Michael A. Gennert, Professor of Computer Science, Worcester Polytechnic Institute and Former Head of WPI’s Robotics Engineering Program
Artificial Intelligence (AI) has recently and rapidly moved from a niche in Computer Science research into the education and business mainstream. Universities are producing increasing numbers of Ph.D.s in AI and closely related fields, including Machine Learning, Robotics, and Computer Vision.
In fact, industry leaders are hiring many of these graduates for product development as well as research roles, competing with universities that are rolling out degrees that include and feature AI. As AI makes a foothold in more areas of our lives, AI-enabled products and services will be commonplace, challenging current educational and business models.
The role of educational institutions
Educational institutions need to adapt to the mainstreaming of AI in two ways. First, they need to think more like businesses, not academics. That is, they need to adopt best business practices, especially when it comes to understanding and serving markets, and also adopting new technologies such as AI.
Second, and as a consequence of the first, they need to offer programs (degrees, concentrations, certificates) to meet the rising demand for professionals with AI competence.
AI has already begun to make inroads into business practices. Google search and maps, recommender systems, and LinkedIn connection suggestions are just a few examples of AI in action every day that serve both businesses and individuals. However, businesses in particular, can benefit from AI in systems such as Intelligent Customer Relationship Management (ICRM) and Intelligent Supply Chain Management (ISCM). These systems increase functionality by uniting intelligent big data analytics to existing CRM and SCM technologies. Universities that adopt these technologies will gain strategic advantages over those that don’t.
Robotics is another area experiencing rapid progress. Warehouse and fulfillment center robots are, by now, commonplace, enabling faster turn-around time and better inventory control, especially coupled with ICRM. However, it is easy to underestimate the potential of robotics to replace human workers, especially in the short-term. Consider that most human tasks require some combination of perception, mobility, manipulation, and judgment. Anyone of these can be automated today with Computer Vision, Vehicles, Grippers, and Data Analytics, respectively. However, integrating any two of these technologies is devilishly difficult; any three or all four impossible. At least so far. The farthest along is autonomous vehicles, requiring perception, mobility, and judgment. Even then, deployment is between 2 and 20 years away, depending on to whom one listens. But in limited domains, such as autonomous shuttles along fixed campus routes, early adopters and experimenters are making inroads.
"As AI makes a foothold in more areas of our lives, AI-enabled products and services will be commonplace, challenging current educational and business models"
A Closer look: AI by the Numbers
Let’s look next at Ph.D.s in AI and Robotics. The Computing Research Association conducts an annual survey—the Taulbee Survey—of student enrollment, degree production, and employment in computer science, computer, engineering, and information. The survey originated in 1974; online data extends from 2002-present. Table 1 summarizes Ph.D.’s in AI and Robotics in the 2002 and 2018 surveys. (The single category of AI/Robotics is now two categories of AI/Machine Learning and Robotics/Vision, reflecting the growth and differentiation of these sub-disciplines. We recombine AI/ML and Robotics/Vision here.
The bottom line? Ph.D. production in AI and Robotics has increased 2.5 times from 2002-2018, with the overwhelming majority of the additional Ph.D.s going to non-academic (industry, government, self-employed) positions. That isn’t even counting the additional 130 Ph.D.s in Informatics/Information Science/Information Systems in 2018, many of whom work in data science and data analytics, for which no comparable specialties (Databases/Information Systems is too broad) in 2002.
Undergraduate computing degrees generally have no such recognized specializations, although some programs offer concentration, focus areas, threads and minors in addition to degree programs. However, there has been a clear trend to add majors, such as Data Science, and courses, such as Data Analytics, AI applied to specific domains, advanced courses in AI/ML/Robotics, and various kinds of Informatics. Without a doubt, opportunities abound, as it is an exciting time for educational innovation to meet real needs.