Nearly all organisations are using artificial intelligence in some capacity, but most remain in the experimentation or piloting phase rather than scaling enterprise-wide, according to McKinsey's latest global survey on AI.

The report, titled "The State of AI in 2025: Agents, innovation, and transformation," reveals that 88 percent of respondents report regular AI use in at least one business function, up from 78 percent in 2024 and continuing an upward trend that began with just 20 percent in 2017.

Despite this increase, approximately two-thirds of respondents say their organisations have not yet begun scaling AI across the enterprise, with only 7 percent reporting fully scaled deployment and 31 percent in the scaling phase.

"Most organisations have not yet embedded AI technologies deeply enough into their workflows and processes to realize material enterprise-level benefits," notes the report authored by McKinsey's QuantumBlack AI practice leaders.

The survey found significant differences in AI maturity based on organisation size. Nearly half of respondents from companies with more than $US5 billion in revenue have reached the scaling phase, compared with just 29 percent of those with less than $US100 million in revenues.

"While only one-third of all respondents say they are scaling their AI programs across their organisations, larger companies - both in terms of revenues and the number of employees - are more likely to have reached the scaling phase," the report states.

AI high performers - defined as organisations reporting EBIT impact of 5 percent or more from AI use and "significant" value attribution - represent only about 6 percent of respondents but provide important insights into successful implementation approaches.

These high performers distinguish themselves through three key practices:

1. Setting transformative goals: "AI high performers are more than three times more likely than others are to say their organisation intends to use AI to bring about transformative change to their businesses."
2. Pursuing growth and innovation: "While most respondents report that efficiency gains are an objective of their organisations' AI use, high performers are more likely than others are to say their organisations have also set growth and/or innovation as an objective of their AI efforts."
3. Redesigning workflows: "High performers are nearly three times as likely as others are to say their organisations have fundamentally redesigned individual workflows."

The survey revealed growing experimentation with AI agents, defined as "systems based on foundation models capable of acting in the real world, planning and executing multiple steps in a workflow." Sixty-two percent of respondents say their organisations are at least experimenting with AI agents, with 23 percent already scaling these systems somewhere in their enterprises.

However, agent use remains concentrated in specific functions. "Looking at individual business functions, agent use is most commonly reported in IT and knowledge management, where agentic use cases such as service-desk management in IT and deep research in knowledge management have quickly developed."

By industry, AI agent adoption is most prevalent in technology (22 percent scaling), media and telecommunications (16 percent), and healthcare (15 percent) sectors.

Fifty-one percent of respondents report experiencing at least one negative consequence from AI use, with inaccuracy (30 percent), cybersecurity concerns (10 percent), and regulatory compliance issues (8 percent) being the most common challenges.

The share of respondents reporting mitigation efforts for risks such as personal and individual privacy, explainability, organisational reputation, and regulatory compliance has grown significantly since 2022, with organisations now addressing an average of four AI-related risks compared to just two risks previously.

Despite the focus on efficiency in AI implementations, the employment impact remains uncertain. While a plurality (43 percent) of respondents expect to see little or no effect on their organisations' total number of employees in the year ahead, 32 percent predict an overall reduction of 3 percent or more, and 13 percent predict an increase of that magnitude.

"Respondents at larger organisations are more likely than those at smaller ones to expect an enterprise-wide AI-related reduction in workforce size, while AI high performers are more likely than others are to expect a meaningful change, either in the form of workforce reductions or increases," the report states.

At the same time, most respondents note that their organisations hired for AI-related roles over the past year, with software engineers and data engineers being the most in demand.

The survey collected responses from 1,993 participants across 105 nations between June 25 and July 29, 2025, representing diverse industries, company sizes and functions. Thirty-eight percent of respondents work for organisations with more than $US1 billion in annual revenues.

McKinsey's full report can be accessed here.