AI and Knowledge Work: Job Augmentation, Creation and Displacement

Overview

Three measures are now in hand for how far AI has reached into knowledge work, and they do not all point the same way. Announced layoffs attributed to AI led every other reason for five months of 2026 and still lead the year at 120,136 cuts through September, about one in five of all announced cuts, but the monthly share fell below 10% in August and September.[1] Realized payrolls show a narrower effect: the BLS Information supersector is down 376,000 jobs (12%) from its November 2022 peak while total employment has grown 6% since January 2022, and the two other knowledge-work supersectors are roughly flat.[2] The sharpest signal is in hiring rather than firing: Stanford’s payroll-based “Canaries” series puts employment of 22-to-25-year-olds in the most AI-exposed occupations 19% below where it would be had it kept pace with their less-exposed peers, a gap that has widened every release since August 2025.[3]

The models themselves are arriving faster than the benchmarks can score them. Between mid-July and early October 2026 the four U.S. frontier labs shipped at least nine models, and the top score on Humanity’s Last Exam reached 61%. METR retired its Time Horizon leaderboard in September, and GPQA Diamond saturated in the third quarter, so release cadence and HLE are the two public yardsticks this section carries into 2027.[4]

The money behind the models is the same money that shows up in the Data Centers section. The four hyperscalers spent $411 billion on capital expenditure in 2025 and have guided to $720–745 billion for 2026; Challenger’s reading of the layoff data is that “regardless of whether individual jobs are being replaced by AI, the money for those roles is.” The capex that crowds out personnel budgets in this section is the construction pipeline measured on the Data Center Capex page.[5]

This section focuses on how AI has influenced the white collar (knowledge work) labor market so far and introduces some useful tracking metrics to help us identify and contextualize important milestones as they occur in the coming years.

The pace and scale of AI-related job displacement, creation, and productivity enhancement will play out over the course of the next decade (at least).

Predictions range from little effect to massive job replacement, and will likely fall somewhere in the middle as technology evolves and companies adapt over time.

1. Measuring AI Model Progress

2. AI Workplace Adoption & Job Displacement / Creation

3. AI & CRE Implications for Office, Life Science and Multifamily

[1] Challenger, Gray & Christmas, September 2026 job-cut report (Oct 1, 2026): “Artificial Intelligence was the fifth-most cited reason with 3,961 cuts in September, about 9% of the month’s total. So far this year, AI has been cited in 120,136 job cut announcements, approximately 21% of all cuts, and it remains the leading reason year-to-date.” AI was the leading reason in each month March through July 2026 (monthly releases, Table 4). challengergray.com ↩

[2] U.S. Bureau of Labor Statistics, Current Employment Statistics, seasonally adjusted, September 2026 Employment Situation (Oct 2, 2026), Table B-1: Information 3,115,000 (Nov 2022) to 2,739,000 (Sep 2026); total nonfarm 150,006,000 (Jan 2022) to 159,044,000 (Sep 2026, +6.0%); Financial Activities +1.7% and Professional and Business Services +1.3% over the same span. bls.gov ↩

[3] Brynjolfsson, Chandar and Chen, “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence,” Stanford Digital Economy Lab, revised August 12, 2026 (ADP payroll data through June 2026); monthly series on the lab’s AI Economic Indicators dashboard (vintage Sept 17, 2026, data through August 2026). Earlier vintages reported 13% (July 2025 data) and 16% (September 2025 data). digitaleconomy.stanford.edu ↩

[4] Release dates from the labs’ own announcements: xAI Grok 4.5 (Jul 16), 4.6 (Aug 12), 4.7 (Sep 21); OpenAI GPT-6 Astra (Sep 4), GPT-6 Sol and Luna (Sep 22); Anthropic Claude Opus 5.5 (Sep 22), Sonnet 5.5 (Sep 28), Haiku 5.5 (Oct 7, 2026). HLE: Artificial Analysis leaderboard, Claude Opus 5.5 at 61.4% (read Oct 14, 2026). METR Time Horizons page marked “no longer actively updated” (Sept 8, 2026). Detail and sources on the Measuring AI Model Progress page. ↩

[5] Alphabet, Amazon, Meta and Microsoft 10-K filings and Q2 2026 earnings calls (July 2026), as compiled on the CRE42 Data Center Capex page: $411 billion in 2025, guidance of $720–745 billion for 2026. Andy Challenger quoted in the Challenger, Gray & Christmas April 2026 release (May 7, 2026). ↩

Sources

[1] Challenger, Gray & Christmas, monthly job-cut reports (2026). challengergray.com

[2] U.S. Bureau of Labor Statistics, Employment Situation, September 2026. bls.gov

[3] Stanford Digital Economy Lab, AI Economic Indicators and “Canaries in the Coal Mine?” (Aug 2026). digitaleconomy.stanford.edu

[4] Artificial Analysis, HLE leaderboard; METR, Time Horizons; OpenAI, Anthropic and xAI news indexes. artificialanalysis.ai

[5] Hyperscaler 10-K filings and Q2 2026 earnings calls via CRE42, Data Center Capex vs. Historical Capex Booms.