For most of human history, intelligence had a clear location.

It existed inside human minds.

Humans interpreted the world, created meaning, constructed narratives, and decided what mattered. Technology expanded human ability, but it remained external. Tools could extend human strength, memory, communication, and productivity, yet the fundamental process of understanding reality remained a human activity.

Artificial intelligence changes this relationship.

The most important transformation brought by AI is not simply that machines can perform more tasks. It is that intelligence itself is becoming embedded into the environments where humans live, work, learn, and make decisions.

AI systems increasingly participate in:

• how information is organized;
• how decisions are recommended;
• how institutions operate;
• how people understand themselves and others;
• how societies define problems and possible solutions.

The central question of the AI age is therefore not only:

What can AI do?

The deeper question is:

What happens to human agency when intelligence becomes part of the structure surrounding human life?

This question gives birth to AI Humanities.

AI Humanities is not merely the application of traditional humanities perspectives to artificial intelligence. It is a new field created by a new historical condition: human beings are no longer the only systems capable of producing explanations, narratives, judgments, and representations of reality.

The task of AI Humanities is to understand what this transition means for human agency, institutional legitimacy, education, enterprise, political power, and civilization itself.

It begins with a simple recognition:

AI is not only changing what humans do.

It is changing the conditions under which humans understand.

The Externalization of Intelligence

Previous technological revolutions externalized particular human capabilities.

Industrial machinery externalized physical strength.

Writing externalized memory.

Computers externalized calculation.

Networks externalized communication and coordination.

AI begins to externalize something deeper: interpretation.

An AI system can summarize an event, explain a concept, evaluate options, generate a narrative, recommend a decision, and propose a course of action. It does not merely store or transmit human knowledge. It actively reorganizes information into forms that influence understanding.

This creates a structural shift.

A person who once had to interpret an unfamiliar situation independently may now ask an AI system to frame the problem. An employee may rely on AI to evaluate strategy. A student may use it to understand a subject. A government agency may use it to classify cases. A company may use it to predict behavior, assess risk, or shape customer choices.

In each case, AI participates in the production of meaning.

The result is not simply greater efficiency. It is a redistribution of interpretive authority.

For centuries, the ability to explain was closely connected to the ability to understand. A scientist explained a theory because the scientist understood it. A teacher explained a subject because the teacher possessed knowledge and experience. A judge explained a decision through a visible legal structure.

AI weakens this connection.

An intelligent system can produce an explanation that is useful, persuasive, and operationally effective even when no individual human being fully understands how the explanation was generated.

We are therefore entering a world in which explanation can become separated from comprehension.

This is not only a technical problem of model transparency.

It is a political and humanistic problem.

Because whoever shapes the explanations through which society understands reality also shapes what society can see, question, and imagine.

The Crisis of Interpretation

Human agency depends on more than freedom of movement or formal choice.

It depends on the ability to understand the conditions under which choices are made.

A person remains an active subject when they can ask:

Why did this happen?

Why was I treated this way?

What alternatives existed?

Who designed the system?

What assumptions produced this conclusion?

Can the decision be challenged?

When these questions cannot be answered, agency begins to weaken.

The individual may still be permitted to choose, but the field of possible choices has already been structured by systems that remain invisible.

This is increasingly relevant in recruitment, credit, education, insurance, healthcare, social media, public services, consumer platforms, and workplace management.

An algorithm does not always need to command a person.

It can rank options, change visibility, adjust prices, prioritize information, delay access, or quietly redefine what appears reasonable.

The person experiences a decision.

The system has already shaped the environment in which the decision becomes possible.

This is the crisis of interpretation at the center of the AI age.

People may receive more information while understanding less about the systems affecting them.

They may obtain faster answers while losing access to the reasoning structures behind those answers.

They may become more capable of acting while becoming less capable of explaining why they act.

This is why AI governance cannot remain limited to privacy, bias, safety, and compliance.

These questions matter, but they do not reach the deepest layer.

The deeper issue is whether human beings retain meaningful participation in the production of interpretation itself.

Human Agency Is Becoming a Design Problem

Human agency has traditionally been treated as a philosophical or legal concept.

In the AI era, it also becomes a design problem.

Every intelligent system contains assumptions about the person using it.

Does the system treat the user as someone who should understand?

Or merely as someone who should comply?

Does it expand judgment?

Or replace judgment?

Does it present alternatives?

Or optimize the user toward a preferred outcome?

Does it help people understand uncertainty?

Or conceal uncertainty behind a confident interface?

These are not secondary questions.

They determine whether AI becomes an instrument of human development or an architecture of dependency.

A system can be highly accurate and still weaken agency.

It can be convenient and still narrow human judgment.

It can produce smooth experiences while quietly removing meaningful choice.

