We can connect with far more things than before. We can talk with people far away, investigate information from around the world, and look for the products and services we need. With AI, we can even gather ideas and examples related to a single question in a short time. Looking at these changes, it may seem that the more things we can connect with, the better we can understand the world. Yet two things need to be distinguished here. Being able to reach a piece of information is not the same as understanding why that information matters to our question.
Suppose someone is considering whether to live in a new place. AI can connect that question to a wide range of information, including housing costs, transportation, work, education, healthcare, climate, and safety. In a time when information was scarce, that alone would have been a major help. But having all the information does not necessarily make the decision clear automatically. The meaning of the same information changes depending on whether the person gives more weight to job opportunities or distance from family, how they think about cost and time, and how much they want an unfamiliar experience. Once the number of possible connections increases, another problem appears: how to interpret those relationships.
What matters here is that the world does not come with a single completed “map of relationships” from the beginning. Even the same fact connects to different things depending on the question from which it is viewed. Information that a place receives a lot of rain relates to clothing and transportation for someone planning a trip, to crops and water management for someone thinking about agriculture, and to drainage and disaster preparedness for someone thinking about urban design. The information itself has not changed, but the meaningful connections change when the question changes. Understanding is not only about gathering more information. It is also about grasping which relationships need to be examined for the question at hand.
AI may be able to perform this kind of relating at high speed. Instead of a person searching through materials one by one, AI can divide a question into several issues and connect each of them to relevant knowledge. If AI is seen simply as an answering device, the main role becomes returning a single conclusion as the result of those connections. But if AI is seen as a system that supports connection, another role becomes visible: enabling people to examine questions such as, “Why is this knowledge relevant to this question?”, “What issues have not yet been connected?”, and “From another perspective, which relationships would become important?”
This difference is not small. If AI shows only the information it judges most relevant to a question, we can move efficiently toward an answer. But if we cannot see why those items were selected, we may mistake the relationships presented to us for the structure of the world itself. In reality, different information may become important when viewed from another purpose, another set of values, or another position. The more powerful technologies for convenient connection become, the more important it is to be able to ask not only “What was connected?” but also “Why was it connected?”
That does not mean people must verify every connection from scratch. If they did, much of the value of using AI would disappear. For ordinary questions with limited consequences, it can be useful for AI to organize relevant information and guide people toward an answer by a short route. But for questions that involve major value judgments, or questions in which the problem itself is not yet settled, prioritizing efficiency alone can make important relationships invisible from the outset. What is needed is neither showing everything nor showing only one thing, but allowing the way connections are presented to change according to the question.
People can also change the question itself after seeing what has been connected. Someone may initially think a problem is about cost, only to realize while investigating it that it is really about time. Something that seemed like an individual choice may turn out to involve relationships with family or an organization. In such cases, AI has not helped only by answering the original question correctly. By making new relationships visible, it has enabled the person to reconsider what they should have been asking. Understanding can emerge not only at the moment an answer is received, but also when the relationship between a question and the world is reorganized.
The understanding emphasized by Sai＊aus is also related to this updating of connections. Simply connecting a person’s question to past information and returning an answer similar to an earlier one cannot fully address changed circumstances or newly felt discrepancies. If a question is used to connect relevant knowledge, a person compares and judges those relationships, and new differences or questions emerge from that judgment, those can be connected to the next round of understanding. Knowledge does not have to close as a completed answer; it can retain the possibility of becoming meaningful again through another question.
In this cycle, people are not passive users. Judging an AI-proposed connection by saying, “That relationship is not important to the current problem,” “Adding this condition changes how I see it,” or “These two things should be considered separately” becomes material for new understanding. AI proposes connections, people examine their meaning, and that judgment changes how the next connections are made. Here, efficiency does not mean only reducing human thought. Reaching relationships that would have been difficult for a person to find alone, and then enabling that person to think about those relationships for themselves, is also a form of efficiency.
Increasing the number of things that can be connected expands the possibilities for understanding. But the number of connections itself does not guarantee understanding. What matters is being able to examine, from a given question, what has been connected to what and why that relationship matters. And when necessary, it must be possible to reorganize the connections from another question. After the internet connected us to many people and sources of information, AI is beginning to create new connections between questions and knowledge as well. The issue is not only how much can be connected. It is whether, through those connections, people can update the way they see the world.
Not answers. Understanding.