Financial services firms are scaling artificial intelligence faster than any other enterprise sector, and they are doing it in the places customers never see.
That is the central finding of the latest Enterprise AI Benchmark Report from PYMNTS Intelligence, based on a survey of senior technology executives at U.S. enterprises with at least $1 billion in annual revenue, split across financial services and insurance, healthcare, and media and advertising. The survey tracked adoption across 75 AI-supported tasks.
Financial services firms reached high adoption, defined as at least half of firms actively using AI for a task, in 27 of the 75 tasks. Healthcare hit that mark in just 10, media and advertising in 16.
The pattern behind those numbers matters more than the numbers themselves. Financial firms are deploying AI where outcomes are verifiable, data pipelines are clean and results can be defended to regulators. Revenue recognition leads at 65% adoption, with credit risk assessment and sales forecasting at 60% each. Customer-facing applications trail badly: churn prediction sits at 30%, KYC identity verification at 20% and A/B testing at 10%, the lowest rate for that task in the entire survey.
AI delivers fastest in structured, auditable, rules-based processes, exactly the environment that describes scoring, segmentation, forecasting and compliance workflows in an ARM shop. The harder, riskier bets are on consumer-facing tools where errors carry regulatory consequences, a calculus the industry knows well.
Budgets are climbing regardless. Eighty-five percent of financial services firms plan to increase AI spending over the next 12 months, citing productivity gains and competitive positioning (65% each) and risk reduction and compliance (55%).
The constraint is not appetite. Three in 10 financial services leaders named data quality and fragmentation as the primary barrier to deeper deployment, a notable admission from the sector that has scaled furthest. The firms bumping against that ceiling have learned that AI is only as good as the placement files, account histories and system integrations feeding it.
One more data point worth noting: 80% to 85% of respondents across all sectors expect AI to augment human decision-making over the next five years, not replace it. None anticipate full autonomy.
