Adapted from GBSN Blog
Education as a sector has proven stubbornly resistant to change. Despite significant progress in technology, classroom instruction remains unchanged from decades ago. While students have rapidly adopted social media and online collaboration, they do not fully utilize learning tools in most courses. While the consumerization of other slow-to-change sectors such as healthcare is in full swing, educational technology systems remain cumbersome to use and are far from the ease of use and embedded customer focus seen in online leaders such as Amazon and Netflix. The big data and analytics revolution is sweeping multiple sectors, yet education operates in an environment characterized by poor data and the rare use of analytical tools. It is no surprise that educational institutions today are under pressure to both improve the effectiveness outcomes and to provide more personalized learning delivery in a cost-effective manner.
As part of my day job as a researcher and teacher, digital transformation is my passion. I do research in how the digital revolution is bringing sweeping change in industry after industry. I teach courses on digital transformation to Executives and MBA students at Cornell discussing case studies about how corporate digital leaders are mastering real-time data and integrating technology into every aspect of their business. Yet each time I step into the classroom, I am acutely aware of how little I know about the backgrounds and learning histories each student in front of me and how this constrains me from being able to customize my questions and learning approaches to each student in the class. In a typical MBA class I have 45 to 60 students in front of me and I am unable to keep track of how each student performed on various content areas in prior classes. Even when I was (previously) the Dean of the school, I had no real time overview of the learning performance of students and the classroom effectiveness of faculty members. The best that I could get was a summary of the course evaluations of faculty members and the grade point averages of students a few weeks after the semester had finished. It was all too little data and received too late for taking any corrective action on time. It would not be an exaggeration if I said that most faculty and educational administrators manage their courses and institutions largely devoid of any real-time performance data.
Enter Artificial Intelligence (AI) in this scenario. AI is today probably the most hyped up set of tools on the technology frontier. It is important to note that AI is not new per se. As a field, it has been with us for almost as long as we have had computers. Since the dawn of modern day computing in the middle of the last century, the dream of building intelligent machines that could emulate the human mind has existed. In fact, when I did my PhD in Computer Science at UC Berkeley in the late 80s, I choose to do it in the field of AI. Strangely despite the current hype, no new fundamental breakthroughs have been witnessed in AI in recent years. And progress in AI was painfully slow for several decades till about 2011 when IBM demonstrated that its Watson system could convincingly beat the human champion in the TV game Jeopardy. The major changes in recent years have been the explosion in data and the abundance of low-cost computing power. This combination has proven very potent and we have seen several rapid advancements in AI systems in recent years which have approached or beaten human experts in a range of fields including games (Go) and medicine (radiology) and have approached meaningful levels of creativity in domains such as music and design.
So what can we hope from AI in education? At one level, AI has the potential to transform education. AI systems, properly deployed can capture learning data, recognize patterns, synthesize insights and feed it through appropriate intelligent (possibly conversational) interfaces to support individualized learning for students and help faculty to achieve higher levels of learning effectiveness. AI can improve the overall learning effectiveness of the educational institution by providing real-time analysis and timely action prompts to business school deans and program administrators. All the purported benefits of AI can be obtained provided two important conditions are satisfied on the assumption that raw computing power is abundant and easily obtained (either directly or through cloud-based services): appropriate data exists and the necessary algorithmic models are deployed. Both are high hurdles to overcome.
As previously mentioned, most educational institutions are largely operating in a severely data constrained environment. The one exception to this are the online courses providers such as Coursera which are capturing voluminous amounts of data about each and every click of their millions of learners. However this is in stark contrast to most physical business schools or universities where data capture is neither sufficient nor real time. Adding to this, inadequate research exists into learning models and how people best achieve their desired learning outcomes. Without an appropriate underlying theory, it is very difficult to build algorithms that can effectively guide students and actions towards effective learning approaches. Note that while building appropriate models is hard in many domains, many digital leaders have invested significant resources (years of time of multiple research teams, all accompanied by generous computational and financial outlays) into building their algorithmic models and even leaders such as Amazon and Netflix continue to keep investing in fine tuning their models based on continuous customer feedback. Educational institutions are years behind in both understanding how individualized learning strategies work and in making the investments in AI and technology needed to be successful.
So while I do not expect AI to transform education anytime soon, we cannot be pessimistic and do nothing. Business school deans have to exert leadership and get their organizations started on the path of experimentation with (albeit) limited AI systems and seeing how they can best integrate AI with their students and faculty. One should not forget that even leaders such as Google and Amazon started down the path of integrating AI within their organizations less than a decade ago. While they benefited from their significant in-house technical expertise and bench strength in AI researchers, integrating AI into the organization continues even today as a high priority process for them. It is going to be a long path of change for educational institutions, but do we really have a choice? The process will get easier as many AI tools get commercialized as plug and play modules on cloud-based services (such as modules to recognized patterns). However, business school leaders have to start today and start exploring how to best leverage AI for a better future of learning in our institutions.
Author: Soumitra Dutta
Soumitra Dutta is a Professor of Management at Cornell University and the Chair of the Board of Directors for GBSN. Previously he was the Founding Dean of the SC Johnson College of Business at Cornell and Chair of AACSB Intl. He also co-chairs the Global Future Council on innovation ecosystems for the World Economic Forum.