I’m going to take you on a brief intellectual journey through the history of information storage and retrieval from prehistoric times to the present. The reason for doing this exercise is so that you can understand what’s new about learning, and what’s still going to be the same.
[Last-minute edit: I fear that this article is too long. It has a loooonng introductory section before we get to the part that really matters. But for the life of me I can’t figure out a good way to shorten it. We have a long human history, and there have only been a few big moments of change in how we pass on information to others. In order to understand what’s happening today with artificial intelligence and learning, we need to have a clear grasp of how we got here. Please, bear with me, and read until the end. For some the conclusion will be obvious – but for many, it will be an entirely new way of understanding the role of AI in human history, and the critical juncture we’re at today.]
Ready? Here we go…
Oral History and Oral Tradition
Before there was any written language, people passed information along verbally. Some of the information was contemporary:
“I see deer tracks – let’s hunt it down!”
“Now it’s time to plant the barley.”
“Put some wood on the fire; it’s about to go out.”
Some of the things people talked about were historical in nature. They were events that happened in the past:
“Your grandfather told me that one year it rained so much that the river came all the way up to the top of this boulder.”
“These are the names of your ancestors for the last fourteen generations. Memorize their names.”
Some of the knowledge that was passed along verbally was educational:
“This is how you roast a bird.”
“To make arrowheads, you need a rock about this size, and a sharp piece of antler, like this one.”
During the period of oral transmission and oral history, information was very insecure. You could only teach people who were your contemporaries. You could only teach what you remembered. Any information that was forgotten, or not told to someone younger, was gone forever when you died.
Early Writing
The earliest writing that we currently know about was Sumerian pictographs impressed in clay tablets. Then came Egyptian hieroglyphs, then Chinese syllabograms, then Old Persian cuneiform. All of these predate the Roman alphabet, which is the source of many modern written languages.
Writing was revolutionary and transformational. With the development of writing, someone could learn something, write it down, and die without ever teaching it to the person who later read what they wrote. You could learn something from someone without ever meeting them. You could pass your hard-won knowledge on to far more people than you could ever meet or speak to.
Durability of Early Writing
The durability of early writing varied by medium and circumstance. For example, information engraved on ancient Egyptian obelisks has lasted to the present day. Early Egyptian paper made from the stem of the papyrus plant existed from the time of the obelisks, but it’s a safe bet that the vast majority of information written on papyrus is lost to us through the decay of the ink, the paper itself, or both. We have many examples of writing on papyrus, but all of those examples together are probably a small fraction of what was originally created.
Mobility of Early Writing
The obelisks, while long-lasting, had a serious shortcoming: transportability. Papyrus wasn’t as durable, but it could be carried by a messenger, shipped to a distant land, or catalogued and stored in a library or archive. Papyrus, invented by the Egyptians and later used by the Greeks and Romans, resulted in two remarkable improvements over oral history and oral tradition: information could outlast its originator, and information could be transported to a different location.
Reproduction and Mass Production
Learning from information was further enhanced by reproduction, or copying. For example, the Jewish holy literature was meticulously hand-copied by scribes many times, for generations, spanning many centuries. Through this process of reproduction, two benefits were obtained: more people could learn the same information, and the durability problem was solved – simply make another copy before the last copy is illegible.
Later, mass production affected the learning process in two distinct ways. First, mass production increased the available supply of paper. And second, the printing press streamlined the process of making copies.
A third benefit of mass production is worth mentioning, and that is reduced cost. A hand-written Christian Bible could only be found in a church, but after the printing press was common, people of modest means could have a Bible within easy reach on a table or shelf in their own home. This reduced cost didn’t just affect religious literature. The number of books and the kinds of books both increased – fiction, non-fiction, science, philosophy, mathematics, instruction in the trades, political treatises – every area of life and thought was affected by, and benefitted from, the reduced cost of books.
