Artificial Intelligence and the Future of K-12 Schools: Part 5: More about the Future of Education in a World of AI

Current research suggests that it has taken about six million years for a particular type of ape to evolve into what we now call prehumans, and for these prehumans to evolve into Homo Sapiens. Evidently it took more than another hundred thousand years for the first Homo Sapiens to evolve into Homo Sapiens with the physical and cognitive capabilities to produce and learn oral languages.

Our oral communication capabilities now are “built in” but require years of education and practice for children to actually gain the skills in language that we adults want them to have.

Contrast this with reading and writing. Evolution provided Homo Sapiens with the cognitive capabilities both to invent and to learn to use reading and writing. But it took until about 5,600 years ago for this invention to occur. I like to think of reading and writing as a type of artificial intelligence (AI), one we have the ability to learn to use. Fluency in reading and writing (that is, in learning to do reading and writing at the level defined by contemporary standards) requires a number of years of instruction, study, and practice.

So, in a nutshell, we have the idea of humans developing reading and writing as a tool that uses and aids their physical and cognitive capabilities, a tool that takes years of study and practice to master. In relatively recent years we Homo Sapiens have developed a very wide range of such tools.

Each tool requires some learning on the part of the person wanting to make use of that tool. A great many of us have the capabilities to learn to drive a car, and we spend the time needed to gain a driver’s license. Many of us also have the capabilities to learn to fly an airplane safely, but in the United Sates less than a half of one percent do so. The percentage with the innate abilities and the drive to become commercial airline pilots is much smaller than this. I find it interesting that the AI-based flight control systems in modern airplanes are able to safely fly such planes.

The basic goals of school-based education were established long before AI-using computers began to be produced. The capabilities of current AI-based computer systems affect most (perhaps all) of the goals of our schools, but the content and processes of schooling have not kept up with these rapidly increasing computer capabilities. This newsletter addresses some of what I and many others believe our schools need to be doing about this situation.

Let’s begin by looking back many thousands of years. We humans have both physical and cognitive abilities. They served our ancestors well over the millions of years leading up to the first Homo Sapiens beginning 200,000 or more years ago. The physical and cognitive capabilities developed by Homo Sapiens have helped us to survive and prosper, while all of the humanoids coming before us are now extinct.

Prehumans were developing and using tools several million years before the first Homo Sapiens were born. Homo Sapiens were better at developing and using tools than their predecessors. Now, most of us live in regions of the world where we enjoy the use of a wide variety of machines, many that have physical capabilities far exceeding our own. Most of us are highly dependent on these machine capabilities. But, we certainly do not spent time worrying about the possibility of such machines taking over the world and enslaving humans.

Over the past 80 years, humans have been developing computer-based artificial intelligence (AI), a type of brain tool. Such AI-using computers are often imbedded into the types of machines that have long augmented our physical capabilities. If the resulting machine has mobility, we typically call it a robot. Such a robot may well have a combination of physical and cognitive abilities that exceed those of humans in a variety of areas. Indeed, since such robots can access the Web, there is no particular reason why such a robot might not have the full range of AI capabilities currently being built into non-mobile computer systems.

We humans have been able to live and work successfully with machines that exceed our physical capabilities. Now we are engaged in the task of learning to live and work with artificially intelligent machines that have a steadily increasing number of intelligent-like capabilities far exceeding our human cognitive capabilities. Such AI makes it possible for me to write and publish this newsletter, and for you to access it.

Here are two important education questions:

1. How have these AI-using computers changed our world and our lives, and what major changes will they produce in the future?
2. What should we be doing in raising and educating our children that will help to prepare them for their adult lives during which such AI-based computers will continue to become more and more capable?

Some futurists talk about the Singularity, a happening (an event) when computers become more intelligent than humans. Predictions about when this might occur range from sometime in the next ten years, to sometime in our current century, or to never. Personally, I am not able to comprehend the possibility of a time when computers become much more intelligent in all areas of human intelligence. So, I do not spend time worrying about the Singularity.

I have found it helpful to learn more about human intelligences as I strive to understand possible roles of AI in education. I became interested in Howard Gardner’s theory of multiple intelligences sometime in 1984 after reading his Reflections on Multiple Intelligences: Myths and Messages published in 1983. Gardner has since devoted much of his academic career to expanding on and elucidating his ideas (Gardner, 1995, link).

