Transforming How We Work | NVIDIA’s Smart Work

Before Answering, Ask the Right Question.」

The Quality of Questions and the Learning Organization in the Age of AI

Change how work is done. Make Smart Work happen.

Generative AI has made it dramatically easier to get answers.

Ask a question, and within seconds AI can organize information, suggest options, draft content, and even conduct analysis.

So, in an age when AI can provide answers so quickly, what becomes more important for human beings?

When we examined NVIDIA—one of the world’s leading AI companies—through the lens of the Human Leadership System (HLS), one intriguing theme began to emerge:

Perhaps what matters increasingly is not simply knowing more answers, but being able to identify what should be questioned in the first place.

From our research into NVIDIA, we arrived at one HLS Insight:

Why Does Jensen Huang Place So Much Emphasis on Questions?

NVIDIA founder and CEO Jensen Huang has spoken at Stanford Graduate School of Business about the way he approaches decision-making and organizational design.

One of the central ideas in his thinking is first-principles thinking.

Rather than beginning with what other companies already do, he goes back to more fundamental questions:

“If we were starting again under today’s conditions, how would we design this?”

“What is this actually meant to accomplish?”

Huang applies this approach not only to technology and products, but also to the way organizations themselves are designed.

One of his most striking remarks is:
“I would like them to question everything.”

He wants employees to question things rather than simply accept information or instructions from above.

He has also described the importance of sharing how he reasons—what the facts are, what the data says, what assumptions
are being made, what is known, and what remains unknown.

That gives us an important clue to NVIDIA’s way of working:

Do not rush toward the answer. First, examine the question itself.

“Why Do Different Companies Have the Same Organizational Structure?”

A distinctive feature of Huang’s thinking is that he does not immediately jump to solutions.

When discussing organizational design, he raises questions such as:

What kind of company is NVIDIA?
Why do companies producing very different things often adopt similar organizational structures?
What are the inputs to an organization?
What are its outputs?
What characteristics of the environment must the organization respond to?

Instead of beginning with:
“What is the best organizational structure?”

he steps back and asks:
What is an organization fundamentally supposed to accomplish?

That distinction matters.
Before searching for an answer, he re-examines what should actually be questioned.

Even After Success: “What Could We Have Done Better?”

Another revealing example is a question Jensen Huang reportedly asked his team:
“What could we have done better?”

What makes the question interesting is when he asked it.

It was not after a failure.

It came after a product launch that had been well received.

After failure, most people can identify what went wrong.

But after success, can a team still ask:

What was missing?
What could have been better?
Where did we fall short despite the positive outcome?

This separates the outcome—success—from the quality of the work that produced it.

Can people acknowledge shortcomings even when things went well?
Can they explain why something could have been better?
Can they turn that recognition into the next improvement?

A question like this reveals more than knowledge.

It reveals the willingness to examine oneself honestly and continue learning.

NVIDIA Describes Itself as a “Learning Machine”

This mindset does not appear to be limited to Jensen Huang personally.

NVIDIA describes its culture with statements such as:
“We are a learning machine.”

It also emphasizes ideas such as:
“The mission is boss.”
and
“Everyone has a voice.”

The important point is this:

A CEO who asks good questions alone does not create a learning machine.

A leader asks a question.
Employees ask questions back.
People acknowledge what they do not know.
They check facts.
They challenge assumptions.
They offer different perspectives.
They answer.
They act.
They examine the results.
And new questions emerge.

Only when this happens throughout the organization do questions become part of organizational learning.

The Connection with PQA — Precision Questioning & Answering

This is where we see an interesting connection with Precision Questioning & Answering (PQA), which SmartWorks has provided
to organizations in Japan for many years.

PQA is not simply a questioning technique.

It is a thinking framework that uses precise questions and precise answers to clarify:

What do we know?
What do we not know?
What do we need to know next?

PQA helps people think more deeply about complex problems, decisions, and situations—and helps reduce the danger of believing
that we understand more than we actually do.

PQA looks at questions through seven broad categories.

