Moineauworld

Tag: Concentrix

AI-Exposed U.S. Jobs Show Early Decline | Analysis by Brian Moineau

Posted on by Brian Moineau

TL;DR

What the source said

Gizmodo’s read of the 2025 OEWS release highlights a 0.2% year‑over‑year decline across 18 BLS‑flagged “artificial intelligence related occupations” while overall U.S. employment rose 0.8% in the same May‑to‑May window. [1][3] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.bls.gov/news.release/ocwage.nr0.htm)

Customer service representatives absorbed the sharpest hit, dropping about 130,180 positions (−4.8%) in the period, according to the same OEWS tables and coverage. [1][3] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.bls.gov/news.release/ocwage.nr0.htm)

The piece notes that “medical secretaries” bucked the trend with strong gains tied to healthcare expansion, which makes the rest of the 17‑occupation basket look superficially stable despite a two‑year slide. [1][2] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html)

Gizmodo frames the evidence as early but directionally negative for back‑office, sales support, and service roles that feature standardized, screen‑based tasks. [1] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602)

Why it matters

Contact centers, clerical back‑offices, field sales support, and entry‑level legal/admin roles employ millions in the United States; these “interface” layers sit exactly where LLMs and workflow tools can shave 10–60 seconds per task. [1][2] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html)

For CFOs in healthcare, retail, insurance, and logistics, the math is simple: if each claim, chat, or form requires fewer human touches, you freeze reqs and let attrition do the rest. [6] (https://www.bls.gov/opub/mlr/2024/article/industry-and-occupational-employment-projections-overview-and-highlights-2023-33.htm)

Two consecutive years of measured declines in the non‑medical cohort suggest hiring bars are ratcheting up quarter by quarter, not just wobbling with the business cycle. [2][3] (https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html, https://www.bls.gov/news.release/ocwage.nr0.htm)

Original analysis

The signal inside the −0.2% headline

The 0.2% dip across 18 AI‑flagged occupations looks tiny until you unpack composition and compounding. [2][3] (https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html, https://www.bls.gov/news.release/ocwage.nr0.htm)

Back‑of‑envelope calculation:

The distribution matters: AI trims high‑volume, routinized, digitized job architectures first, rather than “all jobs” equally. [1][3] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.bls.gov/news.release/ocwage.nr0.htm)

A typology of American jobs with AI exposure

Use a two‑by‑two: Task Structure (Routinized ↔ Non‑routinized) vs. Human Stakes (Low‑stakes ↔ High‑stakes).

This mapping explains why the −1.6% slide clusters in interface roles while medical secretaries—buoyed by healthcare demand and compliance friction—buck the trend. [2][6] (https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html, https://www.bls.gov/opub/mlr/2024/article/industry-and-occupational-employment-projections-overview-and-highlights-2023-33.htm)

A historical analogue—with a twist

Electrification’s productivity payoff lagged factory re‑architecture by roughly 15–25 years, with major gains arriving circa 1915–1930 after plants reorganized around motors. [7] (https://en.wikipedia.org/wiki/Productivity_paradox)

Services in 2026 are already digitized “plants,” so AI can reduce headcount before broad productivity shows up in aggregates, flipping the old lag pattern. [5] (https://www.federalreserve.gov/econres/notes/feds-notes/ai-adoption-and-firms-job-posting-behavior-20260327.html)

Fed researchers, using Lightcast and BTOS data, find no aggregate posting decline so far at AI‑adopting firms, even as early‑career seats thin in high‑exposure roles tracked by the Dallas Fed. [5][4] (https://www.federalreserve.gov/econres/notes/feds-notes/ai-adoption-and-firms-job-posting-behavior-20260327.html, https://www.dallasfed.org/research/economics/2026/0106)

Contrarian read: The “no big deal” benchmark misses the pipeline

You will hear, “Relax—AI hasn’t dented total job postings or overall employment,” and the Fed’s 2026 note indeed shows AI‑related postings at just 1.6% of all firm postings so far. [5] (https://www.federalreserve.gov/econres/notes/feds-notes/ai-adoption-and-firms-job-posting-behavior-20260327.html)

The better benchmark is youth inflow: the Dallas Fed reports a 13% employment drop since 2022 for 22–25 year‑olds in the most AI‑exposed occupations, driven by weaker inflows rather than elevated layoffs. [4] (https://www.dallasfed.org/research/economics/2026/0106)

A single routing layer can remove the business case for hiring level‑1 reps even when aggregate U.S. unemployment looks steady. [1][5] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.federalreserve.gov/econres/notes/feds-notes/ai-adoption-and-firms-job-posting-behavior-20260327.html)

Named‑stakeholder breakdown

  • Teleperformance SE, Concentrix Corp., and TTEC Holdings: Expect margin tailwinds as automated deflection trims agents per client while interaction volumes grow; revenue tilts toward AI tooling and integration services.
  • Salesforce, Zendesk, and NICE: Service clouds turn into automation platforms, shifting pricing from agent seats to interactions and model assists as 2026–2027 product roadmaps harden.
  • Insurers and banks: Claims and credit clerks face hiring freezes, wider spans of control, and flatter promotion ladders as throughput targets rise quarter by quarter.
  • Community colleges and bootcamps: Placement rates for admin, support, and help‑desk tracks risk declines unless 2026–2027 curricula add “AI ops,” audit trails, and exception‑handling labs tied to regulated workflows.
  • State workforce boards in Texas, California, and Ohio: Bridge programs must reroute interface workers into regulated, field‑present roles—health support, inspection, and operations—where AI augments rather than replaces.