In brief summary, his work has focused on nine types of human intelligences:

• Bodily-Kinesthetic Intelligence: The ability to control one’s body movements and to handle objects skillfully.
• Existential Intelligence: The sensitivity and capacity to tackle deep questions about human existence, such as the meaning of life, why we die and how we got here.
• Interpersonal Intelligence: The capacity to detect and respond appropriately to the moods, motivations and desires of others.
• Intrapersonal Intelligence: The capacity to be self-aware and in tune with inner feelings, values, beliefs, and thinking processes.
• Mathematical-Logical Intelligence: The ability to think conceptually and abstractly, and the capacity to discern logical or numerical patterns.
• Musical Intelligence: The ability to produce and appreciate rhythm, pitch and timbre.
• Naturalist Intelligence: The ability to recognize and categorize plants, animals and other objects in nature.
• Verbal-Linguistic Intelligence: Well-developed verbal skills and sensitivity to the sounds, meanings and rhythms of words.
• Visual-Spatial Intelligence: The capacity to think in images and pictures, to visualize accurately and abstractly.

For each of these types of intelligences, researchers can examine the human brain to possibly locate specific regions that are linked to a particular type of intelligence. Also, for each type of intelligence we can study how it is used by people, and how different types of upbringing and schooling affect use of that intelligence. We can explore how our culture, schooling, and other aspects of our lives help to train students so they can make more effective use of a particular intelligence.

A quick perusal of Gardner’s list identifies Bodily-Kinesthetic, Mathematical-Logical, Musical, and Verbal-Linguistic intelligences as four areas being addressed by schools providing specific instruction in each. Interpersonal intelligence is addressed by the need for students to routinely work together as they interact throughout the school day. Naturalist intelligence is addressed in biology and other science courses. Visual-Spatial intelligence is addressed in art and math instruction, and in the use of multimedia as aids to learning.

The extent to which schools address Existential and Intrapersonal intelligences varies considerably from school to school. For example, a school sponsored by a particular religious group will likely address certain aspects of existential issues in ways that are consistent with the beliefs of the religious group. All schools and all teachers include some emphasis on knowing yourself, but relatively few schools provide specific courses in this area.

In brief summary, schools provide an environment in which students make use of a wide range of human intelligences. The use of any one specific intelligence improves one’s ability to use that intelligence. However, an important aspect of schooling is that instruction typically requires students to use many of their different intelligences. That is, the problems and tasks that students gain skill in addressing in any course they study requires the use of multiple intelligences.

A number of people have suggested additions to Gardner’s list of human intelligences. Just for the fun of it, as I read the list I thought about what I might want to add. My brain came up with the response of empathetic intelligence. This led me to ask the question, “Is empathy a type of intelligence?” I went to the Web, and Google provided me with about 47 million results to my enquiry. The answer is certainly yes.

My further Web research led me to a Wikipedia article titled Artificial Empathy (Wikipedia, 2021, link). Quoting from this article:

A broader definition of artificial empathy is “the ability of nonhuman models to predict a person’s internal state (e.g., cognitive, affective, physical) given the signals (s)he emits (e.g., facial expression, voice, gesture) or to predict a person’s reaction (including, but not limited to internal states) when he or she is exposed to a given set of stimuli (e.g., facial expression, voice, gesture, graphics, music, etc.)”.

An empathetic robot would have the knowledge and skills to take appropriate actions based on such perceived information. This is an active area of research, and significant progress is occurring. I was surprised by this. It makes me realize that we indeed are moving toward a time when the intelligence of a computer may well exceed that of a person over a very broad range of human intelligences and capabilities.

Consider my personal thinking and the Web use that I just described. I, a human being, had posed a question that interested me. I made use of the computer facilities in my home and others located in various places to seek answers to my question.

My point is that my curiosity led me to I pose an interesting question about empathy, one that I am sure other humans may have posed as well. However, I doubt that any computer has independently posed this or a similar question, and then set out on its own to find an answer. In that regard, I have and I use a cognitive ability that is quite different from that of the best of current AI-using computers.

