For Teachers: Using This Site in Class
This page is for high-school teachers. It sums up where the site's 13 chapters (258 spreads in all; every spread has an interactive figure, and the spreads about how AI works also have a Python program) connect to the Japanese national curriculum guidelines for "Information I" and "Information II" (Japan's required and elective high-school informatics courses), and what is good to know before using it in class. Quotations from Japanese government materials on this page are our own translations.
What you can do with this site
- One spread = one screen. On the left: a question, an explanation, a formula (with the meaning of each symbol), and the difference between the popular story and the original source. On the right: a figure students can touch and move. One spread is a small lesson unit on one topic.
- No installation, no sign-up. All of the programs about how AI works (146 of them) run as Python inside the student's own browser, and students can edit them and run them on the spot. No account is needed. Nothing students write or any results are sent outside.
- The same Python as the national materials. The main text of the national teacher-training materials for Information I uses Python, and so does the machine learning and neural network part of Information II.
- Original sources and a confidence grade for every event. This gives students material for practicing how to check for themselves whether a site is trustworthy (see the page "How Much You Can Trust This Site").
How the chapters connect to Information I and Information II
Information I topics: (1) Problem solving in the information society / (2) Communication and information design / (3) Computers and programming / (4) Information networks and the use of data. Information II topics: (1) Progress of the information society and information technology / (2) Communication and content / (3) Information and data science / (4) Information systems and programming / (5) Inquiry into finding and solving problems.
This matching is this site's own judgment (it is not a correspondence set by the government).
| Chapter | Information I (everyone) | Information II (elective) | Nearest place in the national materials and commentary | Number of spreads and the examples you can run |
|---|---|---|---|---|
| Ch. 01 Introduction | (4) Collecting, organizing and visualizing data (counting how often words were used each year) | (1) History of information technology | The history table in Information II, Lesson 5 (three booms, no fourth) | 19: see the booms by counting words, remove a peak that isn't there, fit a pattern of peaks and valleys |
| Ch. 02 Before the history | (3) How computers work (AND, OR, NOT and truth tables) | (1) History | Information II, Lesson 17 "the original form was in 1943"; the Lesson 5 table entry "proposal of the Turing test (1950)" | 20: inside the chess-playing doll's box, algebra with only 0 and 1, computing with just a tape and a table (the Turing machine), turning nerves into switches |
| Ch. 03 The first boom begins | (3) Algorithms (search) | (3) The ideas behind machine learning | The perceptron in Information II, Lesson 17; the table's Dartmouth 1956 and perceptron 1958 | 17: generating proofs one by one, the perceptron, Pandemonium, GPS |
| Ch. 04 Inside the first boom, and predictions | (3) Comparing search algorithms, (1) The effects of information technology on society | (3) Recognizing handwritten characters | The comparison of searches in Information I, Lesson 15; the table's ELIZA "1964" | 19: how ELIZA's trick works, A* search, fuzzy sets, handwritten ZIP codes |
| Ch. 05 The winter after the first boom | (3) Logic circuits, efficiency of algorithms (combinatorial explosion) | (3) Neural networks (what one layer cannot separate) | Information II, Lesson 17 "the winter years" and "have students recognize why the problems were resolved" | 19: the wall of one line (XOR), combinatorial explosion, DP matching, the Neocognitron |
| Ch. 06 The second boom begins | (3) Conditional branching (if ... then) | (1) Effects on human intellectual activity | The commentary on Information II (1), on having AI learn the intellectual assets of experts; MYCIN in the table | 20: chaining certainty factors, forward and backward reasoning, stopping a subway with fuzzy logic |
| Ch. 07 The Fifth Generation | (3) Programming (writing in logic) | (1) History, (4) Developing information systems | Only the single line "Fifth Generation Computer Project (1982–92)" in the table | 18: answering from facts and rules (Prolog), unification, waiting until the data arrives |
| Ch. 08 Winter and sprouts after the second boom | (3) The difference between search and learning | (3) Backpropagation, handwritten digits | Backpropagation in Information II, Lesson 17; the table's backpropagation (1986) | 20: Hopfield networks, Boltzmann machines, backpropagation, the earliest autoencoder, convolution, LSTM |
| Ch. 09 The third boom begins | (4) Collecting data (people attached the correct labels), analyzing data (regression, classification, clustering) | (3) Methods for regression, classification and clustering; training data and test data; overfitting | Information II, Lesson 17 "the appearance of GPUs"; the commentary's "what kind of data should be prepared" | 30: least squares, Bayes, logistic regression, principal component analysis, k-nearest neighbors, k-means, hierarchical clustering, decision trees, SVM, random forests, boosting, singular value decomposition, hidden Markov models, EM, reinforcement learning, AIC and cross-validation, training data and test data |
| Ch. 10 The peak of the third boom | (1) Changes in work | (1) Human intellectual activity and employment | Go in Information II, Lesson 5; 47% and 49% and "only a forecast" | 22: word2vec, DQN, transfer learning, object detection, autoencoders and VAEs, GANs, Adam, ResNet, Monte Carlo tree search, what's inside 47% and 49% |
| Ch. 11 The fourth boom begins | (1) Intellectual property and information ethics | (3) How deep learning is used | Current materials have no generative AI. The government's guide to generative AI (2024) stresses "understanding how generative AI works" | 19: self-attention, BERT, GPT, scaling laws, CLIP, diffusion models, RLHF |
| Ch. 12 The fourth boom today | (4) Predicting with a regression line (the form that appeared on the Common Test), (1) Reliability of information | (1) Thinking about future information technology | The predictions (2025, 2029, 2045) in Information II, Lesson 5 | 19: extending a line to estimate what comes next, how the way you ask shifts the answer by 70 years, three futures from the same past |
| Ch. 13 Summary | "Reliability and credibility of information" in every subject | (1) Discussion and presentation, (5) Inquiry | Information II, Lesson 5: "search for sites that explain the history of AI"; the inquiry example in Chapter 5 | 16: the map of predictions, how many years ahead each prediction looked, writing your own prediction so it can be checked later |
How this relates to the national materials
- In the national teacher-training materials for Information II (Lesson 5), students are asked to "search for sites that explain the history of artificial intelligence and investigate how the functions and performance of AI have changed," and teachers are told to "make sure they base their research on reliable sites" (printed page 54). This site was built so that it can be used as one of the sites students consult for that research activity.
- In some places the national materials or white papers differ from the original sources (for example, the materials' table gives ELIZA as "1964," while the original paper is dated January 1966). This site lays out what the original sources say side by side. It is good practice for students to check for themselves "which year is this?" and "whose words are these?"
- The four published years of the university entrance Common Test in "Information I" contain no AI questions. They do contain a question about predicting with a regression line, which uses the same idea as the Chapter 12 spread "Extend the line and estimate what comes next" and the Chapter 9 spread on least squares.
How much time to plan
There are 258 spreads in all. In one spread you read the explanation on the left and touch the figure on the right. In spreads about how AI works, choose "Python code" with the switch at the top right, and you can run ("▶ Run") the program in the same place. We have not written how long this takes, because students have not tried it yet (the estimates on the earlier version of the site were removed because they no longer fit the structure of the spreads).