Technology & innovation
A panel of four economists and workforce experts convened by The New York Times examined how artificial intelligence is likely to transform work, which workers face the greatest risk, and what (if anything) can be done to prepare.
MIT economist and Nobel laureate Daron Acemoglu argued that the vision of AI agents doing most work while humans supervise is "very unrealistic" in the near term, and that manufacturing, health care, and education remain sectors where AI's promise far outpaces its current delivery.
Wharton professor Ethan Mollick said the transformation will be uneven and slower inside large companies with established structures, but that a boom in AI-powered startups is already underway, with early evidence that AI is blurring the lines between technical and business roles.
Clara Shih, co-founder of New Work Foundation, warned that entry-level workers in insurance, claims adjusting, and long-distance trucking face the most acute near-term displacement risk, and said the labor market is already splitting into two groups: those who understand how AI agents work and those who don't.
Dean Ball, formerly an AI adviser in the Trump administration, said change will feel incremental day to day but transformative in retrospect, and cautioned against labor market rigidities that could limit firms' ability to take risks and adapt.
Mollick and Acemoglu both raised concerns about junior workers losing access to the apprenticeship-style learning that has historically allowed entry-level employees to develop judgment and be assessed by managers.
Read more via The New York Times
A growing number of companies are turning to structured interviews, AI-powered assessments, and gamified simulations to replace the gut-instinct hiring process that research consistently shows is a poor predictor of job performance.
A recent Robert Half survey found that nearly a third of U.S. hiring managers said they had made a hiring mistake in just the past two years, with failure to accurately assess skills or cultural fit cited as the top reasons.
LinkedIn has been testing AI-powered pre-interviews for small businesses, where an automated agent asks candidates skills-based questions before a human ever gets involved.
Greenhouse, a hiring software company, said customers conducted 15 million structured interviews last year, up from roughly 500,000 in 2015.
Researchers caution that gamified assessments must be carefully designed to avoid introducing new biases. (Experts say men, on average, have more experience with certain game formats than women, which can skew results if not accounted for.)
Candidates' growing use of AI to craft applications means employers may not be shortlisting the best candidates, but those who present best, according to a Wharton management professor.
Read more via The Wall Street Journal
While return-to-office battles play out across corporate America, employees at post-pandemic AI startups are largely showing up voluntarily, and often staying late, in a dynamic that workplace experts say reflects the unique culture and stakes of early-stage AI companies.
Stanford economist Nicholas Bloom told Business Insider that the combination of young employees and significant equity stakes creates a work mode that is "almost entirely in-person" and "100% work focused," coining the phrase: "They don't work from home, they home from work."
Founders of several AI startups said they have never issued a return-to-office memo. Employees simply come in, often on weekends, without being asked.
Urban studies theorist Richard Florida attributed the pattern in part to the nature of AI innovation itself, noting that AI companies need constant interaction with end users as a core part of the development process.
Founders emphasized that in-person work supports the psychological safety and high-trust environment needed to experiment and fail quickly on hard technical problems.
Read more via Business Insider
White-collar workers are spending an average of 6.4 hours a week supervising, correcting, and cleaning up after AI tools, according to a new report from Glean's Work AI Institute, conducted with researchers from Notre Dame, Stanford, and UC Berkeley.
Highlights from the survey of 6,000 full-time workers in the U.S., U.K., and Australia:
87% said they use AI at work and 75% said it makes them more productive individually, but only 13% said their organization was performing significantly better as a result.
"Botsitting" describes the often-unrecognized work of feeding AI context, checking outputs, debugging errors, and moving information between disconnected systems.
Workers who spend an unusually large share of their AI time botsitting are 73% more likely to be actively looking for another job.
In some cases, workers are being asked to automate the parts of their jobs they find most meaningful, including customer-service employees who value relationship-building but are increasingly expected to supervise AI agents instead.
The organizations seeing the biggest gains are those investing in the work around AI: helping employees access context, building judgment, and establishing clear standards for what good AI-assisted output looks like.
Meta is launching America's Workforce Academy, a nationwide skilled trades training program that covers all costs for participants, awards industry-recognized credentials, and guarantees a job upon graduation, backed by an initial $115 million investment.
