What we at Bonginkan are thinking aboutIt has long been said that philosophy cannot put food on the table.Recently, however, we have begun to feel that this assumption itself may be quietly shifting.At least in our day-to-day work at Bonginkan, philosophy and thought no longer feel detached from reality.They are becoming increasingly inseparable from it.One reason is that values and ways of thinking are now implemented directly as code, products, organizations, and the behavior of AI agents.Questions such as how we think, what we allow, and which ambiguities we choose to keep no longer remain abstract discussions.They appear in systems that actually operate in society, and their consequences return to us with surprising speed.This kind of environment was not common in the past.Within this context, we have found ourselves reconsidering a fundamental question:what is the true nature of intelligence?Intelligence is often associated with accuracy, logical rigor, or the ability to arrive at correct answers quickly.Society ultimately evaluates outcomes through numbers and results, and there are many situations where clear yes-or-no decisions are required.That reality cannot be denied.As a group engaged in AI development, we understand the severity of precise systems very well.A single bug can cause a program to stop entirely.A small oversight in a conditional branch can break the whole system.In implementation, decisions must eventually collapse into either zero or one.At the same time, when we look at companies, societies, organizations, people, and even the requirement-definition level of AI agents, the world appears to be built mostly from gradients.Assumptions are rarely fully fixed.Values differ.Human emotions and relationships influence decisions.What was correct yesterday may not be correct tomorrow.At this layer, making decisions too quickly in black-and-white terms does not always lead to good outcomes.In fact, forcing clarity too early can sometimes damage the complexity of reality itself.Perhaps for this reason, we are sometimes perceived as being “ambiguous.”It may appear as though we are not deciding, or that we are deliberately holding back.From our own perspective, however, it feels less like an inability to decide and more like a conscious separation of which layer and which timing decisions should belong to.Bonginkan does not disregard precision.On the contrary, because we deeply understand how unforgiving precise systems can be, we place great importance on what lies before them:a normative layer of quality.In this layer, accuracy alone is not sufficient.What matters are things that are harder to quantify:whether a judgment is coherent, whether language is precise, whether people can accept the decision, whether a sense of dignity is preserved.These elements, accumulated over time, seem to shape the quality of intelligence itself.What is interesting is that this normative layer, while difficult to quantify, still contains surprisingly strict boundaries.There are communities, circles, and schools of thought.Those who belong can recognize them naturally, while they remain difficult to see from the outside.Here, too, there is a form of rigor—but it is closer to shared sensibility and aesthetic judgment than to logic alone.We tend to think that higher intelligence is not about possessing many correct answers.Rather, it may lie in sensitivity to how questions are framed, how assumptions are placed, and in the ability to hold ambiguity without eliminating it, while still making responsible decisions.The ability to draw clear lines can often be replaced by algorithms or systems.However, sensing how many gradations exist between black and white, and handling them without destroying them, is not easy.Intelligence, then, may not be the power to eliminate ambiguity, but the capacity to endure it.It may be the ability to hold rationality and emotion, efficiency and ethics, speed and dignity together—without discarding any of them—while still moving forward.Today, this kind of intelligence seems, for the first time, to be testable in real society.Thought becomes code, code becomes behavior, and behavior returns as logs and outcomes.Organizations and AI agents honestly reflect the strengths and weaknesses of their underlying design philosophies through friction and sustainability.Philosophy and thought no longer remain as words alone; they are recovered through development and management.This is a harsh environment, but also a remarkably healthy one.At Bonginkan, we do not see ourselves as merely an AI development company.We aim to be a collective of intelligence—one that possesses deep language understanding and perceptual ability that cannot be fully measured by simple cost or efficiency metrics.We seek to build a normative and aesthetic layer on top of precise systems, rather than choosing one over the other.The essence of intelligence may lie in not oversimplifying the world, while still refusing to escape from judgment and responsibility.Not breaking things, not rushing, and still choosing to carry the weight.If this truly is an era in which the nature of intelligence can be tested within real society,then it is undoubtedly a difficult time—and at the same time, an extraordinarily stimulating one.And this experiment, we believe, has only just begun.