Will Artificial Intelligence Save, Replace or even Affect Education Practices?

(a venture capitalist’s view)

AI, a.k.a. artificial intelligence, is a highly efficient artificial pattern recognition system (for recognizing patterns with a finite number of variations).

HI, a.k.a. human intelligence, is a highly efficient natural system for creating solutions to problems which have never been solved before; the central ability of the host of HI is an ability to create a solution to a new problem (this is what I teach my students, no matter what specific subject I teach at the time: http://www.GoMars.xyz/vvli.html; www.GoMars.xyz/vv.htm).

In short:

1. Various advances in AI are becoming a common place (the latest example is AlphaGo Zero: https://www.nature.com/articles/nature24270).

2. Professionals know that there is still a huge distance between AI and HI. However, for general public, a competition between various versions of AI (produced be different companies: https://www.forbes.com/sites/quora/2017/02/24/what-companies-are-winning-the-race-for-artificial-intelligence/#ffe12f0f5cd8) is seen as a competition between different sport teams: for example, self-driving cars = “race cars”; Chess, Go, Jeopardy winners = “sport champions”.

3. From the point of view of attracting public attention, i.e. marketing, the most promising actions should be based on more and more complicated competitions between AI and HI. Personally, I would like to see a human and an AI taking the same standard (commercially available) IQ tests (https://www.mensa.lu/en/mensa/online-iq-test.html).

4. Games like Chess and Go are not considered as popular as TV shows, but various competitions, like science subject Olympiads, may resonate with general public. That is why the next battle between AI and HI, which may attract wide attention of general public, should happen at the next Physics Olympiad (so far, excluding the practical/experimental part, and competing only in the theoretical part, solving problems similar to “F=ma” test; https://www.aapt.org/physicsteam/2016/exams.cfm). AI would be reading the same problems and solving them in real time. If AI will win the competition, that will be a clear proof of its actual power. That company which AI will do it first will win the marketing battle (at least for awhile).

Currently, no company can present AI which could solve physics problems, or math word problems, just by reading from a textbook. This is why so far rumors about AI replacing school teachers exaggerate the actual case.

As an expert in HI, I know that no AI can ever learn on its own (like AlphaGo Zero learned how to play Go) how to solve physics problems, or math word problems randomly taken from standard textbooks. This skill requires a high-level training which can be achieved only as the result of effective interaction with a good teacher. That automatically means that no AI can ever learn on its own how to be become a good teacher (i.e. a teacher who can teach beyond memorizing and repeating various – even very complicated – patterns (that is essentially no different from training animals doing tricks); a teacher who can teach how to create a solution to a brand new problem: http://teachology.xyz/general_algorithm.htm; http://teachology.xyz/sp.htm; http://teachology.xyz/mocc.htm; http://teachology.xyz/la.htm).

However, I know for sure, that some of the top AI developers disagree with the statement I just made.

Some time ago I had a meeting with one of the high-level executives and a businessman at one of large tech companies located in Cambridge, MA.

The mere fact that we met was already extraordinary for me; it was the first time I talked to such a person (BTW; I asked for a meeting three more companies, but only one decided to take a risk).

Even more impressive was the fact that we spent talking for more than an hour.

That conversation helped me to take a look into a mind of a person responsible for guiding multimillion projects.

Of course, we talked about education.

It did not surprise me that people like my opponent formed their view on education via reading or watching science fiction.

In their view, in the future a student will be interacting with AI via screens, speech, gestures, like a student does today when interacting with a teacher.

AI will tutor students, and will do it better than today an average teacher does.

Out conversation has clearly shown two facts: (a) business and tech leaders have a very overestimated view of the role AI will play in 10 to 20 years; (b) business and tech leaders have a very trivial (if not primitive) view on education.

Regarding the first fact I have been writing in the past, for example: http://www.GoMars.xyz/AI.htm.

In part, I wrote: “People working on AI believe that they can make an “artificial brain”. This type of belief is nothing new. For thousands of years, people have been dreaming about flying like a bird. And finally, Wright brothers invented an airplane. At last, men can fly! Yes. But NOT like a bird! What we – humans – created is a device which replicates one function of a bird, i.e. flying above the ground, but to this day there is no device replicating an actual bird. The field of AI is NO different.”

The second fact, that business and tech leaders have a very trivial (simplistic) view of education, is no surprise at all. They have grown up within the same cultural framework as all regular folks have, with only one difference; since they have achieved staggering success in life, they have even less doubts in their abilities to make right decisions than regular folks have.