The program targets trades including electrical work, welding, plumbing, fiber installation, and HVAC, responding to what Meta describes as a shortage of hundreds of thousands of skilled trade workers needed to build out AI infrastructure.
Graduates receive both an NCCER credential and an America's Workforce Certificate, both portable across employers and industry sectors.
The 2026 pilot will launch in Louisiana, Ohio, Indiana, and Texas. Partners include the National Urban League, Associated Builders and Contractors, CBRE, and several regional workforce organizations.
Meta cited its prior Level-Up fiber installation training program, which received 35,000 applications in its first seven days, as the model for this expansion.
Read more via Meta
State Farm CEO Jon Farney told the insurer's annual agent convention last month that existing contracts for the company's 19,000 sales agents are being torn up, with anyone who wants to stay past 2027 required to sign a new compensation deal. The move is driven in significant part by AI and the competitive pressure it has created in the insurance industry.
The new contract ends a deferred compensation program, eliminates health benefits, and changes commissions on several policy types. Depending on their book of business, agents' gross annual income could fall by as much as 40%.
State Farm's new "Next Gen Good Neighbor" initiative will introduce AI-powered digital assistants, household-level product recommendations, and an AI assistant for customers reporting auto losses.
Agents who choose to leave can apply for an exit payment of between $50,000 and $300,000, at State Farm's discretion. Agents told The Wall Street Journal the offer is a fraction of what many agencies are worth.
Meanwhile, competitors are moving in a different direction: Progressive, which sells more than half its personal auto policies direct to consumers using AI and technology, this year displaced State Farm as the nation's largest personal auto insurer, a title State Farm had held since World War II.
KPMG pulled a report on AI after it turned out to contain AI hallucinations. The report made false claims about how UBS, the U.K.'s National Health Service, Swiss Federal Railways, and Transport for London use AI, all of which those organizations disputed. Research group GPTZero identified the inaccuracies and told the Financial Times they stemmed from AI-generated content. A KPMG spokesperson said the firm removed the report while conducting its own investigation and noted that its guidelines require "human oversight to validate content and verify independent sources." Last month, EY withdrew a separate report that appeared to contain fake footnotes and AI hallucinations. (TechCrunch)
Nvidia CEO Jensen Huang is betting that AI will create manufacturing jobs, not just eliminate them. Nvidia announced a $2 billion partnership with Coherent to expand production of a laser component used to connect AI chips into unified systems at a factory in Sherman, Texas, expected to create roughly 1,000 jobs, about 550 of them in advanced manufacturing, engineering, and technical roles. Huang has been a fixture on Trump administration foreign trips, with the president calling him "smart," a "friend," and "amazing." (Associated Press)
JPMorgan Chase plans to deploy AI agents later this year that can operate autonomously for hours at a time. The bank's chief analytics officer Derek Waldron said the arrival of "long-running autonomous agents" marks a new milestone in corporate AI adoption. The bank said AI is already producing results, including a 20% increase in gross sales in its private banking business, and believes the tools could eventually allow individual bankers to expand client coverage by as much as 50%. CEO Jamie Dimon has said some workers will be displaced, and the firm is preparing to retrain and redeploy affected employees. (CNBC)
Honeywell CEO Vimal Kapur said AI is poised to "redefine automation." Kapur made the comments as the industrial conglomerate prepares to spin off its aerospace division and become a pure-play automation company on June 29. He said Honeywell's systems already generate enormous volumes of operational data that AI can now turn into actionable insights, and that labor shortages and aging populations are accelerating demand. "Net workforce is not going to be increasing," he said. "It's going to be decreasing over a period of time." (CNBC)
Mark Zuckerberg told Meta employees in an internal memo that the company has made mistakes in its AI workforce transformation and will "almost certainly make more." Meta laid off 10% of its global workforce in May and transferred 7,000 employees to AI-related roles. Zuckerberg said the company will try to find new positions for workers reassigned to AI model training and reiterated that no further company-wide layoffs are planned this year. He also said Meta will scale back the practice of widening manager oversight ratios, which had reportedly reached 50-to-1 in some units. (Reuters)
