Hogue and his institution are beneficiaries of an influx of artificial-intelligence technologies that are quietly, but inexorably, transforming college campuses. The explosion of data about almost everything that happens in higher education is being fed into new software products that respond with reports, predictions, even conversational answers. As such products pop up in the classroom, via computerized teaching assistants and tutors, chatbots and self-monitoring machines are also spreading throughout the campus.

Machines are unlikely to replace most humans in college operations, but they are making those operations more efficient — potentially freeing up people for other activities.

The new technologies, corralled under terms such as “artificial intelligence,” “machine learning,” and “the internet of things,” involve ways in which machines are able to perform tasks hitherto associated with human intelligence. They can process sensory data, they can make decisions on how to act, and they can even learn. They have little in common with the autonomous, omniscient computer brains of science fiction. Certainly they lack the level of control of sinister futuristic entities like Hal 9000 (of 2001: A Space Odyssey) or Skynet (from the Terminator films). But by crunching masses of data and using advanced computing to interpret the results, these new technologies offer expediencies in areas like enrollment, advising, and facilities.

Artificial intelligence is not, experts say, a flash in the pan that will remain confined to a few campuses. “It absolutely is coming quickly” across higher education, says John F. Bernhards, associate vice president of APPA, a national association of college-facilities administrators. “There’s no question about that.”

The Austin campus is already seeing the results. Hogue, program coordinator for irrigation and water conservation, oversaw the design and installation of a new system to water the landscaping on the main campus. The system, which is networked and largely automated, not only controls and monitors sprinklers but is also connected to instruments that measure evaporation and rainfall. If a certain lawn is baking dry in the sun, it automatically gets more water. If another is still damp in the shade, it gets none. If a pipe springs a leak, the software shuts it down and alerts Hogue and his technicians that a problem needs fixing. He can control the whole system through an app on his phone.

In 2009, Austin used an average of 176 million gallons of water each year. Now the campus is down to about 35 million gallons, Hogue says, and saves about $1 million a year “because the AI allows us to identify problems and make changes very quickly.”

The artificial-intelligence revolution is rooted in advances in both data and technology. In decades past, colleges worked from paper files or isolated computer databases, accessed and interpreted by staff members. Glimpses into potential problems were infrequent and sometimes murky. Faculty advisers, for example, might check in with students only once a semester. Facilities-staff members might be alerted to the impending failure of a cooling unit only when they heard it making a funny sound.

Thanks to information-age advances, “we’re all swimming in a sea of data,” Bernhards says. There are more data available now about many aspects of college operations than humans can be expected to access, interpret, or act on effectively. But there are apps for that.

Georgia State University faced a problem of scale: It enrolls about 52,000 students, almost 60 percent of whom are eligible for Pell Grants. Leaders at the university knew that students from low-income backgrounds, many of them first-generation college attendees, benefit from individualized attention and support as well as financial aid.But Timothy M. Renick, vice president for enrollment management and student success, notes that the financial-aid office receives as many as 2,000 calls a day from students in the weeks before start of each semester. “We’re not American Express,” he says. “We don’t have a call center with 200 people in it.”

Georgia State became the first university to work with AdmitHub, a company that has developed chatbots to communicate with college students through texting.

If the software determines that there’s a high probability that it has chosen the correct answer from its database — the one at Georgia State includes the answers to the 2,000 most common queries — it responds to the student automatically. If it’s less than 95 percent certain, it refers the question to a human staff member, and the correct answer is added to the database.

When it began, in the summer of 2016, AdmitHub’s chatbot answered more than 200,000 questions from Georgia State students and helped decrease the university’s “summer melt”by 20 percent among a control group of students. Not only did it keep students from bouncing from office to office in search of answers about admissions issues, but it also alleviated embarrassment they might have about not knowing an answer or asking about how, say, a divorce in the family might change filling out financial-aid forms.

“These were students who might not have even raised these questions had it required them to go in and talk to some stranger,” Renick says. And unlike human counselors, the chatbot is ready to text back at 2 a.m.

Artificial intelligence is being applied to students’ course-planning and advising, too. Elon University had used software to help students track the courses they were taking and plot their path to graduation, but the technology had trouble integrating study abroad, internships, research, and other important co-curricular activities. “We want to do so much more than plan for four years of college,” says Rodney L. Parks, the registrar.