Let us ask a question, who is smarter, Albert Einstein or Steve Jobs; Elon Musk or Stephen Hawking? Who is smarter, the guy who was smart enough to write the first version of what had become MS-DOS (Tim Paterson? Gary Kildall?), or Bill Gates who was smart enough to buy it from Tim Paterson for $50,000 and make billions of it? Is IBM’s Watson really smarter than Garry Kasparov?

Different people may approach this question differently.

One may say that we would need to give all those people to solve some abstract (not based on a specific content) problems and see how would they do.

Another approach would be saying that this question does not make any sense, because there are multiple types of “smartness” (in the same sense Dr. Howard Gardner talks about multiple intelligences); i.e. there is “programming smartness”, “business smartness”, etc.

Both these approaches fit the field of cognitive psychology.

Many people, however, do not bother with coming up with a psychological definition of “smartness”; they just equate “making big money” with “being smart” (folk question “if you are so smart, why ain’t you rich?” is an indication of this attitude). And people who made big money in tech especially susceptible to this sentiment.

Self-made tech millionaires and people who made it in the tech world from the bottom to the top believe in paradigm: “If I made big money, that means I’m smart.”

And that might be even true.

But even the smartest person in the world cannot know everything. When a car, or a refrigerator breaks, even the smartest people call a professional.

“We have talked to many teachers”, I was told during our conversation.

“How do you select who to talk to?”

“What is the chance that those people are not as good as they present themselves?”

“How do you assess if those people are as good as they present themselves?”

“You told me that at least two thirds of school teachers are not good at teaching. How do you know that? How do you know who is good and who is not? How do you know that people you talk to are from another third?”

“What is the chance that you are so visionary and charismatic person that when you tell people your vision they accept it without giving to it any critical thoughts (the “Halo effect”)?”

“What is the chance that people are much smarter and cynical than you think, and just tell you everything you like to hear, as long as they keep getting from you free stuff (money, books, tablets, computers, etc.)?”

“What is your personal description, definition, of “good teaching”? How does the structure of good teaching look for you? What is the most important result of teaching – for you – and how do you know it was achieved?”

That type of questions I would like to ask my conversation opponent, but I could not fit it in our hour.

If I had more time, I also would ask the same questions about “experts” who are usually hired to support the ideas of the influential “non-experts”. Speaking about experts, there is one question to which I never could get a clear answer: “How did it happen that $200,000,000 spent in Newark, NJ with the help of many experts on “improving education”, didn’t really lead to improvement in education of Newark children?” The fascinating story of Newark can be found in book “The Prize: Who's in Charge of America's Schools?” by Dale Russakoff (https://www.amazon.com/The-Prize-Charge-Americas-Schools-ebook/dp/B00AXS6BIE). A similar question I asked in my open letter to Mss. Laurene Powell Jobs, but, also, without an answer (http://www.teachology.xyz/xq.htm).

At some point in our conversation, I was presented with a vision that in the future, when a student is struggling with a homework, AI would recognize the struggle, and would offer a hint, like “click on this link and watch a movie”, or something else, exactly like a human tutor.

I mentioned that there are already various tutoring systems on the market, and many students just hate them.

Of course, this company has extensively studied those systems and knows how to develop AI tutor which will be much better than the existing ones.

In case it was not very clear, the last sentence was sarcasm.

As a person with an extensive tutoring experience, I know that a human tutor does much more than just offering guiding questions, or hints (http://www.GoMars.xyz/vv.htm). As an example, I used an episode from movie “Sully”. In the episode, test pilots used a simulator to demonstrate that the airplane did not have to land on water, that it could have been brought back to an airport. And Capt. Chesley "Sully" Sullenberger said to the commission: “You do not take into account the human factor” (one of my favorites moments in this great movie). When I was listening to my opponent, I said exactly same words: “You do not take into account the human factor”; in short, learning and teaching is simply much more than maintaining the flow of information, or the sequence of physical actions – that would have been just training (https://www.animaltrainingacademy.com/how-to-animal-training/).

I also used an example from a science fiction story. Long time ago a red a short story about the future. At the age of 16 all children would be assessed with the use of a “mental machine” which would prescribe for each child his or her profession for the rest of their life. Then each child would put a helpmeet, and a technician would charge the machine with a tape holding all professional knowledge needed for the child, and in a minute everyone would learn everything he or she would need to do the work. Except some guys, who could not learn anything from the machine, because their brain was special. Turned out those guys would be going to a regular school to learn how to program those machines and to develop those tapes.