Audio, Photographs, and Video
Sound played an important role in improving learning, even after we moved from oral tradition to writing. Even with the existence of books, a lot of teaching depended – and still depends – on oration and conversation. A teacher lectures, and students listen. After the telephone was invented, people exchanged information in real time over long distances. This was a great improvement in information exchange, but telephone conversations, like the conversations of ancient oral tradition and the lectures of professors, were impermanent.
Learning was also improved by the development of the camera, first in the form of still pictures, and later with moving pictures, or video.
Audio and video recording, like writing, created significant improvements in our ability to store and distribute information. And, as with printing, costs have come down over time.
Wireless Transmission
Wireless systems brought new improvements to the distribution of information.
First, text-based information – the telegraph system and Morse code – went from wired to wireless. Audio information was the next to go wireless, with radio stations springing up all over the world. Finally, video moved from film projected in a local theater to the wireless television.
All of these innovations had a radical multiplying effect on the distribution of information. It was another transformation in what we could collectively know, and how we could perceive our broader knowledge base.
The Nay-Sayers
There have always been people who believed that these improvements in information distribution were more harmful than beneficial, or, at least, capable of harm if not regulated and controlled.
Books: Mass production of books brought pulp fiction to the dime stores. Some people were concerned that these books would corrupt the values of young people.
Radio: There were some who complained that radio programs were a consummate time waster. People would sit around listening to the radio programs when they could be doing something more productive.
Television: The same thing happened when television sets started to gain popularity. Even in the 1960s and 70s people complained that television would “rot your brain.”
Today, in 2023, some people will still say that these arguments about the harmful effects of TV have some validity. Nevertheless, there is equal evidence to demonstrate the benefits of global television viewing. Far more people have far more understanding of politics, religion, philosophy, and global issues than any previous generation could have imagined.
Enter the Internet
The growth of the Internet brought with it an explosion in the area of information distribution and learning. You used to have to go the library to get a book on a topic you wanted to study; the Internet brought the libraries of the world to your computer. You used to have to go to a university to get a professional degree; the Internet brought the university to your computer.
Paper was replaced by disk drives, and ink was replaced by electrical signals. Information – any information – can be stored, transmitted, and copied without end. Some electronic storage media is even more short-lived than papyrus, but the ease of copying information to fresh media means that information need never be lost.
Nay-Sayers Again
Just as with books, and radio, and television, the Internet has given rise to pessimists who see more harm than good. Disinformation and misinformation can spread as quickly as information. The Internet can “rot your brain.” It can be a consummate time waster.
And yet, as with books, radio, and television, the benefits are proving to outweigh the harms.
The amount of information available to humans is increasing exponentially. The supersonic increase in the rate of information growth has to be accompanied by faster means of sorting and finding the information we need. The Internet has made this possible. It is at once a storage device, a cataloging mechanism, and an information search and retrieval tool.
And again, as with books, radio, and television, the costs of Internet access continue to come down. In the United States we currently have several government funding programs which may be generally referred to as rural broadband initiatives. The intent of these programs is to provide equal access to learning and business opportunities to all Americans. We can’t say the goal is accomplished in 2024, but all indicators are positive that more people will have more and better access to Internet resources every year.
Artificial Intelligence
Computational power and methods continue to improve, and now we’re entering another new phase that affects learning in every way. We’ve worked our way through information transmission, information storage, information replication – and now we’re at a time when we have information processing – outside our brains.
An Example (Introduction)
This concept is really new, and this section of the article is the heart of what I’m trying to teach you. If you got bored while I rehashed things you already knew about papyrus and printing presses and libraries and radios and television – well, it’s time to stand up, walk around, get refreshed and alert, and come back for the remainder of this article. I promise you, this next part is important, because this is the part that explains artificial intelligence and its effect on how we learn.
Ready? Here we go.
Computers have given us improved mechanisms for information transmission, storage, and replication, but the processing of information, up until now, has been a biological function of the human brain.