Elon recently started testing new adaptive planning-and-advising software created by a company called Stellic. The software was created by Sabih Bin Wasi, a recent graduate in computer science from Carnegie Mellon University, who was motivated by his own frustrations as a sophomore trying to figure out how dropping a course might affect his path to graduation. It was difficult to parse the ramifications, he says, “unless I had seven tabs open on my browser.” He envisioned “a personalized pathway platform for students where they can take control of what they want to do in upcoming semesters, but also let the administrators and advisers see that experience.”

Drawing from databases of requirements, course schedules, and students’ own data, Stellic allows them to see into their futures, scheduling courses and adapting plans as their paths develop. “When they put a course on there, it alerts them it has a prerequisite, it has a corequisite, or it’s only offered once every two years,” Parks says. At smaller colleges like Elon, which may offer fewer course sections, the program may help keep students from the perennial problem of needing one more class to graduate that they can’t actually take. (Stellic won’t replace traditional advising — Elon students still need someone to sign off each semester.)The technology can offer insights to administrators as well. Rather than send any student with more than 90 credit hours an invitation to apply for graduation, Elon will be able to generate reports showing who’s ready and who needs more time. Stellic can also feed administrators clues about scheduling courses, depending on student interest. “Looking at the metrics gives us a lot more power,” Parks says.

Turning over admissions FAQs or course planning to artificial intelligence involves marshaling data points that may be spread across a number of offices and departments. In the campus-facilities sector, however, data points now come installed in almost everything. Equipment in buildings today “shows up with integrated monitoring points that are ready to go,” says Peter Zuraw, vice president for market strategy and development at Sightlines, a company that advises colleges on facilities. “That makes it easier to contemplate connecting it to systems, and therefore more reasonable to think about instituting artificial intelligence on top of that.”

In fact, most colleges are still playing catch-up on the level of data available to them. Don Guckert, vice president for facilities management at the University of Iowa, says “buildings’ sensing capabilities have run ahead of the organizations’ ability to leverage those capabilities.”

Over the past few years, Iowa has been connecting its campus buildings to computer systems that monitor them for breakdowns and energy efficiency. That has involved a substantial learning curve. A new medical-research building had more than 23,000 data points, and Guckert and his team found that most of them were not worth monitoring. But after some adjustments, the software, devised by Microsoft for its corporate campus in Redmond, Wash., allows him “to look at a building’s operation from a system standpoint rather than a symptom standpoint,” Guckert says.Now, if there’s a hidden steam leak or a malfunctioning part in a cooling unit, it’s noted in a daily computer-generated report so repairs can be scheduled before a professor calls to complain that a classroom is too hot or cold. “It’s like the light going on in your automobile telling you that you need to see a service technician,” Guckert says.

Iowa has already converted 20 of about 100 buildings to computer monitoring, and it plans to bring about 25 more online next year. Thanks to the abundance of built-in monitoring equipment, the cost to connect a building to computer monitoring comes out to “a fraction of a percent” of the structure’s total cost, Guckert says, observing that energy savings should more than pay for it.

Asked about the prospects for artificial intelligence to spread in higher education, Bernhards, the college-facilities consultant invokes smartphone technology. At first it was a cool toy for early adopters. Twenty years later, “everyone has a phone.”

But artificial intelligence brings challenges along with its benefits. As more campus buildings are operated at least in part through such systems, facilities staffs will require more workers with computer and data skills, not just basic mechanical know-how. “There’s going to be a skills gap that we’re all worried about,” Guckert says.

Colleges that use artificial intelligence fed by masses of student and institutional data must also wrangle that data into a form usable for databases. The Georgia State chatbot was such a hit that some students who made it through their first semester with the help of the AdmitHub software wondered, “Where did the chatbot go? I still want to ask it questions,” Renick says. Creating a database of answers that can tackle any question a college student can throw at it is “a heavy lift,” he says, but the university has won a grant from Dell, the computer company, to begin expanding the chatbot to meet that need.

As the technology expands and develops, so do the possibilities. Hogue, the landscape-irrigation specialist at Texas, is using the time that he’s not spending on tweaking sprinklers to devise more innovations. He is working with colleagues on components of a system that could use microphones to listen to aging building machinery to detect frequency changes that indicate a failing part. The building’s computer system could then alert the school of engineering’s 3D printers to create a new part from a library of scanned replacements, consult the building’s class or office schedule to find a good time for a repair, and create a work order to alert a staff member to install the replacement.

Such a system is at least two years away. By that time, Hogue says, maybe autonomous vehicles will be able to deliver the parts.

Lee Gardner writes about the management of colleges and universities, higher-education marketing, and other topics. Follow him on Twitter @_lee_g, or email him at [email protected].