The most advanced AI product may therefore not be the one that makes the greatest number of decisions for the user.

It may be the one that helps the user make better decisions without losing the ability to judge.

This distinction will matter increasingly to technology companies.

As AI becomes integrated into education, healthcare, finance, professional services, and everyday decision-making, trust will depend not only on performance.

It will depend on whether users feel that the system increases their capacity or gradually makes them dependent upon it.

The next generation of responsible AI will therefore require more than technical safety.

It will require the design of systems that preserve:

• judgment;
• interpretive access;
• meaningful choice;
• narrative continuity;
• the ability to question and appeal;
• the capacity to understand how conclusions were reached.

Human agency must become a design objective.

Cognitive Sovereignty

The traditional information economy was organized around access to knowledge.

Those who possessed more information often possessed greater power.

AI changes this equation.

Knowledge is becoming abundant. Answers are becoming inexpensive. The ability to generate content, summarize documents, and retrieve information is becoming widely available.

The new scarcity will not be information.

It will be the ability to determine how information is organized, interpreted, and translated into action.

This is the emergence of cognitive sovereignty.

Cognitive sovereignty is the capacity of individuals, institutions, and societies to participate meaningfully in the processes that shape interpretation and judgment.

It is not simply the right to receive information.

It is the ability to influence the frameworks through which information becomes meaningful.

In practical terms, cognitive sovereignty asks:

Who defines the categories used by an AI system?

Who decides which objectives are optimized?

Who determines what counts as risk, success, relevance, harm, or efficiency?

Who can challenge the system when its interpretation conflicts with lived reality?

Who is allowed to participate in designing the cognitive architecture through which society increasingly understands itself?

These questions will become central to governments, AI laboratories, universities, corporations, and civil society.

They will also create a new form of inequality.

The divide of the AI era may not simply be between those who have access to AI and those who do not.

It may be between:

• those who receive interpretations;
• those who can influence interpretations;
• and those who design the structures through which interpretation occurs.

The deepest form of power will belong not merely to those who own information, but to those who shape the systems that organize meaning.

AI as Cognitive Infrastructure

The danger of AI is that it may weaken human judgment.

Its promise is that it may dramatically expand it.

Human cognition has natural limits.

Attention is limited.

Memory is limited.

Individual experience is limited.

The complexity of modern institutions, global markets, scientific knowledge, technological systems, and public policy increasingly exceeds what one mind can process alone.

AI can help humans carry this complexity.

It can function as a cognitive exoskeleton.

A physical exoskeleton does not replace the body. It supports the body, distributes weight, and expands the range of possible action.

A cognitive exoskeleton can perform a similar function for the mind.

It can help people:

• organize complex information;
• compare competing interpretations;
• identify hidden assumptions;
• maintain continuity across long investigations;
• explore alternative decisions;
• translate technical systems into understandable language;
• reconstruct problems that initially appear chaotic.

Used well, AI does not eliminate human judgment.

It creates the conditions under which judgment can operate at a higher level.

This possibility has major implications for education.

For centuries, advanced intellectual development depended heavily on access to good teachers, strong institutions, specialized books, social privilege, and long periods of training.

AI may lower some of these barriers.

A student can ask questions repeatedly without embarrassment.

A worker can translate an unfamiliar institutional language.

A citizen can examine a policy document.

An entrepreneur can test strategic assumptions.

A researcher can compare conceptual frameworks.

A person confronting a complex life decision can organize possibilities that previously appeared impossible to hold together.

This does not mean AI automatically produces wisdom.

It means that cognitive support may become infrastructure rather than privilege.

Education may therefore shift from the transmission of knowledge to the cultivation of judgment.

The most important educational question will no longer be whether students can reproduce information.

It will be whether they can work with intelligent systems without surrendering intellectual independence.

The New Economic Frontier

The economic value of AI is usually discussed in terms of productivity, automation, and cost reduction.

These will remain important.

But a larger market may emerge around cognitive infrastructure.

As AI systems become more capable, the challenge will not simply be giving people access to intelligence.

It will be helping people use intelligence without becoming disoriented, manipulated, or dependent.

This creates opportunities in:

• human-centered AI interfaces;
• decision-support systems;
• AI-assisted education;
• professional judgment tools;
• institutional translation;
• public-service navigation;
• cognitive health and attention management;
• trust, explanation, and contestability systems;
• tools that help organizations understand the social consequences of AI deployment.

The long-term value may not lie only in systems that perform tasks faster.

It may lie in systems that help people and institutions remain capable of understanding what they are doing.

For companies, this creates a strategic distinction.

One class of AI products will maximize replacement.

Another will maximize human capacity.

The first may generate rapid efficiency.

The second may create deeper trust, stronger institutions, more resilient users, and more durable relationships.

