When people discuss generative AI, conversations often focus on outputs.

How much text can be generated?

How quickly can reports be written?

How many ideas can be produced?

These are important questions, but over time I became increasingly interested in a different one:

Where does meaning come from?

At first glance, the answer appears obvious.

A person asks a question.

An AI system generates a response.

The interaction ends.

However, my experience suggested that something more interesting was happening.

When an AI system produces text, it produces symbols, words, explanations, and arguments.

Yet none of these automatically become meaningful.

Meaning emerges only when a human interprets them within a particular context.

The same AI response may be interpreted differently by:

• different researchers,
• different organizations,
• different professional communities,
• or even the same person at different points in time.

The words may remain unchanged.

The meaning does not.

This simple observation gradually changed how I viewed Human–AI interaction.

Initially, I thought of AI primarily as a source of information.

Later, I began to see AI as something else:

a generator of interpretive possibilities.

The human participant then performs a different role.

Humans connect information to:

• experience,
• context,
• organizational reality,
• goals,
• constraints,
• and consequences.

In other words:

AI generates information.

Humans create meaning.

This distinction may seem subtle, but it has important implications.

Organizations are increasingly capable of generating enormous amounts of information through AI.

The challenge is no longer information scarcity.

The challenge is interpretation.

As AI capabilities continue to improve, we may discover that the primary bottleneck is not computation.

The bottleneck is meaning-making.

This realization eventually influenced our research on Human–AI knowledge systems.

Rather than asking:

How does AI generate information?

we began asking:

How does information become knowledge?

and later:

How does knowledge become credible?

The more we explored these questions, the clearer it became that meaning does not exist solely in AI outputs.

Meaning emerges through interaction.

It is created when information becomes connected to human understanding and practical experience.

Perhaps this is why two people can read the same AI-generated response and walk away with completely different insights.

The text is shared.

The meaning is not.