The Seven PQA Question Categories

Focus
What do we really need to discuss or understand right now?
What is the real issue?

Clarification
What exactly does that mean?
What needs to be made more specific?

Assumptions
What are we taking for granted?
What assumptions are we making?

Evidence
How do we know this is true?
What facts, data, or other evidence support the conclusion?

Causes
Why is this happening?
What factors are creating the situation?

Effects
What is likely to happen as a result?
What consequences or impact might follow?

Action
What should we do?
Who should do what, and how?

The Seven Categories Are Not a Sequence

This point is essential to understanding PQA.

The seven categories are not steps that must be followed in order.

You do not have to begin with Focus, then move to Clarification, then Assumptions, Evidence, Causes, Effects, and finally Action.

You can enter from any category, depending on the situation.

For example, a meeting might begin with an Effects question:

“If we proceed with this plan, what will happen to our customers six months from now?”

That answer may lead naturally to Evidence:
“What evidence makes us believe that?”

Then to Assumptions:
“What assumptions are behind that forecast?”

If an important assumption turns out to be questionable, the discussion may return to Focus:
“Then what problem are we actually trying to solve?”

And eventually move to Action:
“So what should we do next?”

In another situation, the discussion might begin with Action.

In another, Clarification.

The key point is:

The seven categories are not a staircase. They are seven lenses for thinking.

You use the lens that is most useful at that moment.

If the situation requires a different perspective, you move to another category.

And you can move back and forth as often as necessary.

This flexibility is one of the strengths of PQA.

Looking at Jensen Huang’s Questions Through the Seven PQA Lenses

Seen this way, the similarities between Huang’s questioning style and PQA become easier to understand.

For example:

“What kind of company is NVIDIA?”
can be viewed through the lenses of Focus and Clarification.

“Why do companies making different things use similar organizational structures?”
challenges an Assumption.

“What are the inputs and outputs?”
helps Clarify what the system actually does.

Questions about whether previously held beliefs are still valid bring us back to Evidence.

And:

“What could we have done better?”

can open up several different categories at once.

It may lead to questions about Causes.

It may reveal Effects.

It may challenge Assumptions.

It may require Evidence.

And ultimately it may lead to Action.

A single question does not necessarily belong permanently to one category.

What matters is that the question moves thinking forward—and that the conversation shifts to whatever perspective is needed next.

This Does Not Mean NVIDIA Uses PQA

This distinction is important.

At present, we have found no public evidence that NVIDIA has formally adopted PQA.

Nor have we found evidence that Jensen Huang himself has been formally trained in Precision Questioning & Answering.

Therefore, it would be inaccurate to say:

“NVIDIA practices PQA.”

What interests us is something different:

Two approaches that developed independently appear to share important patterns of thinking.

Do not rush to an answer.

Clarify the issue.
Challenge assumptions.
Examine evidence.
Explore causes.
Consider effects.
Think about action.

And then, when needed, question the situation again from another angle.

That flexible movement of thought is where we see an intriguing similarity with PQA.

The Danger of “Thinking We Know”

One important idea addressed by PQA is the Illusion of Knowledge.

In simple terms:
“I believe I understand something better than I actually do.”

People are not always good at distinguishing what they genuinely know from what they merely think they know.

And once we decide that we “understand,” we often stop asking questions.

PQA helps counter this by making questions and answers more precise and by distinguishing:

What do we know?
from
What do we not know?

There is a clear parallel here with Huang’s first-principles approach.

Instead of accepting an existing answer, he goes back to fundamental questions.

What is actually happening?
What facts do we have?
What assumptions are we making?
What do we know?
What do we not know?

The common principle is:
Do not move forward simply because you think you understand.

NVIDIA Through the HLS Lens: Questions and Learning

At this point, it is important to distinguish between the seven PQA question categories and the broader learning cycle within an
organization.

The seven PQA categories do not have a fixed sequence.

