2018 Volume 21 Issue 2

EDITORIAL – Passing the Turing Test

EDITORIAL – Passing the Turing Test

Is that a Human Instructor Online?

Man’s quest to improve calculating performance dates back a long way. In the early 1830s, Charles Babbage, began work on the precursor to the modern computer. During this period, which marked the dawn of the Industrial Age, most business computations were provided in the form of tables, e.g., interest and navigational. These tables, which were almost always generated by human calculators, often contained substantial errors. These mistakes could have a significant impact on a variety of business applications including banking and insurance. Furthermore, other mistakes would often occur when setting the type for printing the tables. These two classes of mistakes, calculations and typesetting, and the overall arduous task of performing manual computations were what motivated Babbage to design the first general-purpose calculator with a printer (the so-called difference engine).

Basically, Babbage’s Difference Engine was a “souped-up” adding machine that could perform very complex calculations. Subsequently, Babbage went on to design a general-purpose computer (the analytical engine). Unfortunately, the manufacturing technology of the day was not capable of meeting his design specifications. Another 100 years would pass before the world would witness the first operational, programmable computer and it would be based on electronics, not mechanical wheels and cranks. The new innovator, another Englishman, Alan Turing, developed a seminal paper in the mid-1930s, postulating that it should be possible to reproduce human intelligence using a powerful enough computer, one that could discover new relationships based on experiential interactions. Nevertheless, Turing is most well-known as the man who cracked the German Enigma Code during World War II. Interestingly, Babbage was also a code breaker and solved the Vigenére cipher-the “Enigma Code” of his era. After the war, Turing continued to explore artificial intelligence and in the early 1950s proposed a screening process that was eventually dubbed the Turing Test. The goal was to define a criterion for a machine to be characterized as “intelligent,” meaning that a human could not distinguish it from another human through a variety of interactions.

Specifically, Turing proposed that a computing machine can be characterized as exhibiting intelligence if it can duplicate human responses under specific conditions. The original Turing Test, also referred to as the Imitation Game, calls for three physically separated terminals. One terminal is operated by a computer while the other two are operated by humans. During the test, one of the humans serves as the integrator, while the second human and the computer function as respondents. The interrogator quizzes both respondents within a certain topical area, e.g., should the country stop using nuclear energy. After a certain length of time or a specified number of entreats, the integrator is then asked to decide which respondent was human and which was the machine.

Society can only be understood through a study of the messages and the communication facilities which belong to it; and that in the future development of these messages and communication facilities, messages between man and machines, between machines and man, and between machine and machine, are destined to play an ever-increasing part. ― Norbert Wiener

Turing predicted that machines would eventually be able to pass the test. Specifically, he estimated that by the year 2000, machines would be able to fool 30 percent of human interrogators in a five-minute test, meaning that the computer must be mistaken for a human more than 30 percent of the time. There is an ongoing debate regarding whether any machine has accomplished this goal to date. However, current estimates suggest that the test will be passed routinely within the next decade. Many have argued that the Turing Test is somewhat passé since the real interest should be in designing machines that are capable of their own thought process, which could be fundamentally different than the way humans think.

You might ask what do these musings on whether a machine can think have to do with education in general and higher education in particular. As it turns out just about everything. Artificial intelligence (AI) is the new zeitgeist for enhancing student learning opportunities and outcomes for the 21st century. Furthermore, AI provides a vehicle to provide educational opportunities on a global basis, especially to underdeveloped countries, which continue to lag behind the advanced nations in educational achievement. In an academic setting the role of the instructor is changing rapidly, thanks in great measure to the Web. To that end, some forecasts suggest that human instructors will begin to be phased out within the next 10 years and replaced by machines. In higher education, where the tenured faculty have lifetime employment, this appears to be somewhat problematic, even in the online learning space. More likely, in the near term, AI-based learning systems will complement human instructors in much the same way that IBM Watson is beginning to complement physicians in the health care industry.

