AI dreams are running into AI realities as companies that have adopted the technology report uneven gains, even in areas where there is improvement, the results are less radical and dramatic than people may have initially thought. 

This is according to a recent survey from business technology consulting firm Gartner, which found teams that implement traditional AI and generative AI are not significantly more likely to report high productivity gains than teams that implement other technologies such as robotic process automation or blockchain. 

Among teams who primarily used traditional AI, 37% reported high productivity gains as a result of implementation, while gen AI-using teams fared marginally worse at 34%. In comparison, 34% of teams using other new technologies also reported high productivity gains, a level that is surprisingly not significantly different from teams using either type of AI. 

Gartner attributed this to several factors. For example, inflated expectations of AI’s capabilities lead to disillusionment. While AI can automate certain tasks and provide valuable insights, it does not automatically translate into substantial productivity improvements across the board. Additionally, measuring productivity gains can be challenging, and implementation lags often delay the realization of benefits.

Of note, though, is that these percentages are averages; some functions have actually benefited quite a lot from AI while others get very little. The biggest beneficiaries have been marketing professionals, with 59% reporting high productivity gains as a result of AI implementation, followed by supply chain specialists at 45%, and sourcing and procurement professionals at 44%. After this, those reporting high productivity gains drop dramatically: only 28% of those in manufacturing, production, quality and R&D functions report high productivity gains from AI; only 27% in the legal, risk and compliance areas; 26% in finance; and, last, IT, which only had 18% reporting high productivity gains. 

Marketing has taken well to AI, according to Gartner’s survey, because it can be used to analyze large customer datasets and pinpoint distinct segment-level buying characteristics, as well as quickly create highly targeted and personalized digital marketing content. By comparison, functions such as legal and HR teams have lots of opportunities for AI deployment, but have lagged, at least partly due to areas such as legal contract review or candidate screening processes, which require teams to invest in significant risk monitoring, governance and rework, effectively capping any time savings and productivity gains.

Similarly, while finance has a lot of opportunities for AI deployment—with most common use cases being things like intelligent process automation, error and anomaly detection, basic financial analysis and forecasting—it lags behind other functions, at least partly due to the culture of finance itself, according to Gartner. 

“Many finance leaders tend to be conservative in their AI deployment due to their high expectations for accuracy, desire to minimize data security risks, and need for auditable reporting and evidence. Coupled with AI-related data and skills gaps in finance and limited funding, most finance organizations are not yet at the point where they can roll out AI more broadly and capture big productivity gains. Only 20% of finance organizations are using AI in production, and only 6% are scaling AI to a larger group of users,” said the report. “The good news is that 66% of CFOs are more optimistic or much more optimistic about the value of AI in finance compared to a year ago.”

When it comes to the teams who do report high productivity gains because of AI, the most common positives have been significant cost savings on the enterprise level, improvements in the creation of more novel products and offerings, and significant improvements in the quality of their enterprise’s products and offerings. 

Gartner found individuals save 5.4 hours per week on average after implementing traditional AI, or 4.98 hours per week for those using generative AI. This is slightly less than what was found by a poll from business solutions provider Intapp, which said AI saves accountants about 31 hours a week; but roughly in line with the results of a Karbon survey which found AI solutions have saved accountants between 3.8 to 6.5 hours over a five-day work week. 

The Gartner poll asked what people were doing with the 5.4 hours saved per week. It found that 0.8 hours were devoted to reviewing and redoing work done by AI; 1.4 hours were devoted to taking on extra work that did not improve team outcomes; 0.8 hours were devoted to developing skills; 0.6 hours were dedicated to “reducing hours worked”; and 1.7 hours were devoted to taking on extra work that does improve team outcomes. A similar pattern emerges for the 4.98 hours saved by those who use gen AI. 

Staff inertia was named as a major factor in why AI has not been saving even more time. It noted, for instance, that 60% of finance staff have a tendency to perform manual work on processes that have been mostly or fully automated, either because they don’t trust the technology or because they have an affinity for legacy work. 

“Changes to ways of working will no doubt come with time, as workers begin to trust AI more and there is effective change management and oversight to reallocate time spend,” said the Gartner report. “Teams reporting higher productivity gains make more strategic use of this time by planning for it in advance.” 

Until that day comes, however, Gartner predicted is unlikely that we will see mass displacement of workers in the near-term future. Gartner found that, so far, the time savings from AI do not yet add up to a full-time employee’s time at the average organization. It’s likely that while AI helps an employee with singular tasks, it does not yet replace an entire employee. However, it did note that, given the uneven productivity gains from AI, this means that larger companies, departments and processes are more likely to quickly realize headcount reductions than smaller groups, making the productive reallocation of that time even more important. Overall the Gartner survey challenged widespread fears that AI is coming for people’s jobs already. 

“Anecdotal evidence abounds about AI-driven job displacement. For example, a technology CEO in a recent earnings call claimed that AI-based conversational agents enabled a 50% reduction in IT support headcount, in much the same way that word processors displaced floors’ worth of typists. To the contrary, Gartner’s AI in Finance Survey found that although 53% of surveyed finance leaders expect headcount reductions from AI, only 5% have actually made headcount reductions,” said the report. 

Leaders should instead think of AI not as a headcount reducer but as something that compresses experience in low-complexity roles and getting new workers up to speed quickly at delivering quality output. AI skills are now common among teams, but they are not a differentiating driver of productivity gains. Teams with the highest productivity gains from AI are better at reorganizing to optimize the impact of AI and taking an open and explorative approach to AI deployment. 

Gartner said that if leaders want to get the most out of AI, they need to adapt their operating model to the technology, not the other way around. Those who have seen high productivity gains adapted both internal structures as well as their team’s ways of working to take advantage of AI’s capabilities.

This includes redesigning structures and workflows to eliminate process bottlenecks and shifting time more quickly to value-added tasks. Leaders should also build AI communities that drive collaboration and knowledge sharing among users that can develop richer models than a siloed team of AI experts. Finally, they need to nurture a culture of AI acceptance through instilling an openness to learn and exploring new AI use cases without fear of AI replacing their jobs. Rather than asking, “Will AI replace us?” the mindset should change to “How can we be more effective at our jobs using AI?”

Overall, Gartner recommended that leaders set realistic expectations for productivity gains in AI investment business cases and drive manager accountability for effectively shifting their team’s time savings from AI use toward value-added activities that improve team outcomes. Beyond scaling back expectations for AI, they should also build a contingency plan for a possible increase in demand for knowledge workers. 

“Despite the excitement surrounding AI, its impact on productivity has been inconsistent, leading to what some describe as the AI productivity paradox,” said Randeep Rathindran, distinguished vice president at Gartner, speaking at its CFO & Finance Executive Conference in Sydney. “While AI has shown potential to boost productivity at the segment level, such as in call centers, broader organizational benefits have been harder to achieve. Therefore, CFOs should recalibrate expectations on how AI will truly impact worker productivity and headcount.”