By Dr. Hope Dugan March 2018
The Infinite Monkey Theorem states that a monkey hitting keys at random on a typewriter for an infinite amount of time will almost surely be able to create the complete works of William Shakespeare. However, the probability that monkeys would be able to create one of Shakespeare's masterpieces is so tiny that the statistical chance of it occurring is almost zero. Put succinctly, all things are possible, but not all things are probable.
If you take the Infinite Monkey Theorem and tweak it a bit, although possible, it is highly improbable that one could create a machine that has a powerful neural network and deeply sophisticated algorithms to replace a good teacher. According to cognitive psychologist, Aldwyn Cooper, “Despite advances in artificial intelligence, humans will always have the edge over machines when it comes to teaching. Below are five reasons why machines will not replace great teachers (2017).
Teaching is relational; not transactional
The most important gift a teacher has is the ability to see children for who they are, who they can be, and the relationship (not the transaction) between the two (Heick, 2018). Research supports that positive student interactions are an essential component of the learner experience and that creating meaningful student relationships is also considered one of the most complex and challenging responsibilities of a teacher (Feimann-Nemser & Remillard, 1996; Ria et al., 2003; Lubbers et al., 2006).
Abraham Maslow, psychologist, indicated that before students begin the intense process of learning, they must first have their basic needs met. In order for an individual to be able to learn, they need to feel a sense of belonging and have meaningful connections. Teachers rely heavily on social interaction to support their students and figure out what they need and so far, no digital system can compete with a human in this realm (Houser, 2018). Computers can provide information, they can calculate, project, and even be a facsimile for human interactions but ultimately, that is all they can do-they cannot actually be human.
There are many nuances to relationships that robots will never replicate in their relationships with humans (Morgan, 2017). Truly knowing someone is more than utilizing an algorithm to recall information about them. Time and again it has been shown that students learn from teachers they like, which is an intimacy outside the grasp of current AI. Robotics and other forms of information communication can only be used to enhance or speed up human correspondence; not establish and maintain meaningful associations.
Because the ability to build relationships is both key and difficult, being great at interpersonal experiences is requisite to being an effective teacher. If one can accept the idea that machines cannot create meaningful relationships, only enhance them, then it follows that machines cannot be great teachers.
Empathy & Trust
According to researchers at McGill University in Montreal, empathy matters. In 2012, researchers found a direct connection between empathy and learning capacity. The study noted that people who receive empathy from others, especially from an early age, develop a higher capacity to learn. When a student needs assistance, part of the need is to be heard and understood. Sometimes students, like adults, just need attention. When a teacher stops, listens, and provides that attention it is an investment in the relationship.
Bob Sornson, founder of the Early Learning Foundation states "Empathy is the heart of a great classroom culture" (Owen, 2015). Empathy allows students and teachers to understand each other and to build trusting relationships. Trust is established by consistent, good behavior from both parties. Small moments build trust among teachers and students and it is that trust that creates an environment where students feel welcome. When a teacher takes time to show empathy, students begin to trust which can lead to a meaningful relationship. Additionally, school programs that intentionally incorporate empathy have better test results. “Empathy pays off,” notes Fast Company writer, John Converse Townsend (2013). Empathy can also help reduce the damaging effects of repeated stress, which also suggests that empathy has tremendous implications for achievement, both socially and intellectually.
In order for humans to thrive, they need to feel connected. When people commiserate over a similar or shared experience, it can create a powerful bond. If you tell a friend that you are sad, the friend will most likely show concern and empathize with your pain; robots cannot genuinely consider human feelings like a person can (Morgan, 2017). If you were to share your sadness with a robot and it responded, "I am sorry for your loss, I can imagine you must be very upset" it would most likely not create that meaningful connection. Why? Because the robot does not have genuine emotions. There is no empathy, merely good programming.
Artificial intelligence may be able to replace some of the rote duties of a teacher (For example, parsing data is something a computer can do better than a human) but there are some things a computer will not be able to do…and being well, human, is one of them.