Don’t be confused here. I need to back up a moment and clarify what computers do. We can use clusters of programming code, called algorithms, to do lots of mechanistic things with information. For example, we can do all sorts of mathematical functions, so we have spreadsheets to help us keep track of inventory and balance the financial accounts. We can use algorithms to compare things: “if the room temperature is below a set number of degrees, turn on the furnace and heat the air.”
So far, all of the things we’ve been doing with computers are deterministic. The outcome is predictable and repeatable. Some of it is super simple, and some of it is complex, but it comes down to this: if the code is written correctly, we’ll always get the same result for a given set of inputs.
Until now, we’ve called what those deterministic systems do “information processing.” But it’s not, not really. It’s information handling. The information is the same, before and after the deterministic system handles it.
So, what’s different with Artificial Intelligence? AI actually processes information. The after isn’t the same as the before.
An Example (Processing Two Topics)
I can do a web search on, say, “Taoism,” and then I can read the information I find in the top ten results. But the computer didn’t teach me Taoism. The information processing necessary to learn Taoism was a biological function of my human brain.
Of course, that’s a vast improvement in learning.
- Centuries ago, to learn Taoism, I would’ve had to go to a Taoist temple and be taught by a Daoshi (Taoist priest or monk).
- Later, I could go to the library and read books on Taoism.
- After that, when books became cheap and plentiful, I could go to a bookstore, buy books on Taoism, read them, and keep them in my own library for later reference.
- More recently, I can do that Internet search to learn about Taoism. Since I’m already paying for a computer and an Internet connection, I can think of the monetary cost as “free.” I didn’t even buy one book.
Now. Two things.
(1) I did learn about Taoism through traditional modern means: I read books and did research on the Internet.
(2) And I did the same thing to learn about Stoicism.
Several years passed.
Now, in the present decade, in my mind – according to my perception – there are a lot of similarities between Taoism and Stoicism.
It occurred to me to wonder: “Taoism and Stoicism may have some similar values and perspectives, but at their heart they are two different systems of thought, arising from two different worldviews. So – how are they different? What distinguishes Taoism from Stoicism?”
Today, there are two inexpensive ways to answer my question.
Method #1: I can read all the books again, do all the Internet searches again, and with all of the information fresh in my mind I can use my own biological brain to do the information processing. At that point, I could even write an article to compare and contrast Taoism and Stoicism. I would have my answer.
Method #2: I can present the question as an input to an Artificial Intelligence (AI) Large Language Model (LLM), like OpenAI’s ChatGPT, or Google’s Gemini, or Microsoft’s Copilot.
And that’s what I did. I chose Method #2.
Specifically, in November 2023 I asked this question to ChatGPT (GPT-3.5): “What are the major similarities and differences between Taoism and Stoicism?”
I let the LLM do the information processing, instead of using biological information processing methods in my own brain.
In mere seconds, the LLM did what it would have taken me hours of work to do. It was a great article. It gave me exactly the kind of information I was seeking.
A search engine can find existing articles about Taoism and Stoicism on the Internet.
An AI Large Language Model can create an article about Taoism and Stoicism. It processes information and creates something that didn’t exist before.
And it saved me hours.
To Sum Up
Here’s the history of human learning, from prehistoric times to the present.
- Verbal. Oral. Speak information to someone next to you.
- Written. Information can be stored and transported. People who never met each other can learn from each other.
- Printed. Printing moved forward from hand-written documents to mass production.
- Audio/video recording. Recording did for verbal and visual communication what printing did for written communication.
- Wireless transmission. Vastly improved the spread of textual, audio, and video information.
- Internet. Vastly improved the systems for storage, search, and retrieval of all three kinds of information (textual, audio, and video).
- Artificial Intelligence. Performs information processing, not just storage/search/retrieval. AI can create new material to assist with learning.
Given the huge and ever-increasing amount of information, it’s time for new methods of processing information. We can’t just store it and retrieve it any longer. There’s just too much information. We need systems that can process that information, discover new relationships between widely disparate facts, form conclusions, and express those conclusions in ways that we can understand.
It’s time for the systems that store the information to learn from the information, and teach us.
–Bob Young
February 19, 2024