“Any teacher that can be replaced by a computer, deserves to be.”
David Thornburg

Whether as part of a talk, training session, or some other event everyone that has worked in educational technology has probably been asked the question “Are you trying to replace teachers with a computer?” Or heard the statement “You are trying to replace teachers with computers!”

Educational technologists are not trying to replace teachers. Educational technology is meant to enhance the teaching and learning experience. The goal of ed tech is to help the teacher and student so they can focus on learning efficiently. Ed tech is meant to solve problems in the learning environment so that teachers can get back to the job of guiding and interacting with the student. One of the uses of educational technology is to help with basic tasks so the teacher can focus on the things only they can do.

The core idea behind David Thornburg statement “Any teacher that can be replaced by a computer, deserves to be.” Is there are things that only a live teacher can do. Therefore, if a computer can do the things a teacher is doing, there is no point in having the teacher. Teachers should focus on the tasks that only they can do. However, like all types of technology in the classroom, we should discuss the idea of computer run education before it happens.

As a starting point for a discussion of computer run education, we should start with the questions; “what can technology do right now?” The reason for this discussion is that at some point technology is going to reach a point that from at least a financial standpoint people will consider replacing teachers with computers.

There are two places where computers can replace teachers in the face-to-face classroom and the online classroom. Let’s start with the online class. Most online classrooms are asynchronous. That means the students do not have a lot of real-time interactions with their teachers. Most of the communication is through email, text messages, and hopefully phone calls or audio conferences. While it is possible it is difficult to make connections and have real personal interactions with students in this environment.

Starting with communication can computers answer email, text, and voice chats/calls? Historically, automated computer response systems suffer from the inability to understand common spoken or written language what is called natural language. However, in 2011 IBM’s unveiled an Artificial Intelligence (AI) questions answering system (Q&A) named Watson. The systems competed on the game show Jeopardy and won, you can read about it here. Watson showed that natural language understanding in a computer had arrived. Since then Watson has only improved and is now used in many industries including finance, healthcare, and education.

Two of Watson’s tools are especially useful to education. A one-to-one tutoring tool that gives real-time tutoring to students when its covenant for them. The other is a natural language chatbot. These AI tools allows the students to ask questions and get feedback in real-time. Additionally, both tools can respond to typed and spoken language.

So a natural language Q&A system like Watson can cover the basic communication needs, usually done through email, text messaging, and voice. What about assignments and tests? All current generation learning management systems can already handle multiple choice, true/false, matching, and single word answer questions.

That leaves the short answer, discussion boards, essays, and reports. There is a software tool called General Architecture for Text Engineering (GATE) a natural language text analysis tool that uses both rule-based analysis and neural net learning. Using this tool students would be able to get feedback on written assignments as well as automated grading on final drafts.

The tools discussed here are only some of the options developed in the last 10 to 15 years. However, looking at just the tools discussed in this blog and based on the modern structure of most online classes we already have the means to replace teachers. Again our research and discussions are lagging what we can do.

Concerning face-to-face classes, the national language AI’s are not quite at the point where they can run multiple participant discussions and give lectures — assuming we except lectures out of thin air. However, natural language AI systems will probably have the ability to run multi-person discussions or give lectures within the next 5-10 years. Augmented reality, virtual reality, and projection systems will also provide natural language AI systems the ability to have a physical presence in the next 10 to 20 years.

Contrary to previous times AI systems with natural language understanding are reaching a point where at least in theory they could replace teachers. We are desperately in need of research that shows what impact AI “teachers” have on learning and the classroom. Additionally, now is the time to have real thoughtful discussions about AI tools in education not in a couple of years when government/administration starts adding them to classrooms. We need to decide not merely what we are willing to let the AI system do in education but why we are making these decisions.

Thanks for Listening To My Musings
The Teaching Cyborg