EY AI Pulse Survey: How AI Is Turning Promise into Payoff

Originally published by EY (Ernst & Young). Source: AI Survey: How AI Is Turning Promise into Payoff – EY, December 2025

EY’s fourth wave of its US AI Pulse Survey, published in December 2025, captures a meaningful inflection point: the “gold rush” phase of AI adoption is maturing into something more measured—and more valuable. But the path from promise to payoff is unevenly distributed.

The Productivity Numbers Are Real

The headline finding is harder to dismiss than analysts might expect:

96% of organizations investing in AI report AI-driven productivity gains over the past year
57% are experiencing significant AI-driven productivity gains
97% report positive ROI from their AI initiatives
Organizations investing $10M or more in AI are 36% more likely to see significant gains than those investing less (71% vs. 52%)

These numbers don’t support the narrative that AI adoption is mostly hype. At the organizations that have committed meaningfully—in spending, governance, and change management—AI is producing measurable results.

The Measurement Problem

But here’s what’s missing from the success story: clarity. 65% of leaders struggle to directly attribute productivity gains to AI adoption. And 92% agree that more work is needed on how to measure and report AI-driven productivity gains.

This is a critical gap. Organizations that can’t tie outcomes to specific AI investments can’t optimize those investments, can’t defend them to boards, and can’t scale them confidently. The measurement infrastructure for AI—baselines, attribution models, leading indicators—is still underdeveloped at most organizations.

The Investment Ambition-Reality Gap

The survey reveals a significant disconnect between what leaders forecasted and what they delivered:

65% predicted spending $1M+ on AI a year ago; only 58% actually did
Only 24% currently allocate 25%+ of total budgets to AI
48% expect to allocate 25%+ next year—double the current rate

As EY characterizes it: “Ambition is outpacing execution.” The organizations that plan to double their AI budget share next year will need to demonstrate that their execution infrastructure—governance, talent, data foundations—can support that ambition.

Responsible AI Is Moving from Principle to Practice

One of the more encouraging findings in Wave 4 is the concrete movement on responsible AI:

60% increased Responsible AI training for employees over the past year
68% plan to increase focus on ethical AI operations in the coming year
63% will increase transparency with customers about AI use (up from 55%)

EY’s Global Risk Leader Kapish K. Vanvaria noted that responsible AI is showing “meaningful follow-through” compared to other stated priorities. This matters because organizations that embed governance early avoid the costly retrofits that come when regulators, customers, or employees push back.

The Real Breakthrough: Amplification, Not Just Automation

EY Americas Technology Leader Colm Sparks-Austin put the core insight plainly: “The real breakthrough isn’t automation—it’s amplification.”

The organizations generating the most value from AI aren’t replacing people with algorithms. They’re using AI to make their best people dramatically more effective—expanding the scope of what knowledge workers can do, the speed at which they can operate, and the quality of the outputs they produce.

EY Global Consulting AI Leader Dan Diasio framed the strategic question organizations should be asking: not just “what can AI automate?” but “what value can we add with the time and capacity AI creates?”

EY Americas AI and Data Leader Traci Gusher identified what separates the leaders in Wave 4: they “pair bold strategic goals with AI investment goals”—ensuring that the ambition level of their AI spending matches the ambition level of their business objectives.

The organizations that close the measurement gap, close the spending ambition-reality gap, and make responsible AI operational—not just aspirational—are the ones who will compound their early productivity gains into durable competitive advantage.

What This Research Misses

The 97% positive ROI figure is in direct tension with PwC’s CEO survey from the same period. PwC’s 2026 Global CEO Survey found only 1 in 8 CEOs have seen AI deliver both cost and revenue benefits. EY’s survey respondents are “senior leaders” in organizations actively investing in AI — a self-selected group that excludes organizations still evaluating or with stalled programs. PwC’s broader CEO panel captures the full population of decision-makers, including those whose AI investments haven’t produced results. The 97% figure is almost certainly inflated by selection bias.

65% unable to attribute productivity gains to AI is a methodological confession that undermines the headline. If nearly two-thirds of respondents can’t measure whether their productivity gains came from AI, then the 96% productivity claim is largely self-reported conviction, not measured outcome. Academic research on the “productivity paradox” — the historical difficulty of finding macroeconomic productivity gains from IT investment — suggests this measurement problem is structural, not just a data gap. Researchers Erik Brynjolfsson and Daniel Rock at MIT have documented that general-purpose technologies like AI show measured productivity impact with significant lags (often 10–20 years), because the organizational co-investments needed to fully exploit them take time to accumulate.

The responsible AI investment data is more encouraging than the practice data supports. While 60% increased RAI training and 68% plan more ethical AI focus, Slalom’s 2026 research found only 50% of organizations have formal AI audit or ethics boards and only 42% implement model controls like bias or drift monitoring. EY’s survey captures intent and investment in training; Slalom’s captures actual governance infrastructure. The gap between training investment and governance implementation suggests responsible AI is still primarily a communications and training initiative rather than an operational practice.

The ambition-reality spending gap is real but the direction of causality is unclear. EY presents the gap (65% predicted $1M+ spending a year ago; only 58% delivered) as evidence that ambition outpaces execution. But it could equally reflect rational learning: organizations that planned large AI budgets discovered deployment barriers and adjusted. This interpretation — careful calibration rather than execution failure — changes the recommended response from “execute harder” to “plan more realistically.”

Research Methodology: 500 US senior leaders across diverse industries. Wave 4 of EY’s ongoing AI Pulse Survey series, December 2025.

Source: EY AI Pulse Survey: How AI Is Turning Promise into Payoff