The speaker connects today's problem of understanding with Alan Kay's older vision. In this vision, a computer is not merely a machine for watching finished videos or playing finished games. It is a tool that raises human capabilities.
In the image the speaker presents, children change a game's code while playing it. Through this activity, they learn physics. What matters is not that the game is finished or that the computer behaves intelligently. What matters is that the children's understanding and abilities change when they change the code.
Figure: The slide shows Alan Kay's name, the title, and a drawing of two children. It supports the speaker's explanation of the computer as a medium that supports children's learning. The details of the historical image and the exact wording of the original essay cannot be confirmed from this material.
If people only watch a finished game or video, they receive the result. The mechanism behind the screen remains invisible. This is using a computer as a tool for passive consumption.
In an environment where the game code can be changed, the learner can act on the mechanism. First, the learner predicts, "If I change this, how will the movement change?" Then the learner changes the code, observes the result, and compares the prediction with reality. This repetition connects computational rules with visible movement.
Example (to make the explanation clearer): If you only watch a character fall in a finished game, you can imagine that a calculation like gravity is involved. But you cannot test that calculation yourself. If you change the in-game code and change how the character falls, you can compare the changed part with the result. From there, you can think about how rules in a computation produce movement.
This example does not teach the implementation of a particular game. The speaker's central claim is that a computer should do more than display answers: learners should be able to change the environment and think from the results.
This idea connects with the microworlds introduced earlier in the talk. In Seymour Papert's turtle example, children create instructions that make a robot draw pictures. The robot itself is not the purpose of learning. The purpose is to develop a feel for mathematics through the activity of moving the robot.
The game example also puts playing and changing code in one environment. The learner does not merely listen to an explanation from outside. The learner enters the mechanism, changes it, and observes the result. That experience changes the learner's way of thinking and abilities. In other words, the computer does not learn instead of the human; it amplifies the human's ability to learn and create the next idea.
The speaker thinks AI may make this older vision newly practical. The next examples are temporary tools that AI creates for understanding. They include user interfaces tailored to the immediate purpose, dynamic simulations, debuggers, and playgrounds.
The important point is not only having AI write code. People must operate the tool and understand the mechanism. This is different from taking humans out of the work loop. Instead of leaving everything to AI and receiving only the result, people can change and test the mechanism and enter more deeply into the loop that produces the next idea.
However, there is no guarantee that using AI will automatically deepen understanding. In the direction the speaker proposes, people need not only tools but also a mindset that values understanding and the creativity to build environments suited to learning. The question connecting Alan Kay's vision with today's AI is not only, "What can computers do instead of humans?" It is also, "By using computers, what can humans understand, and what can they become able to do?"