Why AI Companies Are Hiring Philosophers

I recently came across an NPR article about AI companies hiring philosophers to help develop their models.

That headline immediately caught my attention.

I have worked in technology for more than two decades. I have built software, designed systems, managed engineering teams, and helped businesses use technology to solve practical problems. When I think about the people needed to build artificial intelligence, I naturally think of software engineers, data scientists, mathematicians, researchers, and product leaders.

Philosophers were not the first people who came to mind.

That made me curious.

What problem had AI companies discovered that engineering alone could not solve? Why would some of the most advanced technology companies in the world need people trained in ethics, logic, knowledge, consciousness, and the meaning of human experience?

The more I thought about it, the more I realized that the answer goes to the heart of what makes artificial intelligence different from traditional software.

The Problem Was Not Intelligence. It Was Judgment.

Traditional software usually follows rules written by people.

A programmer can tell a system that an invoice above a certain amount requires approval. We can define the conditions, write the logic, test the result, and decide what should happen when something goes wrong.

Artificial intelligence does not always operate within such clearly defined boundaries.

Imagine asking an AI assistant whether an employee who lied to a customer should be fired. There is no simple rule that can produce the right answer in every situation.

The AI may need to consider the seriousness of the lie, the employee’s intention, the harm caused, the possibility of forgiveness, the company’s values, and whether trust can be rebuilt. Different values may point toward different conclusions.

This is not only a software problem.

It is a question about honesty, responsibility, fairness, consequences, and human behavior.

AI companies learned that making a model more capable does not automatically make it honest, wise, fair, or safe. A model can become better at answering questions without becoming better at deciding which answers it should provide, how certain it should sound, or what it should do when important values conflict.

Intelligence does not automatically create wisdom.

That is where philosophy becomes useful.

What Does It Mean for AI to Be Helpful?

At first, it may sound easy to tell an AI system to be helpful.

But what does helpful actually mean?

Should an AI help a student complete an assignment if doing so prevents the student from learning? Should it agree with a user to make the user feel supported, even when the user may be wrong? Should it provide information that has a legitimate educational purpose but could also be used to cause harm?

Even a simple instruction such as “be helpful and do not cause harm” creates more questions.

What counts as harm? Who gets to define it? Should the AI focus on immediate harm or possible long-term consequences? What happens when helping one person creates risks for someone else? When should an AI respect a person’s choice, and when should it refuse to participate?

These questions do not have purely technical answers. They belong to ethics, a branch of philosophy concerned with how we should act and what we consider good or right.

Philosophers have spent centuries studying situations where values conflict and no available choice is perfect. They are trained to examine assumptions, define terms carefully, test the logic behind an argument, and consider a problem from more than one moral perspective.

Their role is not simply to give the AI a list of rules. It is to help companies think more clearly about what those rules mean, where they may fail, and what kind of behavior they are trying to create.

From Rules to Character

Anthropic offers one of the clearest examples of this work.

The company has developed a written constitution for Claude that describes the values and behavior it wants the model to develop. Amanda Askell, a philosopher at Anthropic, is the primary author of the current constitution and has played a central role in shaping Claude’s character. Anthropic describes the constitution as an important part of its training process, not merely a public statement about ethics.

This distinction fascinated me.

The question is no longer only:

What rules should the AI follow?

The deeper question is:

What kind of AI should this be?

Should it be honest even when the truth is uncomfortable? Should it admit uncertainty? Should it respectfully disagree instead of simply pleasing the user? Should it care about the interests of people who are not part of the immediate conversation?

Anthropic’s Constitutional AI approach allows a model to use written principles to evaluate, critique, and revise its own responses. The goal is to develop an assistant that remains helpful while reducing harmful or misleading behavior.

This is philosophy being translated into engineering.

The philosopher helps explore what honesty, helpfulness, care, fairness, and responsibility mean. The engineer helps turn those ideas into a system that can influence model behavior.

Neither discipline is enough by itself.

Philosophy Is About More Than Ethics

Before reading about this subject, I mostly connected philosophy in AI with ethics. But the role is much broader.

Philosophers also study knowledge.

How do we know whether something is true? What counts as reliable evidence? How should we respond when sources disagree? How certain should we be before making a claim?

This area of philosophy is called epistemology, and it is directly connected to one of the biggest challenges in AI.

An AI model can produce an answer that sounds confident even when the information is incomplete, outdated, or wrong. Teaching a model to say “I do not know” is not only a technical challenge. It requires us to decide what knowledge means, how uncertainty should be expressed, and when confidence is justified.