The vision of education presented to me by my opponent was of sort similar to the one described on the science fiction story.

Instead of machines with tapes – computers with AI.

For masses, teaching essentially would be no different from training animals to do tricks.

I admit, that many students – those who today do not have a good teacher – would be getting better knowledge from AI tutor than from a human teacher.

However, parents with resources would be lining up into elite schools where human teachers good at teaching would be teaching their children (of course, with the help from AI).

To answering my title question, AI will affect education, and it will affect it greatly.

Mass education will become less dependent on the quality of teaching cadre. Knowledge and skills of an average student will increase. Businesses will have better prepared workforce for doing more complicated but still mostly routine work. Teachers will not disappear. Most of them will be working in public schools using all technologies offered by AI (in a way, this will be similar to construction workers who replaced a simple shovel with an automatic trench digger).

The best teachers will be concentrating in elite schools where students will learn more than just a very specific set of skills. The will also learn how to use those skills to create knew knowledge.

At some point in our conversation I was told: “My child complains that it is hard to write with a pen on paper. But using a stylus and writing on a screen is easier!”

I said: “Easier does not mean better.”

My phrase resulted in a long pause.

For me, who has been teaching physics and math for many years to all types of students, it was obvious that in education “easier” does not always mean “better for students”. On the contrary, true learning happens via overcoming obstacles and difficulties we call “mistakes”. Another known fact is that learning how to manipulate with fingers (including writing) helps children’s brain development.

Overcoming mistakes is the essence of learning. Guiding through this process is coaching (training, instructing). Teaching includes instructing but also has more (https://www.smashwords.com/books/view/665204).

But for my opponent, and many others, “easier”, “funnier”, “down to earth” automatically means “better teaching”.

Based on their view of education, they only support ventures which fit their view.

I was asked “Is the project method of teaching better than a traditional teaching?”

I said: “No”, and I was met with a bewildered glance, and an “attack”: “But when students do projects, it is fun, they are active”.

“OK”, I said, “I take my answer back. I change it to – there is no evidence for either to be better”.

That is exactly the situation.

I know that many so called “project based” approaches make it look like students learn more (students definitely look happier, though). But I also know that if the same students would be taught by a really good teacher who would use “traditional” approach, those students would learn even more. But I could never prove it; as well as no one can prove the opposite.

Today there is no scientific evidence that “project based teaching” is better than “traditional teaching”, or the opposite (http://www.GoMars.xyz/msm.html).

First, there is no commonly accepted definition of either type of teaching.

Second, there is no commonly accepted measuring procedures which would allow to compare the learning outcomes of students.

Third, there is even no commonly accepted list of the learning outcomes expected from students at the end of a school, or a given grade.

Today, measuring students’ learning outcomes is like measuring temperature using different devices and scales without any conversion factors.

The majority of the papers describe teaching “experiments” like – paraphrasing –

1) “We want our students to do better. For that we plan on trying this.” – if the project mostly involves faculty or teachers who directly teach students.

or

2) “We want our school teachers to teach better. For that we plan on trying this.” – if the project mostly involves faculty from a school of education.

Which leads us to a simple conclusion: nowadays, every single statement about how good or bad some form of teaching is, represents no more than a personal opinion and can be challenged by the opposite statement, and there is no scientific data to support either.

That means that today there is no such thing as science of education (the scientific field does exist, but there is no yet science).

Here we finally have come to the goal of my visit – to discuss the state of the science of education.

Only after the meeting, reflecting on our conversation, I realized that for more than an hour we talked about two different things (“apples and oranges”). I was talking about science of education. My opponent was talking about education. No wonder, we did not understand each other (which is completely my fault – I was not clear enough).

Before the meeting, I sent a letter, which had this part:

“This book (“The Structure of Scientific Revolutions” by Thomas Kuhn) is a very famous book. First time I read it about 20 years ago. It was a Russian translation. At the time, it was just one of many books I read on the history and philosophy of science. But recently I decided to read it again because I started seeing symptoms described in this book. Symptoms of a paradigm change in a specific area; in the field of education. The crisis is not here yet, but close, and the new paradigm has not been yet formulated, but it's in the air. The old paradigm, which is still the current one, is very simple. It says that learning is basically a process of imprinting the previously collected knowledge into student’s mind, and a teacher is a person who knows what students need and don't need to know.

