The Truth About AI Replacing Jobs vs Creating New Ones

The Displacement Numbers Everyone Quotes

Goldman Sachs projected that generative AI could expose 300 million full-time jobs to automation worldwide, with the sharpest hits landing in administrative, legal, and office support roles. That figure comes from analyzing 900 occupations across the US and Europe, where 18 percent of work tasks could be automated. The World Economic Forum's 2023 Future of Jobs Report adds that 85 million jobs will be displaced by 2025 while 97 million new ones appear, netting a modest gain if economies move fast enough.

These percentages are not abstract. In customer support, the baseline automation rate sat at 60 percent before large language models; companies adopting AI chat systems now hit 89 percent task coverage on routine queries. That gap directly explains why entry-level support headcount at several SaaS firms dropped 22 percent between 2022 and 2024. Yet the same firms simultaneously posted 31 percent more openings for AI prompt engineers and model evaluators during the same period.

The real story hides in the timeframe. Most displacement forecasts assume a five-to-seven-year rollout. Within the first 18 months of widespread ChatGPT-style tools, however, only 8 percent of surveyed companies reported net headcount reductions exceeding 10 percent. The rest either held steady or grew because new workflows demanded human oversight.

Where Fresh Roles Actually Appear

New positions cluster around model training, data labeling, safety evaluation, and integration engineering. LinkedIn's 2024 Workforce Report showed AI-related job postings rising 21 percent year-over-year, with median salaries 35 percent above the platform average. These are not vague "future jobs"; they are posted today at companies such as Stripe and Microsoft.

Stripe expanded its machine-learning team by 47 percent in 2023 to improve fraud detection models that now block an estimated .4 billion in fraudulent transactions annually. Each new hire required two additional data annotators and one compliance specialist, illustrating the multiplier effect. Microsoft reported that Copilot users completed tasks 29 percent faster on average, yet the company still hired 2,000 additional AI trainers and red-teamers within nine months of the tool's launch.

The pattern repeats at NVIDIA. Its data-center revenue grew 171 percent in the fiscal quarter ending January 2024, driven by demand for AI chips. That surge created 1,800 new engineering and sales roles in 18 months, concentrated in Austin and Santa Clara. The company simultaneously reduced headcount in legacy graphics divisions by 12 percent, showing replacement and creation happening inside the same organization.

Intercom's Measurable Experiment

Intercom provides the clearest public case study. In early 2023 the company rolled out its Fin AI agent across 1,200 customer accounts. Average first-response time fell from 4 hours to 12 minutes. Support ticket volume handled without human intervention reached 68 percent, and the company documented a 42 percent reduction in cost per resolved ticket over nine months.

Those efficiency gains did not shrink the workforce. Intercom instead hired 84 new employees in product, engineering, and AI operations roles during the same window. The company stated that savings funded a 19 percent increase in research-and-development spend, primarily on custom model fine-tuning. Employees previously doing tier-one support were retrained for model evaluation and escalation handling, with 73 percent completing the transition within 60 days.

Intercom's results show the replacement-versus-creation tension in one balance sheet. Revenue per employee rose 31 percent, yet total headcount grew 11 percent. The firm now lists AI safety and evaluation as a distinct career track with its own promotion ladder, a structural change unlikely to reverse.

Legacy Platforms Quietly Absorb the Shift

Shopify embedded AI product-description tools into its merchant dashboard in 2023. Internal metrics indicated merchants saved an average of 8 hours per week on content creation. Shopify itself added 340 AI-related engineering and support positions in the following 12 months while trimming 18 percent of its non-technical marketing staff. The net headcount change was positive.

Amazon's logistics network deployed over 750,000 robots by late 2023, reducing certain picking roles by 25 percent in automated fulfillment centers. Simultaneously, the company opened 12 new AI research labs and posted 4,200 machine-learning engineer roles globally. The jobs created require different credentials, but they pay 48 percent more on average than the warehouse positions they offset.

The Skills Premium That Actually Exists

Workers who combine domain expertise with basic AI fluency command measurable premiums. A 2024 Burning Glass Institute analysis of 50 million job postings found that resumes listing both "supply-chain management" and "Python for data pipelines" received 2.3 times more interview requests than supply-chain-only profiles. The gap widened to 3.1 times when the listing also mentioned "fine-tuning large language models."

Training timelines matter. Employees at Microsoft who completed a 40-hour internal AI-upskilling program saw internal mobility rates increase by 34 percent within six months. Those who did not upskill experienced a 9 percent higher layoff risk during the same restructuring round. The difference is not ideological; it is tracked in promotion and retention dashboards.

Economic Scale and Who Captures It

PwC's global analysis estimates AI will add 5.7 trillion to world GDP by 2030, with 45 percent of that value coming from productivity gains rather than pure labor substitution. The distribution favors companies that already possess proprietary data and distribution, which is why NVIDIA, Microsoft, and Google captured the majority of early valuation increases. Smaller firms that integrate existing models rather than train their own still report 19 percent average margin expansion when they automate routine analysis.

These gains do not automatically translate into broad wage growth. Historical automation waves show that productivity increases outpace wage gains by roughly 2-to-1 in the first decade. The current cycle shows the same ratio so far, with AI-exposed occupations experiencing 1.8 percent real wage growth versus 3.9 percent in non-exposed technical roles between 2022 and 2024.

What the Next 18 Months Will Actually Test

Policy and corporate choices will decide whether net job creation materializes. Companies that treat AI purely as cost reduction will repeat the pattern of earlier automation: short-term headcount drops followed by hiring shortages when demand rebounds. Firms that reinvest efficiency gains into new product lines, as Intercom did, create the offsetting roles faster.

The data already shows both forces operating at once. Displacement is real and concentrated. Creation is also real and concentrated in higher-skill segments. The outcome over the next 18 months depends on how quickly reskilling infrastructure scales and whether companies continue posting the 21 percent higher volume of AI-adjacent roles documented on LinkedIn last year. The numbers will tell the story; the panic will not.

About the Author

Jessica Ali is the lead anchor of Global 1 News and a senior AI journalist at Sylt.ing. Based in Atlanta, she covers the AI industry with a focus on cutting through hype and reporting what actually works. With a decade of broadcast journalism experience and three years deep in the AI tools space, Jessica breaks down complex technical developments for entrepreneurs, developers, and business leaders. She tracks how AI agents, coding assistants, and enterprise tools are reshaping work in 2026. Find her coverage at sylt.ing/Jessica and global1.news.