We are beginning to trust AI with high -risk tasks, including the execution of automated factories and guide military drones through hostile airspace. But when it comes to administering data centers that drive this AI revolution, human operators are much more cautious.
According to a New survey of more than 600 data centers around the world for Activity InstituteAn inspection and qualification company of data centers, only 14 percent say they would trust AI systems to change equipment configurations, even if it is trained in years of historical data. In the same survey, only one in three operators says they would trust AI systems to control the data centers equipment.
Its skepticism can be justified: despite seeing tens of billions of US dollars in AI systems, 95 percent of organizations so far lack a clear return on investment, according to a recent MIT report of generative use-AI. Advanced industries, which include factories and data centers, classified near the bottom of the list of sectors transformed by AI, if they do.
OPERATOR Trust in AI systems
Even before the impulse driven by the AI to expand data centers, it is known that data centers operators themselves are a relatively required crowd by the change that has been disappointed by the technologies of the past, he says Rose WeinschenkAn associate of research at the ATS time institute. Operators often have technical or electrical engineering environments, with training in the functioning of critical facilities; Others work on the side of the IT or network system and are also considered operators.
OPERATOR Trust in AI Declined Every Year For The Three Years Following OpenAI’s Release of Chatgpt in 2022. WHEN ASKED by Uptime If they trusted through 2024. While the public has marveled at the seemingly All-Knowing Nature of New Large Language Models, Operators to Feel This Type of ai is too limited and an predictable for use in data centers.
But now, operators seem to have entered a “careful test and validation period of different types of AI systems in certain data centers, said activity research time analyst Max Smolaks In a public web seminar of the latest survey results. To capture changing feelings, the activity time asked the operators in 2025 what requests for AI could serve as a reliable decision maker, assuming adequate past training. More than 70 percent of the operators say they would trust AI to analyze the sensor data or predict maintenance tasks for the equipment, according to the survey.
“Data centers are very, very happy to do certain things using AI, and they will never trust that AI do other things,” said Smolaks in the Web Seminar.
The unpredictability of AI in data centers
A reason why trust in AI is low for the critical control of the teams is the unpredictability of technology. Data centers are executed in “good and outdated engineering”, as the programmed logic if/then Robert Wrightthe main data data officer in Ilkari data centersA Startup of Data Centers with two centers in Colombia and Iceland. “We say that we cannot run lucky, we have to run with certainty.”
Data centers They are a complex series of systems that feed each other. You can pass a few seconds before catastrophic failures occur that result in damaged chips, lost money, angry customers or Fatal fires. In the high -risk environment of the data centers, the anonymous posters in the R/Datacenter Reddit forum that responded to a IEEE Spectrum consultation He could generally see a reason to justify the risk that AI could bring.
Distrust can also mask a underlying work insecurity. The workers of many industries are concerned that AI takes their jobs. But the 2025 activity time survey found that only one in five operators sees the AI as a means to reduce the average levels of personnel.
“Operators believe that today’s AI will not replace the personnel required to execute their facilities,” said Smolaks in the activity web seminar. “It can reach office workers, but the data center works seem to be safe from AI for now.”
But it is understandable that early race operators feel Like this technology comes for your work, says the electrical engineer Jackson Fahrneythat has worked in data centers for more than eight years. Someone of only six months at work can see an AI system as if they were told: “Here, train their replacement,” he says. Actually, he does not believe that AI replaces him or others within the data centers. However, AI has a more “sinister” presence in the workplace than automatic learning tools, which for a long time have been part of an operator’s tools game and are destined to help operators make decisions.
It could be that AI is the cherry in addition to a trend of the entire industry to reduce the number of operators within the data centers, he says Chris McLeana design consultant and construction centers.
While 60 engineers could have executed a data center in the past, now six are needed, says McLean. Less than six are also required, since more and more critical maintenance are subcontracted to specialists outside the data center. “Now you compensate all your risk with a low -cost human and a high -cost AI,” said McLean. “And I have to imagine that this scares for operators.”
That said, there are more database jobs than qualified applicants, as previously reported by Spectrum. Two thirds of the operators fight with the retention or recruitment of the personnel, according to the 2025 activity survey, similar to the responses of the surveys during the previous two years.
EFFICIENT ALGORITMOS FOR DATA CENTERS
Even so, there are useful algorithms built in decades of automatic learning research that could make the operation of the data center more efficient. The most established AI system for data centers is predictive maintenance, says Wright of Ilkari. If the readings of a particular HVAC unit are increasing faster than those of other units, for example, the system can predict when that unit should be repaired.
Other AI systems focus on optimizing cooler plants, which are, in effect, the refrigerator systems that maintain the fresh data center when driving cold water and air. The coolers represent much of the energy consumed by the data centers. Data on climatic patterns, the network load and the degradation of the equipment over time, all feed on a single AI system that is executed in hardware within the installation to optimize the total energy consumption, says Michael Bergerwho directs research and development in the Australian Energy Software Company Keep it.
But Berger rushes to notice that his Optimization software AI-AI does not control the equipment. It is executed at the top of the basic control loop and refine the parameters to use less energy while achieving the same result, he says. Berger prefers to call this automatic learning system instead of AI due to how specialized it is to the needs of a data center.
Others completely adopt the AI, both the name and technology, as well as Joe MinarikThe director of Operations in Data bankA company of Dallas headquarters with 73 data centers in the United States and the United Kingdom. He attributes his attitude admitted to AI to his 17 years working for Amazon Web Services, where the software is the king. Currently, Databank uses AI to write software, and there are plans to implement AI systems for automated input generation and monitoring, as well as monitoring and network settings for the end of the year. The AI for larger tasks, such as cooling, is tentatively scheduled by the end of 2026, subject to the time that has been training AI in sufficient data, he said.
Ai makes hallucinations: MINARIK has seen it give incorrect information and send to his team along the wrong way. “We do it, we see what happens today. But we also see that it improves and better once we give it more time,” he says.
The key is “tremendous amounts of data points” so that AI understands the system, says MINARIK. It is not different from the training of a human data centers engineer in each possible scenario that can occur within the halls of a data center.
Hyperscalers and companies data centers, whose unique client is the company that has the data center, are implementing AI at a faster rate than commercial companies such as Databank. MINARIK Listen to AI systems that execute complete networks for internal data centers.
When Databank implements AI for more significant data centers operations, it will remain with a tight strap, says MINARIK. Operators will still make final executions.
While the AI will undoubtedly change the way in which data centers are executed, MINARIK sees operators as a central part of that new future. Data centers are physical places with activity on the site. “The AI cannot leave and change a spark plug,” he says, or listens to a strange burst from a server shelf. Although MINARIK says that one day there could be sensors for some of these problems, they will still need physical human technicians to fix the team that keeps data centers into operation.
“If you want a sure job that can protect it from AI,” says Minarik, “Go to data centers.”
Of the articles of your site
Related articles on the web
#Data #Center #Trust #Operators #skeptical