This led me ask another question, “Is curiosity a type of intelligence?” So, I once again went to the Web where I found two general answers. One is that some researchers consider curiosity to be a type of intelligence. The other is that curiosity is considered to be part of every intelligence.

My Google search of the term artificial curiosity produced about 24 million results. Needless to say, I did not read all of these papers. But I did spend enough time browsing to be able to satisfy my curiosity. (If I said that to a class I was teaching, I would expect a chuckle, if not outright laughter. Did you chuckle? Hmm. Is humor a type of intelligence?)

I found an article about curiosity written by Jeremy Dean that I want to share with you, A Curious Sign that Your IQ Is High (Dean, 2011, link). Here are several short quotes from the article:

People who are curious ask lots of questions, look for surprises, seek out sensations and make time to search out new ideas.

The results showed that students who are curious do better in their school work.

Taken together, conscientiousness and curiosity were just as important as intelligence in students’ performance.

In brief summary, curiosity is an innate human characteristic that appears to be a component of each type of human intelligence. Whether curiosity is classified as a type of intelligence is much less important than the fact that it is a component of many different types of intelligences. Schools can provide an environment that encourages curiosity.

Consider Go, a popular board game long regarded as being beyond the AI capabilities of computers. In recent years, a computer program was developed that defeated one of the world’s best Go players. A couple of years later a different computer program taught itself to play Go and defeated the original program 100 times in a row in a match between the two computers (Hutson, 10/18/2017, link). Notice my emphasis on taught itself. It was not necessary for human programmers to write a detailed step-by-step set of instructions for the computer that would direct it in playing so well. To me, this was an amazing achievement and it ushered in a major step forward in AI.

For a simpler example, you  probably know how to play Tic Tac Toe on a 3 x 3 board where one wins by getting three in a row. Suppose a child were provided with a somewhat more detailed explanation of the rules of the game and how to tell a game has ended by someone winning or in a draw. I can imagine a child learning to play the game by playing against himself or herself, and developing a strategy that never loses. The game is no longer much fun after such an achievement, but this certainly illustrates learning without the direct aid of a teacher.

This success in a computer teaching itself  to play and win the game of Go was a major breakthrough in AI, yet it is only the tip of the iceberg of AI’s growing capabilities. For example, consider the task of learning an oral language and also to read and write in that language. Humans have this capability. Do you think a computer could teach itself these skills? It now seems that this may well be possible. Researchers are working on this task because success in the area would help a great deal in the efforts to create computer systems that are equally as good or better than our human capabilities at translating and communicating in human languages.

Next, let’s explore an education situation in which AI is making a major contribution. People have been developing, testing, and using computer-assisted learning (CAL) for about 60 years (Suppes, 1988, link). CAL systems making extensive use of AI are now in routine use in our schools .

It has turned out that developing high quality effective CAL has proven to be a very difficult task. One challenge is to decide on measurable goals—and determining how to measure success in achieving these goals.

Suppose we look carefully at the learning outcomes typically achieved by a human teacher working with 25 or so students. We decide to use test scores as our measure of success, with an emphasis on tests developed by persons other than the teacher. We now have substantial research-based evidence that CAL has become more effective than are typical human classroom teachers in a variety of teaching tasks in which success for students can be measured by their ability to perform well on tests.

I find it to be somewhat amusing to think about how we are now using AI-based computers to help students pass tests that such computers themselves can pass. This is somewhat like teaching humans to compete with computers in areas in which computerized machinery is much more capable than humans. This situation tells me that schooling based on preparing students to pass the types of tests that have traditionally been used to measure the level of success being achieved by these students is rather obsolete. Such tests are no longer authentic.

However, we know that schooling and education are far more than just achieving high scores on tests. For a long time, the needed computer facilities were quite expensive, but this is no longer true today when such computers have become cost effective.

Many years of research and development, accompanied by substantial decreases in cost and by improvements in the facilities that can be made available to students, have led to significantly increased amounts of the use of CAL. The Covid-19 pandemic has hastened this adoption and use of CAL.