Investors should pay attention to this difference.

The next major AI opportunity may not only be a more powerful model or a more capable robot.

It may be the infrastructure that allows human beings to live, learn, decide, and remain psychologically and politically functional inside increasingly intelligent environments.

The question for long-term capital is therefore not only:

Which company possesses the strongest intelligence?

It is also:

Which companies are building the conditions under which intelligence becomes usable, trustworthy, and compatible with human agency?

Why Universities Need AI Humanities

Universities often divide AI into technical and ethical domains.

Computer science studies how AI systems work.

Law studies regulation.

Policy studies governance.

Philosophy studies moral questions.

Social science studies institutions and behavior.

Each field contributes something important.

But the transformation now underway crosses all of them.

AI simultaneously changes:

• how knowledge is produced;
• how people form judgments;
• how institutions make decisions;
• how authority is justified;
• how education functions;
• how identity and narrative are constructed;
• how social reality is organized.

No single existing discipline fully contains this problem.

AI Humanities should therefore not become merely another ethics course added after technical development is complete.

It should function as an interface discipline.

It should connect:

• computer science with human meaning;
• product design with agency;
• governance with interpretation;
• education with cognitive development;
• economics with human capacity;
• institutional power with the structure of explanation.

Its research object is not only AI.

Its research object is the changing relationship between intelligence and human life.

Universities that recognize this early will not simply create a fashionable new program.

They will help build the intellectual infrastructure required for the AI age.

Why Think Tanks and Governments Need a New Framework

Governments and policy institutions often approach AI through familiar categories:

national competitiveness, regulation, safety, privacy, labor disruption, security, and innovation.

These are necessary, but the framework remains incomplete.

Once AI systems participate in interpretation, public policy must also address cognitive power.

A government may regulate whether an automated system discriminates.

It must also consider whether citizens can understand and challenge the system.

A company may comply with privacy rules.

It must also consider whether its product gradually erodes independent judgment.

A public institution may introduce AI to improve efficiency.

It must also ask whether the system makes administration more legible or more opaque.

Future governance frameworks may therefore need to include principles such as:

• the right to understand consequential automated decisions;
• the right to contest machine-generated interpretations;
• safeguards against excessive cognitive dependency;
• protection of meaningful human judgment in critical domains;
• public participation in defining the values embedded in intelligent systems.

These questions sit between law, design, philosophy, social science, and institutional strategy.

That is precisely why AI Humanities is needed.

A Field for the Age of Intelligent Environments

AI Humanities is not anti-technology.

It does not begin from fear of intelligent machines.

It begins from the recognition that intelligence is becoming environmental.

When intelligence enters schools, companies, hospitals, cities, media systems, public institutions, and personal decision-making, the human question can no longer remain separate from the technical question.

The field must study at least five transformations:

First, the transformation of interpretive authority.

Who explains reality when AI systems participate in explanation?

Second, the transformation of human agency.

How can people remain active subjects rather than passive recipients of optimized environments?

Third, the transformation of cognition.

How does human thinking change when intelligence becomes external, continuous, and collaborative?

Fourth, the transformation of institutions.

How must companies, governments, schools, and legal systems change when decisions are produced by hybrid human-machine structures?

Fifth, the transformation of civilization.

What kind of humanity emerges when intelligence is no longer located exclusively inside the human mind?

These questions are already present.

The institutions capable of recognizing them early will shape how AI enters society.

Those that ignore them may discover that they have built intelligent systems without building a society capable of living with them.

Conclusion: The Next Race Is for Human Agency

The first AI race has focused on capability.

Which model can reason, code, write, predict, and generate most effectively?

The next race will be broader.

It will concern the relationship between intelligence and human agency.

Which companies can build AI that strengthens judgment rather than merely capturing attention?

Which universities can educate people for a world of externalized intelligence?

Which governments can preserve public legitimacy when algorithms participate in decision-making?

Which investors can identify the infrastructure required for human beings to remain capable inside intelligent environments?

Which societies can use AI to expand human possibility without quietly reducing human autonomy?

These are not secondary questions to be answered after the technology is complete.

They are part of the technology’s meaning.

AI Humanities begins from a simple proposition:

The central challenge of the AI age is not only to make machines more intelligent.

It is to ensure that intelligence—wherever it exists—continues to enlarge the human capacity to understand, judge, create meaning, and participate in shaping reality.

The future will not be determined only by who builds the most powerful intelligence.

It will also be determined by who defines what that intelligence is for.

And ultimately, the success of AI should not be measured only by what machines become capable of doing.

It should be measured by what human beings become capable of becoming.

Jiayu Li is a student at Chung-Ang University. His writing focuses on artificial intelligence, governance, technology and society, and the human consequences of emerging intelligent systems.