Robots will replace teachers by 2027. ― Anthony Seldon

No two students learn in the same way, just as no two medical patients are the same. This is why the class of AI technologies used to diagnosis and prescribe for patients can be applied to student learning. Addressing the wide variance in student backgrounds and capabilities becomes one of the major challenges for every instructor. Basically, what is needed is a customized lesson plan that resonates with every student in the class (traditional or online). Creating robotic instructors or robotic teaching assistants that can meet these demands will be challenging but could ultimately be used to solve many of our most pressing educational requirements.

For example, an AI-based learning paradigm embraces many options for delivering content to students in both individual and collaborative contexts and as such supports the transition from the three pillars of traditional higher education-fixed time, fixed location, and fixed learning pace-with a more flexible and dynamic learning environment and experience. This same approach can also be employed to identify students at risk early on and to develop interventions based on modifying the pace and level of content via artificial intelligent agents. There is an ongoing debate concerning whether robotic instructors will ever completely replace human instructors due to the complexity and range of the tasks needed to provide an effective learning environment. This debate resembles the controversy as to whether robotic physicians will replace human doctors. Perhaps a better perspective is to develop processes where both human and artificial resources can be utilized together to improve outcomes in both education and health care.

Chatbots have the potential for holding a meaningful conversation with a student, essentially making it into a tutor that can walk them through even the toughest, most subjective topics.― Bill Gates

In the educational arena, one-near term goal for AI-based tutoring systems is to approach Bloom’s two-sigma learning performance standard. Some thirty years ago, Benjamin Bloom observed that students receiving individualized tutoring tended to perform two standard deviations better than students receiving conventional classroom instruction. The task at hand now is to find technology-based methods of instruction that are scalable, which is not the case with the traditional Bloom model, and that approach the effectiveness of one-to-one human tutoring. Most likely the phasing in of AI learning support systems over the next few years will occur in a similar manner as the transition to autonomous vehicles and robotic health care providers.

AI-based learning systems not only offer the promise of improving learning performance, but they also can assist in containing the unbridled growth in college tuition, which continues to rise at rates well above inflation. Cost containment is now the byword on many college campuses because of the increasing blowback from the public on skyrocketing prices. Some specific AI cost-saving applications in this regard include replacing undergraduate teaching assistants (TAs) with TA-bots, substituting bots for financial and enrollment staff, automating the help desks, which are notorious for employee turnover, enhancing the supply chain, and reducing the use of consultants (e.g., branding firms). For example, instead of assigning ten TAs for a class of 300 undergraduates imagine that each student has their own TA customized around their specific background and learning needs. Furthermore, these TA-Bots would be available on a 24/7 basis. The above AI applications alone could result in significant cost savings for the higher education universe, which again continues to be under enormous pressure to rein ever-rising tuitions and fees.

The past fifteen years have seen considerable AI advances in education. Though quality education will always require active engagement by human teachers, AI promises to enhance education at all levels, especially by providing personalization at scale. Over the next fifteen years the use of intelligent tutors and other AI technologies to assist teachers is likely to expand significantly, as will learning based on virtual reality applications. ― Peter Stone

In this brave new learning world, will instructors be able to tell if they are engaging human students or student-bots and will human students know if they are receiving instruction from human professors? And the bots go on…

Print Friendly, PDF & Email
Author of the article
Owen P. Hall, Jr., PE, PhD
Owen P. Hall, Jr., PE, PhD
Owen P. Hall, Jr., PE, PhD is a former Corwin D. Denney Academic Chair and is a Professor of Decision Sciences at Pepperdine University’s Graziadio School of Business. He is a Julian Virtue Professor and a Rothschild Applied Research Fellow. Dr. Hall received the Harriet and Charles Luckman Distinguished Teaching Fellow in 1993, the Sloan-C Effective Teaching Practice Award in 2013, and the Howard A. White Teaching Excellence Award in 2009 and 2017. He is the vice-chair of the INFORMS University Analytics Programs Committee. Dr. Hall has more than 35 years of academic and industry experience in mobile learning technologies and business analytics.
More articles from 2018 Volume 21 Issue 2
Related Articles