Which industries have been hit hardest by AI adoption?

Information and finance/insurance sectors reported relatively high AI adoption alongside declining demand for workers, suggesting some companies may be using the technology to contain labor costs. These were identified as possible exceptions to the broader pattern of no clear relationship between AI exposure and employment decline.

Are younger workers being affected differently?

Entry-level employment presents a less encouraging picture. Unemployment among workers aged 22 to 27 has risen from its 2023 lows and remains above 2019 averages, and BofA said AI may be contributing to weaker employment outcomes for this group, though trade-policy uncertainty also played a role.

AI Safety & AlignmentYahoo Finance AI3h agoAug 15, 2026, 10:00 JST4 min read

AI replaces tasks, not jobs; BofA finds flat employment in exposed sectors

Send on LINE

Key takeaway

BofA Global Research examined employment across 206 US industries and found little correlation between an industry's AI exposure and job losses, suggesting artificial intelligence is replacing specific tasks rather than wiping out entire occupations.
While sectors heavily exposed to AI showed flat employment since ChatGPT's release in late 2022, the researchers found almost no evidence that companies have reduced working hours or broadly cut labor demand in response to the technology.
However, entry-level workers aged 22 to 27 have experienced rising unemployment, which BofA says AI may be contributing to, though offsetting job gains in AI infrastructure construction and manufacturing are creating new positions.

3 Key Points

What happened
BofA Global Research examined employment across 206 US industries and found almost no correlation between an industry's AI exposure and its employment growth since ChatGPT was released in late 2022. Industries with the highest AI exposure recorded broadly flat employment, while the least-exposed industries saw employment rise by about 2% over the same period.
Why it matters
The analysis suggests AI is more likely to replace individual tasks than entire occupations, limiting the risk of widespread job destruction. Aggregate hours worked showed little connection with AI exposure, and labor demand measured by employment and job openings showed no clear relationship with AI usage rates between January and June 2026—though information and finance/insurance sectors did report relatively high AI adoption alongside declining worker demand, suggesting some companies may be using the technology to contain labor costs.
What to watch
Entry-level employment presents a less encouraging picture, with unemployment among workers aged 22 to 27 rising from 2023 lows and remaining above 2019 averages; BofA said AI may be contributing to this weakness. On the positive side, non-residential construction added 95,000 jobs during 2026 through the report's publication date, and AI-related manufacturing industries created another 32,000—together accounting for roughly one-quarter of new private-sector jobs during the period.

Summaries like this, in your inbox every morning.

Get news like this every morning — free

In Depth

Read the full story

BofA Global Research released a report examining employment patterns across 206 US industries to assess whether artificial intelligence has driven widespread job destruction since ChatGPT was released in late 2022. The headline finding: industries with the highest exposure to AI have recorded broadly flat employment over that period, while industries with the least AI exposure have seen employment rise by about 2%.

More striking than the absolute numbers is the lack of correlation. When researchers looked for a relationship between an industry's AI exposure and its employment growth, they found almost none. This disconnect held across multiple measures. Aggregate hours worked showed little connection with AI exposure, suggesting companies have not broadly reduced employees' working hours in response to the technology. Labor demand, measured using employment and job openings and tracked between January and June 2026, also showed no clear relationship with AI usage rates.

Two sectors stand out as possible exceptions: information and finance/insurance both reported relatively high AI adoption alongside declining demand for workers. BofA interprets this as evidence that some companies in these sectors may be using the technology to contain labor costs rather than maintain headcount. The broader pattern, however, supports the idea that AI replaces individual tasks rather than entire occupations—limiting the risk of wholesale job destruction.

The employment picture for younger workers, however, is less encouraging. Unemployment among workers and recent college graduates aged 22 to 27 has risen from its 2023 lows and remains above 2019 averages. BofA said AI may be contributing to weaker employment outcomes for this cohort, though the report also notes that trade-policy uncertainty played a role. On a more positive note, spending on AI infrastructure is creating jobs elsewhere. Non-residential construction added 95,000 jobs during 2026 through the report's publication date, and AI-related manufacturing industries created another 32,000. Together, these two sectors accounted for roughly one-quarter of new private-sector jobs during the period. BofA expects such gains, along with the creation of new tasks and roles, to offset some initial displacement from AI adoption.

Context & Analysis

BofA's analysis of 206 US industries provides empirical grounding for skepticism about AI-driven mass unemployment. The researchers examined three separate labor metrics—employment growth, aggregate hours worked, and labor demand measured by job openings—and found no consistent relationship between an industry's exposure to AI and adverse employment outcomes. This lack of correlation is itself significant: it suggests that other factors (such as aggressive hiring in the years following 2019) may explain employment weakness in highly exposed sectors, rather than the technology itself.

However, the report acknowledges important caveats. Information and finance/insurance sectors broke the pattern, showing high AI adoption alongside declining worker demand, which implies that some employers are indeed using the technology to contain labor costs rather than create new roles. More concerning is the entry-level labor market: unemployment among workers aged 22 to 27 has risen above 2019 levels, and BofA attributes this partly to AI's influence. The offsetting job creation in AI infrastructure construction and manufacturing—together accounting for roughly one-quarter of new private-sector jobs in 2026—suggests that while displacement may occur in certain occupations, new roles are emerging elsewhere. Whether these newly created positions match the skills, geography, and compensation of displaced workers remains an open question the body does not address.

FAQ

Which industries have been hit hardest by AI adoption?: Information and finance/insurance sectors reported relatively high AI adoption alongside declining demand for workers, suggesting some companies may be using the technology to contain labor costs. These were identified as possible exceptions to the broader pattern of no clear relationship between AI exposure and employment decline.
Where are new jobs being created related to AI?: Non-residential construction added 95,000 jobs during 2026 through the report's publication date, and AI-related manufacturing industries created another 32,000. Together, these two sectors accounted for roughly one-quarter of new private-sector jobs during the period.
Are younger workers being affected differently?: Entry-level employment presents a less encouraging picture. Unemployment among workers aged 22 to 27 has risen from its 2023 lows and remains above 2019 averages, and BofA said AI may be contributing to weaker employment outcomes for this group, though trade-policy uncertainty also played a role.

Yahoo Finance AIRead Original Article