The past year has seen the application of gen AI in operations move past early flirtation and into a more committed relationship. But as plans for 2025 get started, including increasing discussions on the topic of agentic AI, it seems the honeymoon is over, and nothing—or at least not much—is happening. Time, perhaps, for a serious chat. And there are several reasons why it’s in all our interests for this to work out—not least because successful adoption and deployment of AI and gen AI on an industrial scale mean productivity growth.
This was a key topic of discussion at the Gartner CIO Symposium in Barcelona, where we presented our latest findings and further explored the ideas with some work. Here are the highlights.
Why increased productivity is important
Productivity growth could be a counterbalance to the asset price inflation of the past two decades, which has created about $160 trillion in “paper wealth” and even larger amounts of debt growth. Furthermore, an additional investment equivalent to 8 percent of global GDP is needed to fund the net-zero transition, along with continuously improving living standards if we are to achieve sustainable, inclusive growth.
An aging population in most advanced economies and China represents the demographic headwinds on the horizon. The ratio of global workers per person over age 65, for example, is set to shrink from 6.6 in 2022 to 3.8 in 2050. Productivity growth was not stellar before the global financial crisis, but since then, advanced economies have seen a steep decline in productivity. In Europe, that decline was even steeper than in North America, and unlike North America, Europe has not yet seen a recovery.
Europe’s complicating factors
Skills shortages and misalignments are a feature of the European productivity growth conundrum. In our future of work report, executives said they expect to retrain 32 percent of their workforce. Europe is also subject to macroeconomic uncertainties, including dependencies on China for critical raw materials for battery production. Input costs have also escalated, with electricity and gas prices two to three times more expensive in Europe than in the United States. On the flip side, Europe has the edge in producing clean energy, with approximately 60 percent low-carbon sources compared to approximately 40 percent in North America and approximately 34 percent in China.
Then there is AI technology itself. Advancements in AI are still regularly mistaken for transformation silver bullets. And even where there is promising potential, many hard yards of “traditional” business transformation and change work are needed to get initiatives to work at scale. Recent McKinsey research found that 90 percent of AI projects are stuck in the experimentation phase. The honeymoon, it seems, has bred a few misunderstandings and unrealistic expectations.
At the macro level, European companies are spending 60 to 70 percent less than their US counterparts on external AI infrastructure, software, and services and 45 to 50 percent less on internal IT. Yet it is in adoption (rather than creation) where most of the economic potential lies. While 89 percent of large companies globally have a digital and AI transformation underway, they have only captured 31 percent of the expected revenue lift and 25 percent of the expected cost savings from the effort.
It’s also important to remember that gen AI usually only makes up about 15 percent of any solution. Value comes from organizations having all the right elements in place—like domain reimagining, a robust operating model, and proprietary data—not just having access to the right technology.
The handshake that makes it happen
We estimate a 3 percent annual contribution potential to EU productivity growth through 2030 from technology, AI, and gen AI. But this contribution can only really happen if there is a partnership or “handshake” between chief investment officers (CIOs) and COOs. This is the relationship (supported by the CEO) that is crucial for maximizing the value derived from technology investments and ensuring that tech initiatives drive operational efficiency and innovation (exhibit).
Together, the CIO and COO can work out how a particular business can capture value from gen AI and what needs to change so that the business can be successful with the application and integration of AI.
Remember, just three in ten digital transformations capture the full value at stake. The reason for the shortfall is most often not the tech itself but the usual suspects of talent, culture, and change management. Our experience shows that for every $1 spent on new technology, another $3 should be invested in change management. Making progress on these elements is another reason why the CIO and COO can and must team up, supported, in turn, by a forward-facing and united C-suite.
Tactics for the long haul
Working out where to focus energies and where to scale will be important through 2025. For example, a board-led initiative to assess the value of gen AI across business functions may land on, say, five end-to-end processes to be automated and AI-assisted. A start like this provides the foundations for a primary gen AI plan and a gen AI center of excellence being established in-house.
It’s also important that any responsible gen AI steering group be cross-functional, bringing together business heads with legal and compliance, who should review and validate risk assessments for all use cases. For AI specialist teams, we see a move to more experienced, higher-value, and smaller teams and models that favor in-house expertise over outsourcing.
Converting AI initiatives into business dividends is about people, culture, and models. And as we know very well, change across these areas is never turnkey. So, as efforts in those areas are ongoing, what else will increase the likelihood of converting AI initiatives into business dividends?
One thing is to make sure your executive teams are well-versed in AI technologies and their implications so they can better lead through change. Understanding what AI is and what it isn’t, as well as what it can and can’t do, needs to be broader and deeper.
Another is to encourage safe experimentation. When you see teams exploring new AI applications as part of their day-to-day work within well-managed environments, it’s a sign that though the honeymoon is over, the marriage is still on track.