McKinsey - AI adoption is leveling off, but AI leaders are pulling ahead

McKinsey’s Global Survey on the State of AI is in its fifth year and provides a nuanced picture of how organizations are making use of the technology. This year’s report shows that whilst the adoption of AI is leveling off, more capabilities are being used now than they were a few years ago - and that leaders in the field are pulling ahead of the rest of the pack. 

In addition, and this will likely not come as a surprise to most, the survey shows that AI teams are largely lacking in diversity. As diginomica has noted time and time again, this is worrying given that these AI models need to build trust with all sections of the population and prove their decision making is unbiased (which even with diverse teams is uncertain, given datasets themselves inherently contain bias). 

McKinsey surveyed 1,492 participants representing the ‘full range of regions, industries, company sizes, functional specialities, and ensures’. Of those respondents, 744 said their organizations had adopted AI in at least one function and were asked about their organizations’ use of AI. 

The survey is useful for consistently tracking how organizations are using AI, what their pain points are and provides insight into how companies should be thinking about their development of artificial intelligence. 

Since the first survey was done in 2017, McKinsey has seen AI adoption increase from 20 percent of respondents to approximately 50 percent. However, adoption peaked in 2019 at 58 percent and has since leveled off. 

However, the average number of AI capabilities that organizations use has also doubled - from 1.9 in 2018 to 3.8 in 2022. 

Of the capabilities used, some 39% said that they’re using RPA in at least one business process or product, the highest performing. This is closely followed by computer vision (38%), natural language text understanding (33%), virtual agents (33%), deep learning (30% and knowledge graphs (25%). The capabilities with the lowest penetration at the moment are transfer learning (16%), generative adversarial networks (11%), and transformers (11%). 

However, McKinsey notes that the top use cases have remained relatively stable, with optimization of service operations taking the top spot each of the past four years. 

The most commonly adopted AI use cases, by function, are: service operations optimization (24% of respondents); creation of new AI-based products (20%); customer service analytics (19%); customer segmentation (19%); and new AI-based enhancements of products (19%). Interestingly, predictive service and intervention comes in last for use case adoption, with just 14% of respondents citing it as applicable. 

However, where organizations see value from AI is evolving. The report notes: 

In 2018, manufacturing and risk were the two functions in which the largest shares of respondents reported seeing value from AI use. Today, the biggest reported revenue effects are found in marketing and sales, product and service development, and strategy and corporate finance, and respondents report the highest cost benefits from AI in supply chain management. 

The bottom-line value realized from AI remains strong and largely consistent. About a quarter of respondents report this year that at least 5 percent of their organizations’ EBIT was attributable to AI in 2021, in line with findings from the previous two years, when we’ve also tracked this metric.

McKinsey also highlights that the level of investment in AI has increased alongside its rising adoption. Five years ago, 40 percent of respondents at organizations using AI reported more than 5 percent of their digital budgets went to AI. In 2022 this number has risen to more than 50 percent of respondents. 

Worryingly though, investment is not increasing in the level of risk mitigation organizations engage in to ‘bolster digital trust’. In other words, companies are spending more on AI tools, but aren’t increasing spend in mitigating any AI-related risks. 

Commenting on the findings, Michael Chui, Partner at McKinsey Global Institute, said: 

Over the past half decade, during which we’ve been conducting our global survey, we have seen the “AI winter” turn into an “AI spring.” However, after a period of initial exuberance, we appear to have reached a plateau, a course we’ve observed with other technologies in their early years of adoption. We might be seeing the reality sinking in at some organizations of the level of organizational change it takes to successfully embed this technology.

In our work, we’ve encountered companies that get discouraged because they went into AI thinking it would be a quick exercise, while those taking a longer view have made steady progress by transforming themselves into learning organizations that build their AI muscles over time. These companies gradually incorporate more AI capabilities and stand up increasingly more applications progressively faster and more easily thanks to lessons from past successes as well as failures. 

They not only invest more, but they also invest more wisely, with the goal of creating a veritable AI factory that enables them to incorporate more AI in more areas of the business, first in adjacent ones where some existing capabilities can be repurposed and then into entirely new ones.

The leaders

As highlighted by Chui, there is a trend emerging that AI leaders are investing more wisely in their automated tools, people and organizational structures. These ‘AI high performers’ are defined as organizations that are seeing the biggest bottom line impact from AI adoption - that is, 20 percent or more of EBIT from AI use. The proportion of respondents falling into that group has remained stead at approximately 8 percent. 

McKinsey notes that high performers are more likely to follow best practices, such as linking their AI use to business outcomes. In addition, they’re more likely to have a data architecture that is modular enough to accommodate new AI applications - so they’re not adopting AI in a rigid way that doesn’t allow for iteration. 

Interestingly, these high performers are also 1.6 times more likely than other organizations to engage nontechnical employees in creating AI applications. 

These high performers are also investing more than other organizations. The report notes: 

AI high performers are poised to continue outspending other organizations on AI efforts. Even though respondents at those leading organizations are just as likely as others to say they’ll increase investments in the future, they’re spending more than others now, meaning they’ll be increasing from a base that is a higher percentage of revenues. 

Respondents at AI high performers are nearly eight times more likely than their peers to say their organizations spend at least 20 percent of their digital-technology budgets on AI-related technologies. 

And these digital budgets make up a much larger proportion of their enterprise spend: respondents at AI high performers are over five times more likely than other respondents to report that their organizations spend more than 20 percent of their enterprise-wide revenue on digital technologies.

These factors also mean that AI performers have a better chance of recruiting AI talent. Hiring is still a challenge, but they’ve got better approaches to recruitment (attacking it from more areas) and have greater success.

Skills and diversity

McKinsey notes that all organizations report that hiring AI talent, particularly data scientists, is difficult. Software engineers (39%) and data scientists (35%) are the highest sought after skills according to respondents. But if you look at the chart below, most respondents say that hiring for AI-related roles has been difficult in the past year and hasn’t become easier over time:

In good news, when it comes to securing talent, the most popular strategy among all respondents is reskilling existing employees - with nearly half doing so. But a trend in the research suggests that high performing respondents are more likely to tap into multiple recruitment channels - such as looking at universities, technology companies, training academies, or diversity focused programs. 

However, diversity in AI teams is still lacking. McKinsey says that there is “significant room for improvement at most organizations”. The report states: 

The average share of employees on these teams at respondents’ organizations who identify as women is just 27 percent. The share is similar when looking at the average proportion of racial or ethnic minorities developing AI solutions: just 25 percent. 

What’s more, 29 percent of respondents say their organizations have no minority employees working on their AI solutions.

The report also found that there’s more work being done to improve gender diversity than ethnic diversity, with 46 percent of respondents saying their organizations have active programs to increase gender diversity within the teams that are developing AI solutions, whilst only one third say that they have programs to increase racial or ethnic diversity. 

However, the research shows a correlation between diversity and outperformance. As the report states: 

Organizations at which respondents say at least 25 percent of AI development employees identify as women are 3.2 times more likely than others to be AI high performers. Those at which at least one-quarter of AI development employees are racial or ethnic minorities are more than twice as likely to be AI high performers.

My take

Plenty to take away from this for organizations thinking about their use of AI. But the key thing is that it’s a multi-pronged approach that will lead to success. Adopting the tooling isn’t enough, teams, systems, processes and people are also essential.