Transference
AI, after many years of frustration, has made great strides. Today Siri, Alexa, and other personal assistants can turn off and on lights, thermostats, and play music upon request. But as anyone who has given a command to one of these personal assistants only to be told, “I am sorry, I did not understand, what was that again?” or “I cannot find Justin Bieber Despacido” or “there are no devices with that name” can attest, AI still has a long way to go before it can pass the Turing test. Part of the cause of these issues is that computers cannot yet reliably transfer learning to a new situation, as humans can.
Machines do not perform well when confronted with new situations. Robots are designed to draw inferences based on pre-programmed possibilities. Personal experiences are unique to humans and cannot be transferred into machines thus ‘learning’ is a dynamic process and not a fixed set of rules that can be written as an algorithm. Human beings practice transference when we encounter something new. As a basic example, in math class we may learn to raise our hand when we have a question. When we move on to our English class, we will apply that same knowledge and raise our hand there when we have a question, despite that it is a different set of circumstances. Machines are not that sophisticated yet.
To see why modern AI is good at some things but ineffective at others, it helps to understand how deep learning works. Deep learning is, essentially, math: a statistical method where computers learn to classify patterns using neural networks. Deep learning employs an algorithm that adjusts the mathematical weights between nodes, so that an input leads to the right output. It may appear that the machine is employing ‘reasoning’ but it is not; it is following a series of if-and-then statements. Deep learning is currently the dominant technique in artificial intelligence, and by design cannot lead to an AI that abstractly reasons and generalizes about the world. Deep learning’s advances are the product of pattern recognition where neural networks memorize classes of things so they can be fairly consistent when they encounter the same situation again. But almost all the problems in human cognition aren’t classification problems. Unlike humans, AI is unable to transfer it’s prior ‘learning’ into new situations. When an AI is given a “transfer test” where it is confronted with scenarios that differ from the examples used in training, it cannot contextualize the situation and frequently breaks (Pontin, 2018).
Teaching is a dynamic profession requiring split-second decision-making. The unique demands placed on teachers make this position different from many other jobs and therefore unlikely to be a good candidate for automation. Students all learn differently, and a good teacher must attempt to deliver lessons in ways that resonates with each child. Effective teachers must be able to navigate many hurdles while satisfying rapidly changing circumstances. To be an effective teacher it is essential to be able to adapt and apply learning in meaningful ways-ways that current neural networks and deep learning are unable to consistently reproduce.
Rose Luckin, a professor at the University College London Knowledge Lab stated, “I do not believe that any robot can fulfill the wide range of tasks that a human teacher completes on a daily basis, nor do I believe that any robot will develop the vast repertoire of skills and abilities that a human teacher possesses,” (Houser, 2018). While machines can handle a variety of specific tasks, they haven’t yet come close to the artificial general intelligence (AGI) needed to nimbly respond to ever-changing situations in modern classrooms.
Inspiration
William Butler Yeats is quoted as saying, “Education is not the filling of a pail, but the lighting of a fire” meaning that teaching is really about inspiration, not information. Effective teaching focuses on the why and how, not the what, with the goal to spark imagination and to find a bridge to learners’ hearts and minds so that they are compelled to learn. A computer might be able to motivate a student, but to really inspire takes a human being. (If you need a concrete example, take a look at the rather creepy ‘inspirational’ posters and memes created by InspiroBot…)
To create the conditions where curiosity and passion ignite requires a gifted, human, teacher. Even Steve Jobs famously noted, “I’ve helped with more computers in more schools than anybody else in the world and I am absolutely convinced that is by no means the most important thing. The most important thing [to education] is a person. A person who incites your curiosity and feeds your curiosity; and machines cannot do that in the same way that people can.” One of the uniting aspects of the human experience is having an inspirational teacher. If you speak to any successful person they will often tell the story of the teacher that either incited them or inspired them to attain their current goals. Teachers are trained to inspire learners and inspiration cannot be programmed into artificial intelligence.