The story shows up not in mass layoffs but in non‑backfills and frozen requisitions, exactly what the OEWS and coverage already imply. [1][2][3] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html, https://www.bls.gov/news.release/ocwage.nr0.htm)

What others are missing

The crucial angle is hiring inflows versus layoffs: Dallas Fed decomposition attributes the 2022–2025 deterioration for 22–25 year‑olds in high‑exposure jobs mainly to weaker inflows from “not in the labor force” into these roles, not to elevated separations. [4] (https://www.dallasfed.org/research/economics/2026/0106)

This matters because inflow declines rarely appear in corporate announcements or WARN data, yet they quietly erase local entry‑level ladders months before unemployment rates budge. [4] (https://www.dallasfed.org/research/economics/2026/0106)

Automation in professionalized service work typically starts by closing junior reqs and absorbing marginal load with software, which aligns with the 1.6% multi‑year slide outside medical secretaries. [2] (https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html)

What to watch next

  1. By Q4 2026, OEWS will show a third straight −1% to −2% year‑over‑year decline across the 17 non‑medical AI‑related occupations combined, confirming multi‑year substitution beyond a one‑off blip. [3] (https://www.bls.gov/news.release/ocwage.nr0.htm)
  2. By Q1 2027, earnings from at least two major BPOs—Teleperformance SE (EPA: TEP) and Concentrix Corp. (NASDAQ: CNXC)—will explicitly attribute gross‑margin expansion to AI call deflection/assist while reporting flat or declining average agents per client despite higher interaction counts.
  3. By June 2027, Federal Reserve/Lightcast analyses will still show neutral aggregate postings, but a higher share of service‑org postings will specify “AI‑assisted workflows,” alongside further drops in early‑career employment for 20–24 year‑olds in high‑exposure occupations. [5][4] (https://www.federalreserve.gov/econres/notes/feds-notes/ai-adoption-and-firms-job-posting-behavior-20260327.html, https://www.dallasfed.org/research/economics/2026/0106)

My take

The market has already repriced “interface jobs,” and the repeated −1.6% declines outside a single booming healthcare subcategory confirm where substitution lands first. [2] (https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html)

We are swapping people for software exactly where workflows are standardized and stakes are low, which will not crash headline U.S. employment but will starve tomorrow’s talent bench in 2027–2029. [1][3] (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602, https://www.bls.gov/news.release/ocwage.nr0.htm)

Service leaders should create new entry‑level roles—AI ops, compliance QA, and exception handling—with measurable skill ladders, or accept brittle orgs that fail the first time a model hallucinates on a regulated workflow. [6] (https://www.bls.gov/opub/mlr/2024/article/industry-and-occupational-employment-projections-overview-and-highlights-2023-33.htm)

Sources

[1] Gizmodo — American Jobs with AI Exposure Really Are Starting to Disappear, Data Show (https://gizmodo.com/american-jobs-with-ai-exposure-really-are-starting-to-disappear-data-show-2000759602) — News peg and core figures: −0.2% for AI‑flagged roles vs. +0.8% overall; −4.8% (~130,180) for customer service reps.

[2] Business Standard (summarizing Bloomberg) — US starting to witness heavy job losses in occupations exposed to AI (https://www.business-standard.com/world-news/us-starting-to-witness-heavy-job-losses-in-occupations-exposed-to-ai-126051600082_1.html) — Corroborates 18 BLS‑flagged occupations (~10M jobs), the −0.2% YoY dip, and the repeated −1.6% decline for the other 17 roles.

[3] U.S. Bureau of Labor Statistics — Occupational Employment and Wages, May 2025 (https://www.bls.gov/news.release/ocwage.nr0.htm) — Primary OEWS release underpinning the year‑over‑year comparisons.

[4] Federal Reserve Bank of Dallas — Young workers’ employment drops in occupations with high AI exposure (https://www.dallasfed.org/research/economics/2026/0106) — Shows a 13% decline since 2022 for 22–25 year‑olds in high‑exposure jobs and attributes it to weaker inflows.

[5] Board of Governors of the Federal Reserve System — AI Adoption and Firms’ Job‑Posting Behavior (https://www.federalreserve.gov/econres/notes/feds-notes/ai-adoption-and-firms-job-posting-behavior-20260327.html) — Finds no aggregate posting declines tied to AI adoption and notes AI‑related postings at 1.6%.

[6] Monthly Labor Review, U.S. Bureau of Labor Statistics — Industry and occupational employment projections overview and highlights, 2023–33 (https://www.bls.gov/opub/mlr/2024/article/industry-and-occupational-employment-projections-overview-and-highlights-2023-33.htm) — Context on sectoral demand, compliance friction, and where augmentation is more likely than substitution.

[7] Wikipedia — Productivity paradox (https://en.wikipedia.org/wiki/Productivity_paradox) — Background on historical lags, including the 1915–1930 period when reorganized factories converted electrification into measurable productivity.



Related update: We recently published an article that expands on this topic: read the latest post.