Technology & innovation
New research finds that 71% of AI-related job postings by S&P 500 companies are for senior-level positions, with just 13% open to junior candidates, raising questions about how the next generation of workers enters a field that is reshaping the economy.
Researchers from the AIDE Institute analyzed more than 161,000 LinkedIn job postings by major corporations and found AI hiring is, in their words, "overwhelmingly senior, leaving few openings for beginners."
In AI-exposed occupations, young workers saw a 6% decline in employment between late 2022 and September 2025, while older workers in the same fields saw gains of 6% to 9%, according to Stanford research.
The unemployment rate for recent college graduates stood at 5.6% as of March, well above the overall rate of 4.2% that month, according to the New York Fed.
The task at hand is to give young people a boost to that second rung of more complex tasks and responsibilities that are higher value and what the firms are hiring for. The entry point has just shifted upward."
ADP Chief Economist Nela Richardson said the challenge for employers is not discarding the career ladder for young workers but shifting the entry point upward toward higher-complexity tasks that AI cannot yet perform.
Read more via CNN
Wells Fargo CEO Charlie Scharf pushed back on binary takes about AI's impact on jobs at a recent Bernstein investor conference, saying the reality inside his own company is more complicated than the debate suggests.
Scharf said most of the bank's 201,000 employees have already been given basic AI tools, with executives assessing efficiency gains in areas including auditing, legal, contracts, patent filings, investment banking pitchbooks, and credit memos.
The bank plans to hire more people who can build AI models or use them to better serve customers, but Scharf acknowledged a timing mismatch between where efficiency gains land and where new roles emerge.
It's so obvious to me, looking at the way we're using AI inside the company, it is both of those things. The risk is that they're not totally aligned, in terms of the same people and the timing of it."
Read more via HR Dive
BCG's fourth annual AI at Work survey of nearly 12,000 workers across 14 markets finds widespread adoption but an organizational failure to convert time savings into measurable value.
Nearly half of respondents (47%) now report spending more time managing and directing AI than doing the actual work, and 72% say AI has considerably changed skills expectations in their roles.
Frontline employee AI adoption has surged to 74% regular users, up more than 20 percentage points over two years, with India, Brazil, and the Middle East leading adoption globally.
While 42% of regular frontline users report saving at least a full workday per week through AI, 66% say they get little or no guidance on what to do with that time.
Strategic clarity proved a stronger driver of AI impact than better tools, lifting measurable business outcomes by 25 percentage points compared to roughly 5 points for tools alone.
Read more via Boston Consulting Group
New peer-reviewed research from Penn State and USC finds that how employees use AI matters as much as whether they use it, with passive reliance on AI-generated content producing measurable psychological costs.
Participants who copy-and-pasted AI responses rather than collaborating with AI reported nearly 20% declines in feelings of ownership over their work and roughly 10% declines in perceived meaningfulness.
The declines in self-efficacy and meaningfulness persisted after participants returned to working without AI, suggesting the effects are not easily reversed.
Passive AI use produced an initial spike in task enjoyment and satisfaction, but when participants returned to manual work, outcome satisfaction fell 21% below those who had worked manually throughout.
Collaborative AI use, where workers used AI to develop and refine their own ideas, showed outcomes comparable to working without AI at all.
Read more via Penn State
An HBR analysis of nearly 13,000 real-world AI use cases finds people turning to generative AI for an increasingly wide and intimate range of tasks, with emotional support growing fastest and workplace use growing messier.
Therapy and companionship ranked as the top use case for the second year running, growing from 5% to 11% of the dataset in a single year, with users increasingly naming AI tools, assigning them genders, and in some cases describing grief when a model is updated.
Shadow usage is widespread, with employees quietly using AI tools their employers haven't sanctioned and, in some cases, building agents to automate significant portions of their own jobs without disclosure.
The researchers flagged a pattern they call "thinkslop," where low-effort AI use erodes users' ability to think independently, develop original ideas, or maintain intellectual rigor over time.
Read more via Harvard Business Review
AI-powered chatbots are great, until they help hackers access your company's data: A bug in the company's customer service AI-powered chatbot allowed hackers to reset passwords for roughly 34,000 Instagram accounts, including one belonging to a senior Space Force official and a former presidential social media account, by simply asking the chatbot to make the change; 20,000 accounts were fully breached before Meta fixed the flaw, and the company said it would not make major changes to its broader AI product plans as a result. (The New York Times)
AI impersonation attacks are outpacing corporate defenses: More than half of organizations were hit this year, yet three-quarters still rely on limited monitoring or purely reactive responses, according to a survey of 1,100 cybersecurity and risk leaders by security firm Outtake; the report found only 4% of businesses have full visibility and active controls over their AI agents' external interactions. (Cybersecurity Dive)
PepsiCo is delivering Doritos without a driver: The company's 41 driverless trucks are now operating across Arizona, Texas, and Arkansas, making it the first major U.S. consumer goods company to run autonomous vehicles at scale on public roads; the trucks have logged a 99% on-time arrival rate and no accidents since launching in June 2025, but the Teamsters have lobbied multiple states to require a human operator in any autonomous commercial vehicle, and PepsiCo acknowledged it expects to hire fewer drivers overall as the program expands. (The Wall Street Journal)
AI may be able to spot breast cancer six years early: A Swedish study published in Radiology tested three commercially available AI detection systems on mammogram data from nearly 32,000 patients and found the systems identified early signs in roughly 20% of individuals who were eventually diagnosed, well ahead of when radiologists detected them. (PR Newswire)
