
Fed conference weighs AI productivity, regulation and job losses
Executives at a Federal Reserve conference reported fivefold engineering productivity, doubled Stripe business starts and uneven job-displacement risks shaped by regulation.
- By
- Tom Whitfield
- Filed
- Channel
- AI & Compute
- Read
- 4 min read
By Tom Whitfield
Affirm CEO Max Levchin told a Federal Reserve task-force conference that AI has made the company’s engineers five times as productive and cut the cost of building new features by 80%. Stripe CEO Patrick Collison said weekly new business starts on Stripe’s platform had doubled from a year earlier, which he called unprecedented in the company’s 15-year history.
The disclosures came during a panel moderated by Federal Reserve Chair Kevin Warsh at the first conference of his productivity-and-jobs task force, held at the Federal Reserve Bank of San Francisco. The measures cover different boundaries: Affirm reported gains in internal software work, while Stripe counted new businesses beginning on its payment platform.
Collison tied Stripe’s performance to US leadership in AI use. “The US is in a definitively preeminent position today,” he said. “We see it directly in our data: AI usage intensity is higher in the US than elsewhere.”
Why would regulation limit both job losses and growth?
Axios reported that Marc Andreessen, who co-leads the Federal Reserve’s task force on productivity and jobs, told Warsh on Monday that regulation and weak competition will keep AI from transforming large parts of the economy. He said that outcome would limit job losses and productivity gains at the same time.
Andreessen focused on regulated, subsidized sectors, including health care, education and housing. Those sectors are “not legally allowed to have rapid change,” he said, while prices keep falling in faster-moving sectors such as technology and media.
His mechanism links the pace of institutional change to both sides of the AI debate. Rules that slow adoption can limit job displacement, but the same restrictions can prevent rapid workflow changes and large productivity gains; in faster-moving sectors, prices continue to fall.
Andreessen made the argument during a closing fireside chat with Warsh and former Commerce Secretary Gina Raimondo. His co-leadership placed him at the link between executive operating data and the Fed’s policy recommendations. The implication is twofold: rules that reduce disruption can also slow productivity gains.
What did the executives tell the Fed?
Stripe’s US observation came from company data rather than an economy-wide output statistic. On a later panel, Providence CEO Erik Wexler said AI transcription tools save some physicians two to four hours of paperwork each day. His estimate applies to selected users and specific administrative tasks; it does not quantify the tool’s effect across health care employment.
Palo Alto Networks CEO Nikesh Arora said job displacement remains some way off because companies have not yet given AI independent authority to act on people’s behalf. In his view, that missing capability marks where displacement begins, because autonomous action can change work previously assigned to people.
The reported gains concern software output, business formation and time saved on selected tasks. Arora’s account of job displacement depends on a capability companies have not yet deployed. The evidence supports a sector-by-sector assessment rather than a national jobs forecast.
What did Warsh ask the task forces to produce?
Warsh opened the conference with three staff scenarios for AI. One scenario assumes the AI boom disappoints and stock valuations fall, giving policymakers a downside case before they consider recommendations.
He asked all five of his task forces to propose improvements to Federal Reserve policymaking by year-end. Andreessen shares the productivity-and-jobs group with Microsoft executive Asha Sharma and Stanford economist Charles Jones.
Warsh said some groups may miss the deadline, leaving the proposals in progress. The conference offers a starting point for Fed policy, not a completed framework.
What do the disclosures establish?
Taken together, the conference produced several measurable company claims but no single estimate for national productivity or employment. Affirm’s fivefold figure and 80% cost reduction concern engineering work; Stripe’s doubled weekly starts measure business formation on one platform; Wexler’s estimate concerns transcription; and Arora’s timing argument concerns independent AI authority.
The evidence points to uneven adoption, with Affirm reporting faster output and lower development costs in selected workflows. Regulation and weak competition can restrain change in large service sectors, even as technology and media prices fall.
The year-end recommendations will test whether the Fed can address those competitive constraints. The near-term commercial picture should remain split between measurable workflow gains and slow change in regulated services.
Source: mlq.ai
More from Tom Whitfield
Show full bio
Staff writer covering companies, deals and strategy in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,054 articles
Related articles
fed-chair-warsh-to-grill-ai-s-economic-claims-in-san-francisco
Fed Chair Warsh to Grill AI's Economic Claims in San Francisco
ai-delivers-zero-financial-payoff-outside-big-tech-apollo-economist-says
AI Delivers Zero Financial Payoff Outside Big Tech, Apollo Economist Says
warshs-three-ai-scenarios-test-a-trillion-dollar-investment-bet
Warsh’s three AI scenarios test a trillion-dollar investment bet
fed-s-barr-warns-ai-agents-could-trigger-bank-runs-in-minutes
Fed's Barr Warns AI Agents Could Trigger Bank Runs in Minutes



