Explainer Collection

AI Adoption

Yu (2026)

AI has reached core operations, but the productivity payoff is still missing

Advanced AI adoption among S&P 500 firms more than quadrupled between 2022 and 2025. Deep integration was associated with higher profit margins, but not with higher revenue per employee.

21%: of S&P 500 firms using AI in production or integrating it deeply by 2025
67%: of advanced adopters concentrated in the technology sector

Acharya (2026)

Governance maturity makes agent fleets safer, but the evidence is still simulated

A 750-run multi-agent simulation finds that higher governance maturity sharply reduces agent sprawl and risk incidents while improving task completion and composite Net Business Value. Its pivotal claim is that Level 3 is the minimum viable standard, not a decorative middle rung.

Level 1: reactive governance with high shadow-agent and orphan-agent exposure
Level 2: an early control stage that barely improves adversarial outcomes

Hitzig (2026)

Coding agents make a software background optional, but expertise still decides who succeeds

An analysis of roughly 400,000 Claude Code sessions finds a clear split of labour: people decide what to build, the agent decides how. Command of a domain, not the ability to write code, is what makes sessions succeed.

70%: of planning decisions ('what to do') are made by the person, while Claude makes about 80% of execution decisions
2×: Expert-rated sessions reach verified success more than twice as often as novice-rated ones

Cambridge CCAF (2026)

Finance has gone all-in on AI, but the supervisors watching it have not

A 628-organisation, 151-jurisdiction survey finds 81% of financial firms now using AI, while regulators trail on adoption, data collection and the supervisory tools needed to keep up.

81%: of surveyed financial firms are adopting AI at some level, but only 14% see it as transformational to strategy
40% vs 20%: share of industry vs regulators reporting advanced AI adoption (Scaling or Transforming)