Organizational learning, however, often involves a recurring movement such as:

A question arises.
People engage in dialogue.
They think.
They reach an answer or judgment.
They act.
They observe the result.
They reflect.
They learn.
And a new question emerges.

We might describe this as:

Question → Dialogue → Action → Reflection → Learning → Better Question

Within that learning cycle, the seven PQA categories can be used whenever they are needed.

They function as:
tools for deepening thought.

That distinction is important.

In the Age of AI, Is Asking More Important Than Answering?

We are not arguing that:

“The ability to answer no longer matters.”

Quite the opposite.

The quality of both questions and answers will become even more important.

Generative AI can respond extraordinarily quickly once it receives a prompt.

But what if the assumption behind the question is wrong?

What if we are asking about the wrong problem?

What if the evidence behind the AI-generated answer is weak?

Human beings still need to ask:

What do we really need to focus on?

What exactly does this mean?

What assumptions are being made?

What evidence supports this?

What is causing the situation?

What effects could follow?

What action should we take?

The goal is not to ask AI seven questions in a predetermined order.

The human role is:

to recognize which question is needed at this moment.

That ability may become increasingly important as AI becomes more powerful.

Developing Self-Directed People Means Developing People Who Can Question

This issue also connects directly to human capital management and people development.

Companies today are focusing on themes such as:

human capital management,
self-directed employees,
reskilling,
AI and digital talent development,
and psychological safety.

But simply telling employees:

“Think for yourself.”
does not make people self-directed.

If managers always provide the answer, employees learn that:
“The boss has the correct answer.”

By contrast, consider an environment where leaders regularly ask:

“What do you think?”
“What evidence supports that view?”
“Is there another way of looking at it?”
“What could happen if we continue this way?”
“What should we do?”

And employees themselves become accustomed to raising questions.

Over time, people may move from:
waiting for answers
to
thinking for themselves.

NVIDIA’s idea that “Everyone has a voice” can also be viewed as one of the conditions that make this kind of organizational learning possible.

Where Should You Start Asking in Tomorrow’s Meeting?

When using PQA, there is no need to think:
“We have to begin with Focus.”
You might begin with Effects:

“If we follow this proposal, what will happen six months from now?”

Then move to Evidence:
“What makes us believe that?”

Then to Assumptions:
“What are we assuming in making that forecast?”

If that exposes a questionable assumption, you might return to Focus:
“Then what problem are we actually trying to solve?”

And eventually arrive at Action:
“So what should we do next?”

In another meeting, you may begin with Action.

In another, Clarification.

The important thing is not to use the seven categories in order.

Use the question that is needed when it is needed.

HLS Insight

Before Answering, Ask the Right Question.

Our research into NVIDIA has led us to one HLS Insight:

Before answering, ask the right question.

We are not yet defining this as a universal principle of HLS.

The next question is:

Is this distinctive to NVIDIA？

Or:

Is it a Smart Work principle shared by other hight-performing organization around world?

We will continue testing this hypothesis as we examine companies such as Microsoft, Netflix, and Amazon.

Study one company.

Identify an Insight.

Test it against the next company.

Look for differences.

Modify the hypothesis where necessary.

If a common pattern begins to emerge, test it again elsewhere.

Through this process, we aim to understand:

What kinds of systems enable people and organizations to work better—and more productively?

That is the purpose of SmartWorks’

HLS Global Case Research.

Dash × Peemo | HLS Global Case Research

Dash
Akira Chida, President of SmartWorks. He has worked for many years in leadership development and human resource development and is currently studying how leading companies around the world work through the lens of HLS — Human Leadership System.

Peemo
Dash’s AI Research Partner. Peemo researches and organizes publicly available information on companies, leadership, and organizational culture, and works through dialogue with Dash to develop HLS Insights and hypotheses.

At present, we have found no public evidence that NVIDIA has formally adopted PQA or that Jensen Huang has received formal PQA training. This article represents SmartWorks’ independent analysis comparing publicly available information about NVIDIA with the principles of Precision Questioning & Answering.

Akira Chida / Dash
SmartWorks
HLS Global Case Research