Can Machine Learning Bridge the Employability Gap and Make Higher Education More Relevant?

Last year, an expert panel at OEB18 debated Bryan Caplan’s book, “The Case Against Education”. Amongst other things, Caplan argued that higher education was largely a waste of money and more specifically does little to prepare students to be productive members of the workforce. In his words: “we have to admit academic success is a great way to get a good job, but a poor way to learn how to do a good job.” While universities can debate their purpose and intellectual pursuits as an end in themselves, one also has to acknowledge some truth in Caplan’s arguments. A labor force with ‘wrong skills’, as per Allen and van der Walden (2001), represents a significant employment problem, that frustrates employees and job seekers, and that requires costly retraining by employers. We have a significant skills imbalance worldwide and higher education institutions are part of the problem.

How big is this problem? Really big. In 2017, the OECD estimate as much as 80 million workers in European countries are mismatched by qualifications (OECD, 2017). For example, on average across the OECD, over 35% of workers are mismatched by qualifications, both over-qualified (17%) and under-qualified (19%) (OECD, 2018). In Science, Agriculture, the Humanities and Arts, not only do most graduates struggle to find employment but they are more often than not employed in a position unrelated to their specialization (OECD, 2018). Indeed, a survey of 45,000 European entrepreneurs cite two related factors as the labor costs (42.3%) and lack of skilled labor (41.7%) as their two biggest challenges in the short term (Eurochambres, 2019).

The impact can be severe. From the demand side, unfilled openings and delays in filling vacancies can slow down innovation adoption, delay production, reduce productivity, increased costs (related to retraining), and impact competitiveness. From the supply side, skill surpluses can result in frustrated employees and job seekers, reduced employment or lower wages. Despite our increased investments in education, our global labor pool often has the wrong skills, lags demand or is in the wrong place. At the same time, higher education and education systems, more generally, are not adapting quickly enough to meet the existing or future needs of the labor market, particularly where they are being shaped and reshaped by technology advancements and digital transformation. In this respect, one has to agree with Caplan, the Academy is not set up to rapidly change curriculum and academics may not wish to invest in their own reskilling, on an ongoing basis, to be able to teach a new rapidly evolving curriculum. Academics are not immune to change but they may be resistant. Curriculum reform takes time; the one commodity we don’t have in this case. To make matters worse, we need continuous curriculum reform to accommodate an ever-evolving labor market.

Technology may be the solution.

Advances in machine learning are transforming labor matching; the task of matching potential candidate to given job. Parsing, matching and ranking curriculum vitae against job advertisements is a well-defined problem space with industry-hardened solutions from companies such as, Sovren and Textkernel. Such technologies are transforming the recruitment space. Can the same approaches help higher education address the employability gap? We think so.

At Dublin City University, we are working with to design a Continuous Curriculum Improvement System (CCIS). CCIS is designed to classify and compare:

  • Employer requirements as evidenced by job advertisements or generic job descriptions;
  • Candidate (student) profiles as evidenced by resumes;
  • Professional body requirements as evidenced by individual membership and qualification accreditation requirements; and,
  • Curriculum and course content as evidenced by course descriptions.

CCIS is a combination of machine learning techniques to identify misalignments automatically and quickly, and thus can be used to improve the quality and reduce the cost of not only curriculum improvement but graduate recruitment and training, and professional accreditation of courses. Furthermore, students and graduates could use CCIS to identify suitable jobs, education requirements, professional body membership eligibility, and suitability of courses and courses for specific jobs.

Skill shortages, skill gaps and qualification mismatches are significant threats to economic competitiveness throughout the world. Higher education providers face increasing pressure to be responsive and relevant to the needs of employers worldwide. Failure to respond appropriately to changing requirements in the workplace, may jeopardize the reputation of individual courses and institutions and by inference, funding from industry, government, and students who may be attracted to courses and organizations who are better aligned to their needs and market requirements. The first challenge is to accurately identify the mismatches on an ongoing basis. We think CCIS can do this sufficiently well over time.  The second challenge, and by far the more difficult challenge, is to implement the changes to curriculum.

And this is the nub of the matter. Caplan’s pessimism in “The Case Against Education” is couched in the oft-cited cliché – people are resistant to change. That, indeed, may be the case. However, people here may be only one part of the solution. By understanding the mismatches between the curriculum, job requirements, and a student’s existing knowledge, we can create a personalized learning path that does not necessarily require academic retraining or hiring new faculty. It could equally be met by e-learning or technology-based solutions. Producing employable graduates is and should not be the primary purpose of higher education but it should be given significantly more tangible emphasis than it is today. If higher education institutions truly want to act in the best interests of their students and society then a change of mindset is needed. But technology can remove the heavy lifting and make change easier. To quote Cardinal Newman, “To live is to change, and to change often is to become more perfect.”

Written by, Theo Lynn and Andreea Wade

Theo will be presenting in the panel on The Future Is Bright – On AI, Curricula and Skills Gaps.

Leave a Reply

Your email address will not be published.