This home use of CAL has not proven to be particularly effective for a variety of reasons. One is that students have not been educated to take responsibility for their own learning in such an environment. A second is that students are very accustomed to the set of controls and rules provided in a classroom and school. A typical home does not provide such a structure. This is another indication that good schooling is far more than just having students achieving high test scores.

A much higher goal for CAL is to develop systems that are as good as a human tutor. Such a tutor has empathy, and so is good at connecting with a student as a human being to help foster learning. We are a long way from achieving such CAL!

AI makes it necessary for us to carefully reconsider some of the widely accepted goals of education. Let’s use Basic Skills as an example. Quoting from the discussion of 14 goals of education in a recent IAE Newsletter.

G6. Basic Skills: All students gain a working knowledge of speaking and listening, observing (including visual literacy), reading and writing, mathematics, logic, and storing, retrieving, and communicating information. All students learn to solve problems, accomplish tasks, deal with novel situations, and carry out other higher-order cognitive activities that make use of these basic skills.

The steadily improving capabilities and availability of computers leads to the question of what now constitutes a basic skill. Do we still consider reading and writing cursive to be a basic skill? Some schools no longer consider it to be one and are dropping it from their curriculum. Is using a word processor that includes a spelling and grammar checker now a basic skill? How about reading and writing multimedia documents? How about being skilled in learning from CAL systems? These are some of the provocative questions that need to be addressed by our current schools.

A still larger issue is developing and implementing a curriculum when AI can solve many of the problems and accomplish many of the tasks that we currently teach students to accomplish without computers. Let’s use math as an example. Two goals of math education are:

1. To help students develop number sense and more general math sense. We want students to have good insights into the capabilities and limitations of math to help represent and solve a myriad of problems that they encounter in their everyday lives.
2. To develop skills used in solving math problems. Here, we now need to consider mental skills, paper and pencil skills, calculator-based tools, and computer-based skills.

We have long had AI-based computer systems that can solve the full range of the types of math problems students currently study in their math courses up through the first two years or so of college. Software to accomplish these problem-solving tasks is available free on the Web, with Wolfram Alpha being an excellent example (Wolfram Alpha, 2021, link).

The challenge that our educational systems now face is to achieve an appropriate balance between math education school time and effort spent on the two general goals listed above. Computers make it possible to spend more time on the first of the two goals, and less time on the second. Within the second category, how much emphasis should be placed on each of the four commonly available approaches?

This same type of analysis can be done for the entire Prek-12 curriculum. In each subject area currently being taught, we need to help students learn to make effective use of computers to help solve the problems and accomplish the tasks that make up the discipline. We also need to help students gain the knowledge and skills to be responsible, productive adults in a world that is increasingly becoming computerized.

The next newsletter in this series will explore some of the dangers inherent to the world’s growing dependence on computers. I do not fear the possibility that AI-based computers will take over the world.

What I do fear is that a small number of human users of AI-based computers might take over the world or parts of the world. Consider a despot running a group, an organization, a company, part of a government, or the government of an entire nation. AI-based technology can be a powerful aid to such a person or group of people.

Dean, J. (2011). A curious sign that your IQ is high. Psyblog. Retrieved 2/22/2021 from https://www.spring.org.uk/2020/09/iq-is-high-sign.php.

Gardner, H. (1995). Reflections on multiple intelligences: Myths and messages. Phi Delta Kappan. Retrieved 2/16/2021 from https://www.academia.edu/download/50668263/Reflections.pdf.

Hutson, M. (10/18/2017). This computer program can beat humans at Go—with no human instruction. Science. Retrieved 2/16/2021 from https://www.sciencemag.org/news/2017/10/computer-program-can-beat-humans-go-no-human-instruction.

Suppes, P. (1988). Computer-assisted instruction. The encyclopedia of educational media communications and technology. Retrieved 2/21/2021 from https://suppes-corpus.stanford.edu/sites/g/files/sbiybj7316/f/computer-assisted_instruction_278.pdf.

Wikipedia (2021). Artificial empathy. Retrieved 2/20/2021 from https://en.wikipedia.org/wiki/Artificial_empathy.

WolframAlpha (2021). Wolfram Alpha computational intelligence. Retrieved 2/24/2021 from https://www.wolframalpha.com/.