It does not have to be an all-or-nothing proposition
According to Webster’s, a dichotomy is defined as, a division or contrast between two things that are opposed or entirely different. Frequently, people frame decisions in education as dichotomies. For example, depth versus breadth; ‘sage-on-the-stage’ versus ‘guide-on-the-side’; or qualitative versus quantitative. As a more generic example, if you and a group were lost and one pulled out a map and said, we can go north or south (leaving out east or west) you might be inclined to assume that your only options were North and South. East and West might actually be better options, but by setting the decision up with only two options, what might have been the most important is now not even considered. Setting up dichotomies limits possibilities and when we consider how a question is framed, it influences the outcome.
When we limit our conversation to “will AI replace teachers?” we have effectively ruled out all other options. Why not leverage the best from machines and the best from humans? The thinking is not to replace people with machines but rather, for robots to function as an aid in the classroom and add value to the learner experience (Mubin & Ahmad). Maybe both can live side-by-side, harmoniously and that partnership may enhance the experience in ways that using just one or the other would not allow. For example, AI and automated systems could have collaborative roles in the education system. That would enable teachers and students to take advantage of the tech to benefit them both. We already employ computers to do a variety of things that make the classroom better like grading tests, keeping attendance records, and using technology to differentiate learning and through adaptive systems. As a caution, however, teachers should not be relegated to the role of overseers of the machines. They are leaders, coaches, guides, facilitators, and mentors. Whereas machines can motivate a learner through badges and game-based design, teachers encourage students when they struggle, and inspire them to set and reach for their goals. A computer can provide information, but a teacher can lend a hand, or an ear, and discern what each student needs to succeed. So, although technology should play a critical role in the future of education, it will not supersede that of a human teacher.
References
Cooper, A. (2017, October 2). Robot teachers won't replace us. Retrieved from Times Higher Education website: https://www.timeshighereducation.com/opinion/robot-teachers-wont-replace-us
Feiman-Nemser, S., & Rémillard, J. (1996). Perspectives on learning to teach. In F. B. Murray (Ed.), The Teacher Educator’s Handbook: building a knowledge base for the preparation of teachers (pp. 63–91). San Francisco, CA: Jossey-Bass.
For Every Child, Multiple Measures: What Parents and Educators Want From K-12 Assessments - NWEA. (2012). Retrieved from NWEA website: https://www.nwea.org/resources/every-child-multiple-measures-parents-educators-want-k-12-assessments/
Heick, T. (2018, February 12). Are schools prepared for great teachers? [Web log post]. Retrieved from https://www.teachthought.com/pedagogy/schools-prepared-great-teachers/
Houser, K. (2018, January 17). The solution to our education crisis might be AI. Retrieved from https://futurism.com/ai-teachers-education-crisis/
Jenkins, S., Williams, M., Moyer, J., George, M., & Foster, E. (2017). The Shifting Paradigm of Teaching: Personalized Learning According to Teachers. Retrieved from Knowledge Works website: http://www.knowledgeworks.org/sites/default/files/u1/teacher-conditions.pdf
Mertler, C. A. (2014). The data-driven classroom: How do I use student data to improve my instruction?
Morgan, B. (2017, August 16). 10 things robots can't do better than humans. Forbes. Retrieved from https://www.forbes.com/sites/blakemorgan/2017/08/16/10-things-robots-cant-do-better-than-humans/2/#1bccd70ef7f3
Mubin, O., & Ahmad, M. (2016, November 15). Why teachers should not fear robots taking over their jobs. Newsweek. Retrieved from http://www.newsweek.com/robots-teachers-classroom-students-wall-e-education-521442
Owen, L. (2015, November 11). Empathy in the Classroom: Why Should I Care? | Edutopia. Retrieved from https://www.edutopia.org/blog/empathy-classroom-why-should-i-care-lauren-owen
Pontin, J. (2018, February 2). GREEDY, BRITTLE, OPAQUE, AND SHALLOW: THE DOWNSIDES TO DEEP LEARNING. Wired. Retrieved from https://www.wired.com/story/greedy-brittle-opaque-and-shallow-the-downsides-to-deep-learning/
Townsend, J. C. (2013, June 6). Why We Should Teach Empathy To Improve Education (And Test Scores). Retrieved from https://www.forbes.com/sites/ashoka/2012/09/26/why-we-should-teach-empathy-to-improve-education
Comments