Philosophers also study logic. They can examine whether a conclusion follows from the available evidence, whether an argument contains hidden assumptions, and whether the same reasoning is being applied consistently.

Then there is the philosophy of mind.

What do we mean when we say that an AI understands something? Is intelligence the same as consciousness? Can a system reason without experiencing the world? When an AI sounds caring, is it actually caring, or is it producing language that resembles care?

These questions may sound abstract, but they affect how people relate to AI. As models become more natural and emotionally convincing, people may trust them, depend on them, or form attachments to them.

We need to understand not only what the technology can do, but also how the experience of interacting with it may affect us.

Why This Became Personal for Me

This subject arrived at an interesting time in my own journey.

I recently rebranded my work as Mindful AI Guy. I made that change because I felt something important was missing from the way we often talk about artificial intelligence.

Most conversations focus on capability.

What can AI generate? What can it automate? How much time can it save? How many people can it replace? How much money can it make?

Those questions matter. I am an engineer and a business owner. I care about building systems that work and solving problems that create real value.

But capability is not the same as purpose.

Just because we can automate something does not mean we should automate all of it.

A local business owner may ask me to automate customer service. From a technical point of view, I can think about the model, the tools, the workflow, the integrations, and the return on investment.

But there is another question that deserves equal attention:

Which parts of the customer relationship should be automated, and which parts are valuable precisely because a human is present?

That is not an abstract philosophical exercise. It is a practical business question.

We can automate a response, but should we automate empathy? We can make communication faster, but will it still feel personal? We can reduce the time employees spend talking with customers, but are those conversations sometimes where trust is created?

The best answer may not be complete automation.

It may be using AI to remove repetitive work so that people have more time and attention for the moments where their presence truly matters.

Where Mindfulness Meets Artificial Intelligence

My mindfulness journey has taught me that mindfulness is not only about meditation or feeling calm.

It is about awareness.

It means paying attention to what we are doing, why we are doing it, and what we are experiencing while we do it. It asks us to notice our intentions, our habits, our reactions, and the consequences of our choices.

That is why I believe mindfulness has an important place in how we build and use AI.

Mindfulness asks us to pause before acting.

Philosophy asks us to examine our assumptions before deciding.

Engineering helps us turn our decisions into systems.

These are not separate worlds. They need one another.

Engineering without reflection can become blind optimization. We may become very good at building something before asking whether it should exist or what effect it may have.

Reflection without technical understanding can become vague. We may discuss what responsible AI should look like without understanding what the systems can actually do or how they are built.

Mindfulness helps connect the two. It keeps our attention on both the technology and the human experience surrounding it.

This is the idea I keep returning to:

AI is a powerful tool. Let us understand it clearly, use it mindfully, and keep asking what kind of humans we want to become while using it.

I intentionally use the word mindfully rather than thoughtfully.

Thoughtfulness is important, but mindfulness has a larger surface area. It includes thought, intention, attention, experience, awareness, and impact. It asks not only whether we have made a good decision, but also whether we are present enough to notice what the decision is doing to us.

Hiring Philosophers Is Not Enough

I am encouraged that AI companies are bringing philosophers into the room, but I also believe we should remain cautious.

Hiring philosophers does not automatically make a company ethical.

A company can publish principles and still make decisions based primarily on competition, speed, power, and profit. A philosopher can raise difficult questions, but that matters only when the organization is willing to let those questions affect what gets built, how it is released, and what the company chooses not to do.

The real test is not whether an AI company employs philosophers.

The real test is whether philosophy has enough influence to change decisions.

Philosophers also cannot solve these problems alone. Responsible AI requires engineers, psychologists, social scientists, legal experts, security researchers, domain specialists, policymakers, business leaders, and the communities affected by these systems.

Perhaps the most valuable contribution philosophers make is not that they always have the answer.

It is that they are trained to stop us from rushing toward an answer before we have understood the question.

The Questions We Cannot Automate Away

The more capable AI becomes, the more important these questions will become.

What should remain under human judgment? What should never be fully automated? How do we protect human dignity while gaining the benefits of technology? Who decides which values are built into systems used by millions of people?

And perhaps the most personal question of all:

What happens to us when we regularly hand our thinking, writing, decisions, relationships, and creativity to machines?

I do not believe the answer is to reject AI. I also do not believe the answer is to automate everything simply because we can.

The answer begins with awareness.

We should understand the technology clearly, use it with intention, and remain honest about both its benefits and its costs.

The future of AI will not be shaped only by what machines become capable of doing.

It will also be shaped by what humans decide is worth doing.

And by who we choose to become along the way.