interest rates favor long-term borrowing for single family and multifamily homes, nonresidential investments, and government spending.
Lower interest rates stimulate construction
During the first and second stages, slower job creation could offset some benefits of lower interest rates. Simply put, unemployed workers don’t buy homes, nor do they consume as much. In turn, this can lead to higher vacancy rates among some nonresidential properties – diminishing the strength of its recovery in 2027 and beyond.
It is important to keep in mind that the AI impacts take place in the context of the greater economy. As interest rates and inflation drift down, partially due to AI, they will stimulate economic growth. Furthermore, the tax benefits from the One Big Beautiful Bill Act and regulatory reforms could add further to economic strength. While AI could slow job growth and eventually materialize in job displacement, other factors going on in the economy could diminish this impact on construction activity.
Oddly, the potential slack in job growth generated by AI could materialize just when overall construction activity is heating up. The construction industry has endured a persistent labor shortage on the jobsite. AI holds the potential of marginally reducing this shortage through concrete mix optimization, estimating, scheduling, design automation, and project management. Perhaps more importantly, if AI slows hiring, particularly for younger and entry-level workers, some could turn toward construction – potentially reducing the labor shortage.
About The Sullivan Report
The Sullivan Report delivers subscription-based economic forecasts and market intelligence for the cement, concrete, construction, and building materials industries. Led by award-winning economist Ed Sullivan, its flagship U.S. Cement & Construction Outlook provides five-year forecasts and expert analysis to support strategic decision-making. Learn more at TheSullivanReport.com or contact info@thesullivanreport.com.