The Real Reason Companies Cling to AI: A CNBC Slip Reveals Worker Fear

David Kelly didn’t set out to expose corporate America’s quiet preference. The chief global strategist at JPMorgan Asset Management appeared on CNBC in early September 2026 to discuss the puzzling state of the labor market. Wages had slowed sharply. Unemployment sat at a healthy 4.1 percent. Yet workers stayed put. They stopped asking for raises.

“They think this is a scary economy,” Kelly said of American employees. “They think they’re threatened by AI and they don’t have the guts, basically, to ask for a wage increase or to demand a wage increase in this economy.”

Host Leslie Picker jumped in. She called it a great point. “Because the machine could potentially do your job cheaper. So you better put your head down and keep doing the work.”

The exchange, captured in a Gizmodo article published September 4, 2026, peeled back the official story. Executives insist AI drives productivity. Studies have yet to prove consistent gains across the board. The candid moment suggested another motive. AI keeps labor costs in check by making workers afraid.

Wage growth hit just 3.1 percent year-over-year in August 2026. Inflation ran at 3.4 percent in July. Real earnings slipped. Union membership covers less than 6 percent of private-sector workers. Employees enjoy few protections if they push back. The combination creates a powerful deterrent.

But this is only part of the picture. Federal Reserve officials and business surveys add layers. They show companies use AI to slow hiring more than to fire existing staff. The technology reshapes roles. It raises the bar for new hires. And it hands leverage to employers in tight labor markets.

Evidence Mounts That AI Curbs Hiring Ambition

Minneapolis Fed President Neel Kashkari told CNBC in January 2026 that artificial intelligence was causing big companies to slow hiring. Many businesses reported “real productivity gains.” Smaller firms saw less impact. “AI is really a big company story,” Kashkari said. He expected continued low hiring and low firing.

That pattern holds in fresh data. The New York Fed’s regional business surveys, detailed in a Liberty Street Economics post from September 1, 2026, found AI adoption rose again over the past year. Yet few firms cut jobs. Only 4 percent of service firms reported layoffs tied to AI. No manufacturers did. About 15 percent of service firms hired fewer workers than they otherwise would have.

Offsets exist. Thirteen percent of service firms added staff specifically to work with AI. Retraining remains more common than dismissal for current employees. The authors — Jaison R. Abel, Richard Deitz, Natalia Emanuel, and Nick Montalbano — noted these results echo prior surveys. AI augments more than it replaces so far.

But expectations point higher. Thirteen percent of service firms using AI now anticipate layoffs in the next six months. Nearly a quarter of future AI users expect to hire fewer people. The effects concentrate among college-educated workers. Entry-level roles appear especially exposed.

Other recent reporting backs this up. A Dallas Fed analysis of job postings, released September 1, 2026, showed openings for AI-exposed tasks fell after ChatGPT launched in late 2022. Surviving firms posted fewer positions in automatable occupations. The drop reached 8 to 9 percent by early 2026 in some measures. Texas postings alone declined an estimated 2.6 percent in 2025 due to automation exposure.

Productivity data tells a mixed tale. NBER working papers and Goldman Sachs research find gains in high-skill services and finance. Yet aggregate employment effects stay small. Goldman Sachs economists calculated AI reduced monthly payroll growth by about 16,000 jobs over the past year. Augmentation created jobs elsewhere. The net drag fell hardest on younger, less-experienced workers.

Shopify’s CEO offered a blunt example in 2025. He told staff to prove why a task couldn’t be handled by AI before requesting more headcount or resources. Microsoft’s Brad Smith noted AI influences who gets hired. Candidates with AI skills stand out. Call centers and customer support saw early headcount reductions.

Surveys of HR leaders paint a similar scene. A CNBC poll of workforce executives found 45 percent expect AI to affect nearly half or more jobs in the next 12 months. Sixty-one percent said AI already made their companies more efficient. Yet many firms report no direct layoffs. They simply open fewer positions.

The fear factor Kelly described fits neatly here. Low unemployment normally emboldens workers. People quit bad jobs. They demand better pay. Not this time. Workers see AI as a ready substitute. They keep their heads down. Employers save on labor without mass firings.

And the savings matter. Seventy-five percent of organizations that replaced workers with AI later said it cost more than expected, according to industry data cited in August 2026 reports. Hidden expenses — computing power, maintenance, human oversight, error correction — eroded gains. Knowledge loss hurt performance. Some roles still need people.

Yet boards and executives keep investing. PwC’s 2026 AI Jobs Barometer showed companies most exposed to AI posted 52 percent headcount growth from 2018 levels versus 36 percent for the least exposed. Productivity climbed faster for AI leaders. Wage premiums for AI skills hit 62 percent. The technology sorts winners from laggards.

Tech CEOs have walked back dire predictions. Sam Altman, who once warned of major job losses, now says the industry underestimated how central people would remain. Dario Amodei of Anthropic softened his tone on entry-level roles vanishing. The narrative shifted from replacement to augmentation. But the CNBC moment suggests the old logic lingers in practice.

Workers feel the pressure. Mercer’s 2026 survey found only 44 percent of employees “thriving,” down from 66 percent in 2024. Job-loss anxiety, thinner staffing, and uneven AI training contributed. More than a third would consider quitting if left behind on AI tools.

Economists debate the long view. Some models forecast $2 trillion in added U.S. output over time. Others see 20 million jobs displaced eventually. Near-term data shows more task redistribution than outright elimination. Roles evolve. Expectations rise. Junior positions shrink while mid-career hires with proven skills gain favor.

Visier’s analysis of 3.6 million records, released in August 2026, captured the shift. Overall hiring rates dropped 24 percent. Yet jobs like educators, lawyers, architects, and security professionals proved hard to replace. Data and analytics roles grew as a share of headcount. AI engineers surged. Data scientists declined.

The pattern favors experience. Companies now prioritize workers aged 35 to 50 over younger talent in some cases. The rising bar Nicole Bachaud of ZipRecruiter described in recent interviews explains much. Employers want AI literacy on day one. Communication, adaptability, and critical thinking matter more than ever. The labor market tightens around skills rather than sheer headcount.

So the CNBC segment didn’t reveal a conspiracy. It surfaced an uncomfortable truth. AI offers employers a psychological edge in wage negotiations. Combined with slower hiring, retraining instead of firing, and selective augmentation, it delivers cost control without the political backlash of mass layoffs.

Whether those productivity claims eventually prove out at scale remains open. Early evidence is patchy. Gains appear in specific functions. Broader statistics lag. But the worker fear is real and measurable. Wages stagnate even as the economy hums. People stay in jobs they once might have left.

That dynamic may not last forever. New roles will emerge. Training programs expand. Some companies already invest in internal mobility to avoid the higher cost of constant external hiring. Yet for now, the balance tilts toward caution. Employers have discovered AI’s value lies as much in what it prevents — higher pay, bolder demands — as in what it automates.

The candid television exchange cut through the corporate messaging. Productivity is the public story. Control is the private one. And workers, staring at machines that could do parts of their jobs cheaper, understand the difference all too well.