Companies have deployed artificial intelligence in IT service management, recouped their investment, and yet employees complain that they have more work than before. This is exactly what a new survey by SolarWinds describes. According to the survey, 84 percent of respondents believe that artificial intelligence has met or exceeded their expectations for return on investment, but more than half—52 percent—admit that their overall workload has increased since the technology was introduced. Interestingly, only 7 percent of respondents said that the cost of implementing AI matched their original estimates.
On average, teams have been using artificial intelligence in ITSM environments for roughly sixteen months, and most of them are still more focused on keeping it running than on reaping its full benefits.
The workload has not decreased
The savings are real, and the survey quantifies them precisely. Respondents report that artificial intelligence saves them an average of 3.2 hours per week when detecting problems, three hours on end-user requests, and 2.9 hours when categorizing tickets. Altogether, that amounts to almost a full working day per week, so the result looks excellent on paper.
The catch is where the time saved has gone. Nearly half of respondents—48 percent—spend part of their day managing artificial intelligence tools and integrating them with other systems. A similar proportion, 47 percent, reviews AI outputs, while 37 percent fine-tune the model. The amount of work has therefore not decreased; it has simply shifted elsewhere.
Unexpected expenses follow a similar pattern. Forty-eight percent of companies mentioned employee training, 47 percent data cleaning, and 45 percent system fine-tuning. None of these are one-off expenses, and all of them are ongoing. As a result, 83 percent of respondents spend at least three hours per week solely on maintaining the reliable operation of artificial intelligence.
Teams are still dealing with problems
The survey also showed that most IT departments remain in incident-response mode. When asked about the greatest impact of artificial intelligence, 30 percent of respondents cited detecting issues before they affect users, while 23 percent mentioned request categorization. Only 19 percent identified preventing problems before they arise.
SolarWinds draws a recommendation from this. Companies should deploy artificial intelligence for frequent, clearly defined tasks with measurable benefits. At the same time, they should keep it close to existing workflows instead of introducing it across dozens of unrelated tools. The most common reason the technology fails is poor data quality, which must be addressed systematically as part of the overall strategy.
Brad McGinity of SolarWinds' ITSM division says that deploying artificial intelligence itself is no longer the most difficult part. According to him, the challenge now is building enough discipline within a company to ensure that the technology actually delivers results. Successful teams do not merely run a faster service desk; they operate on an entirely different model.
What a company measures makes the difference
The SolarWinds report adds one more finding that helps explain the situation. Organizations that continue to measure activity—the number of tickets resolved and similar figures—report higher workloads 2.4 times more often than more mature teams that track business outcomes. The benefits are most evident in structured and easily measurable processes, with incident resolution leading in return on investment at 26 percent.
The debate over the return on investment in artificial intelligence has occupied IT leaders for the past three years, and some companies are still struggling to resolve it. IDC research, for example, found that one in five companies is pouring money aggressively into AI without properly assessing what it gets in return. According to a recent KPMG analysis, however, returns do not have to be purely financial; companies also track the quality of work performed or the speed and accuracy of decision-making.
Source: itpro.com