There is however a feeling that the old paradigm might be outdated. More and more people write that students need to learn how to think critically. However, there is no common view on what critical thinking is, what is its structure, and more importantly how to teach it, especially on the massive scale. Different groups offer different approaches. And we can see the pockets of studies which, from my point of view, will lead to a formulation of a new paradigm. Right now we are in the transition from the old paradigm to a new paradigm, which has not been yet presented. We are in the pre-paradigm stage”.

 

My intention was to attract attention to the project which would help to advance science of education to a true science (http://www.teachology.xyz/chs.htm).

Only after the meeting I realized that my expectation was premature. I should have not expected from people of such status to think about science: “Hmm, what could we do to change things in science of education?”

I do not think this type of a question has ever popped up in the mind of my opponent before our meeting; due to a simple reason – those people do not think about advancing science, because they do not consider themselves scientists (http://www.GoMars.xyz/30uS.html). And also, because for many people any kind of a search is already seen as a scientific research; which is not actually a case.

In hindsight, I should not have been expecting from my opponent any attention to science of education. Unfortunately, many scientists in the field of education are not involved into scientific projects, too. Even a top-level official at a large research university told me once, that science of education is not possible. Even more, there is no need for science of education at all, because education is more like a craft; we just need more good “craftsmen”. And that was a person who ran at some point a teacher preparation institutional entity. Finally I understood, why my appeal to business leaders (http://www.teachology.xyz/MO.html) and my GoFundMe campaign (https://www.gofundme.com/teachology) were bound to fail.

In this paper, http://www.GoMars.xyz/nsf.html, I have shown that more than 90 % of NSF funded projects are not scientific, but social by their nature. For those projects, their primary goal is not producing new knowledge, but helping teachers teach better (http://www.GoMars.xyz/3pc.htm).

Almost all “experiments” funded by the NSF, or described in various magazines, fit the clear and universal law: “if we take two large groups of fairly similar students, and students in the first group will have a more extensive or diverse learning experience (for example, more contact hours with an instructor, or more time used for guided discussions, or more time spent on certain exercises, or training through more and/or different exercises, etc. than students in the second group), students from the first group, on average, will demonstrate better learning outcomes than the students in the second group” (the 1st Law of Teachology: http://www.GoMars.xyz/6lt.html).

In teaching, this law has the same explanatory and guiding power as the Newton’s Second Law has in physics. There is no need for trying to prove it again and again; it should be used for designing new teaching practices.

I have no doubts that the use of various technologies, including AI, will lead to better education – as a human practice.

To advance science of education someone would need to adopt a “Manhattan Project”, or “An Apollo Program” type approach (http://www.GoMars.xyz/30uS.html).

My hope was to find a tech business leader who would be interested in advancing science of education.

My search is still open (http://www.GoMars.xyz/YP.html).

Although, I doubt that this particular company will invite me back again; my talking stile is too "confrontational". However, if for some people the tone of the conversation matters more than its substance, maybe those people should take a look in a mirror? (https://teachologyforall.blogspot.com/2017/04/polite.html).

And speaking again about how self-made tech millionaires see education - most probably, most of them just have not had a good teacher in their life (they may have had friends, advisers, professional role models, but not good teachers). That is why they do not know what a good teaching is, and how big if a difference between a teacher and a good teacher. And when a person has not met a good teacher, that person just cannot resonate with anything you would try to tell him or her about good teaching. It is like trying to describe the taste of chocolate to someone who has never tasted any sweets; or trying to describe hiking to a person who has never left a room.

In the minds of self-made millionaires, being self-made may equate with being self-taught, and lead to being self-sufficient, or even arrogant. Unfortunately, even the smartest people in the world can let their arrogance to blind them. One of the saddest cases of arrogance is the death of Steve Jobs. When he learned he had cancer he did not immediately take care of it using a traditional medical approach, even though the cancer was treatable. But Steve so strongly believed in his own power that he refused to follow doctors' advice. When he finally gave up and turned to traditional medicine, it was too late. No doubt Steve Jobs was not just smart, he was a genius. And he knew it. But he took it too far.

Please, follow this link:  http://www.teachology.xyz/AI.htm

for some more discussion on the matter.

<<back to publications<<...................>> to Teach0logy.xyz>>