AI applications, testing part of future envisioned for DSU

TELEPRESENCE ROBOTS are currently in use at Dakota State University. In the future, AI technology could be incorporated into these devices so they could do a variety of tasks currently done by the individual using them, such as self-navigation and recognizing facial expressions.

Hardware. Software. Applications. Cybersecurity. Artificial Intelligence (AI).

As the field of computer science has grown, a new vocabulary has slipped into common usage. What was once science fiction has become reality, and with that new reality has come not only a new vocabulary but also a need for expertise.

Dakota State University in Madison has been on the forefront in providing that expertise, developing new programs as fields emerge. This year, two new degree programs will be offered -- a Bachelor of Science in AI and a Bachelor of Science in AI for Organizations.

"I think having both of those bachelor degree programs is very powerful for the future that we're envisioning," said DSU President Jose-Marie Griffiths.

The university has offered a minor in AI, but determined as part of strategic planning in 2017 to grow that minor into a major.

"We built our faculty, and we are now ready to go forward with a major," Griffiths indicated. "The major has some overlap with our computer science courses, because you can't do artificial intelligence without having those fundamental capabilities."

The AI major in the Beacom College of Computer and Cyber Sciences will focus on developing AI applications. The AI in Organizations in the College of Business and Information Systems will focus on using AI applications.

"Artificial intelligence is fundamentally dependent on data, because the intelligence -- whatever you build, your algorithm, your neural net -- has to operate on data to learn," Griffiths explained.

With data analytics and health informatics in the College of Business and Information Systems, placing the applications piece in that college made sense. She emphasized this new field of study enhances other programs and does not detract in any way from the university's cybersecurity program. Rather, Griffiths indicated, it builds on the same core of expertise.

Similarly, she does not believe the new specializations approved by the South Dakota Board of Regents for the South Dakota School of Mines and Technology will detract from the DSU programs. With the specializations in cybersecurity and in artificial intelligence and machine learning, a handful of courses are being added to Mines curriculum which will help graduates to be more qualified in their field.

"Engineers need to know something about cybersecurity," Griffiths said. "There's a hardware component to cybersecurity, which is more engineering, electrical engineering and computer engineering. We deal with the software side. We don't do the engineering piece."

In speaking about the future envisioned for DSU, Griffiths was alluding to research in areas of application, such as AI in cybersecurity, AI in health care and AI in agriculture. This includes not only data but also the sensors needed to acquire the data and the ethics for decision-making.

This is especially true in the third of three core areas in which AI may be used, Griffiths indicated. The first area is analysis.

"Where artificial intelligence can really help is in analyzing huge volumes of data that humans wouldn't be able to do and perhaps identify patterns that people can't," she said.

People simply cannot retain the amount of information that computers can retain, nor review data as quickly. Cures for diseases may be found through the use of AI analysis in reviewing medical literature.

The second area is augmentation, where AI is used to provide people with information so they can make decisions.

The final area is autonomy.

"You have machines and pieces of software that go and do things and don't need a human to tell them what to do. The design of those is going to be very critical for ethical reasons, making sure you design for different populations that you are going to serve," Griffiths indicated.

She pointed out that facial recognition software currently in use was "optimized on the white male face." As a result, the software is not as good on either the female face or the faces of individuals of other races.

"The data that you have to train the intelligence determines its ability. If you're not training on the range of faces they'll be looking at, they're not going to very good, very discerning," Griffiths stated.

Because of the ethical considerations inherent in using AI with different populations, she would like to see DSU become involved in TEVV -- testing, evaluation, validation and verification.

"It's not as sexy as doing the actual design or the actual use, but it's going to be essential. Someone has to do it, and I would love to see DSU do something in that area," Griffiths indicated.