You are here

How artificial intelligence makes higher ed smarter

Technology characterized by machines acting like humans revolutionizes higher ed
University Business, February 2016

Remember IBM’s Watson, the star of Jeopardy! a few years back? The computer system learns by examining data and earlier results it provided, then formulating even better answers and results for the next problem.

Watson is probably the most well-known example of artificial intelligence, which has come out of research labs and onto college and university campuses to aid students and faculty.

“AI in general is the field of trying to design computers, software and hardware that can perform tasks we think only people can do,” says Marie desJardins, associate dean at University of Maryland, Baltimore County and a member of the Association for Computing Machinery.

Machines examine vast data fields much faster than humans, then present information for people to interpret. Studying patterns can help educators ask better questions about students’ needs and more accurately tailor learning plans.

AI’s role in the classrooms of the future

“We think that future education will be highly personalized, and that is only possible with sophisticated technology. The educator will respond to specific personalities, ambitions and live conditions, and the education delivered will be based on all those factors. The only way to do this at massive scale is through technologies like Watson.”

William Confalonieri, vice president and chief digital officer, Deakin University (Melbourne, Australia)

“We have had the classroom model for centuries. Now we have a blending of online and face-to-face that will become more challenging. We will ask where we need high-value faculty members and buildings and refocus attention there. We don’t want a fully online experience, either. Value remains in bringing people together for problem-solving, teamwork, just the right amount of mentoring. All that is a rich component that has to happen outside of technology.”

Michael King, vice president and general manager, IBM Global Education Industry

“As AI permeates education, as it should, as it will, the focus of education will shift from finding the right answers to asking the right questions. Human creativity is majestic and beautiful, but so much of it is consumed with finding answers to routine questions. ... The focus of education will shift to learning about how to ask the right questions—useful, deep, transformative questions, which help unleash the full power of human creativity.”

Ashok Goel, professor, director of the Design & Intelligence Laboratory, and co-director of the Center for Biologically Inspired Design at Georgia Tech

Chalapathy Neti, director of education transformation at IBM, says, “The big theme is the augmentation of human experts who are conducting a piece of the process. The machine becomes a symbiotic assistant to humans to help them become better, to help every teacher be the best teacher.”

AI remains in the very early stages of making education more effective, accessible and affordable—but it’s beginning to transform learning environments and campus services.

Accelerated learning

Picture an intelligent tutoring system that mimics a one-on-one tutor and can figure out what is hard for students to learn, says Kenneth R. Koedinger, professor of human/computer interaction and psychology at Carnegie Mellon University in Pittsburgh.

“AI can use data-mining techniques to identify critical skills, then use those to design better courses to better teach those skills.” The job at hand: Improve what is already available, making courses targeted to each learner.

Kent State University in Ohio has developed AI as a key part of its developmental math program. Since 2010, the school has used ALEKS (Assessment and Learning in Knowledge Spaces), created at UC Irvine and purchased by McGraw-Hill.

“We are not combining traditional classroom work with ALEKS in these classes,” says Andrew Tonge, chair of the math department. “We have gone whole hog and switched to the emporium style of teaching.”

Students learn online at computer stations in a large, monitored classroom. As they progress—with help from faculty members, a graduate assistant and peer tutors on site—the software adjusts the difficulty of math problems based on right and wrong answers. Developmental classes, which used to be full-semester courses, can be completed in half semesters now.

Stronger human collaboration

IBM has made the Watson developmental platform available to administrators and educators, and officials at Deakin University in Melbourne, Australia, have blazed a path in that regard.

Deakin started working with IBM in 2014 and within a few months a student services system was ready for use. Students can ask Watson any question related to university life, from “Where do I meet classmates?” to “I failed my exam—what should I do?”

“In the past, those questions were channeled through phone calls, offices, email and people,” says William Confalonieri, vice president and chief digital officer. “Now Watson delivers that service”—and humans are left for the really important interactions.

Every release (version three is due late 2015) has grown smarter, as Watson gets to know the individuals asking the questions, tailoring answers to each student’s own historical data and then ranking those answers.

While this may seem akin to a simple search, it is more. “Google search engine delivers a ‘syntactic search,’ without much understanding of the concepts behind the sequence of words. Watson is walking in the direction of ‘semantic search,’ where the question is decomposed to understand the intention and then search for the answer based on concepts,” says Confalonieri.

Watson searches for information relevant to Deakin, interpreting and answering with relevance to that person and that university.

While Deakin cannot share investment specifics, Confalonieri sees clear evidence of successful ROI in the increase in student satisfaction and success.

ADA compliance, reaching all students

Officials at Dallas County Community College District, with more than 77,000 students, needed a way to easily translate speech into text for online videos. Professors could upload short videos to enhance course content, but the ADA requirements necessitated closed captioning.

“We created a way for any video uploaded to be processed through the Watson engine,” says project manager Jesus Moreno, who coded the program. “It learned accents and words, and gets more accurate every time it is used.”

When a faculty member uploads a video and sends it to Watson, the in-house software strips out the images and transmits only the audio portion. Within about five minutes, transcribed text comes back and re-attaches to the video. When captioning is complete, Watson alerts the instructor, who can then make revisions as needed.

More than 90 percent of the district’s videos had fallen out of ADA compliance. The average vendor charged $2.50 to $5 per minute to transcribe video—which could have cost Dallas County more than $150,000, says Moreno.

“Watson was not only the most user friendly, but was the least expensive,” says Michael Coleman, a producer at DCCC.

IBM gives the college 1,000 free minutes of audio conversion a month. Every minute over that costs 2 cents.

With so many students relying on services tailored to various disabilities, staying in compliance means avoiding fines and, possibly, litigation.

Barb Freda is a North Carolina-based writer.