Economic ages are defined by work. In the Agricultural age, labor was physical, most people worked the land, and that work drove the economy. In the Industrial age, labor migrated to factories, and manufacturing work gained economic primacy. With the Information Age, value shifted again, toward management and analysis of information as machines replaced a large amount of physical labor. We called it knowledge work.

Each era elevated new forms of work over old ones. Millions of jobs became obsolete, and millions more jobs simply lost their centrality to economic progress. They persisted, but with low or no impact to productivity across the economy.

As we stand at the cusp of the AI Age, it's obvious that artificial intelligence can easily replace most forms of knowledge work. This isn't because AI is uniquely suited to replace most knowledge work (though it is). The inevitable conclusion of the Information Age was always going to be an invention that replaced “knowledge work.”

Every economic transition demands more learning. Call it whatever you want: education, or the search for wisdom. And the defining human challenge of the AI Age will be to realize our capacity for endless curiosity.

What is Going to Happen?

In the face of economic upheaval, one thing has helped societies thrive: education. Learning. Education makes change survivable and so favorable to people that we call the change "progress." Resistance in any form, whether the luddites, the unions, the doomers, gloomers, or the boomers, has consistently failed to stop change. And these forms of opposition tend to have deranging effects on groups and societies that embrace them.

Each time the dominant form of work has changed, prosperity has followed those who re-skilled and adapted. Success went to societies who prioritized immediate learning for affected workers and systems of education for their upcoming generations. The university system as we know it was founded in the transition from Agrarian to Industrial society. The GI Bill and technical training marked the second boom of learning, driving knowledge work in the information age (and accelerating into the oversaturation of college education we see today.)

This will continue to be the case. But today, it will go faster. The Agricultural Age lasted millennia; the Industrial Age two centuries, and the Information Age barely sixty years. Each era compounds progress from the previous one, accelerating transitions exponentially. This means that the AI Age isn’t just coming, it’s already here. This age is defined by systems capable of managing themselves and, increasingly, managing us.

Today, AI writes and debugs code, diagnoses medical images, composes music, summarizes law, and offers strategic recommendations. As AI commoditizes intellectual tasks, knowledge work itself becomes economically less valuable. The marginal cost of research is approaching zero, and tasks that once required specialized experience or extensive training can be completed by AI in seconds, at minimal cost.

As knowledge tasks become automatable, learning itself will become the paramount human work and, therefore, anyone’s most marketable skill. Today’s accelerated cycle—learn, innovate, implement, automate, learn again—means humans must be relentlessly curious. We must ask more questions of our world, faster, across more domains, and with greater sophistication, just to remain economically relevant. In other words, learning will replace knowledge as our dominant economic output. This isn't a radical idea.

At Perplexity, meetings don't have a presentation. They start with a problem, and from there we ask questions. What would we need to know to solve this? AI is good at helping us find that answer, fast. We don't form a committee, hire a consultant, or schedule another meeting. We ask more questions. Suddenly, every meeting is ten meetings’ worth of decision-making.

I think of this as "question work." Aside from the startup furniture and video conference software , these meetings feel nothing like the era of knowledge work. The most powerful people in the room don't come with the best answers, they come with the best questions.

Successful people have known this for centuries. We call it curiosity.

The Question Age

John Deere was not an industrialist when he ushered in the second industrial revolution. He was a blacksmith. What separated Deere from the millions of blacksmiths whose jobs he ultimately made obsolete was curiosity. He wondered, what if there is another way? Why does steel bend when it hits a rock? What would I need to know to harden it? And then, he acted on the answers.

This pattern repeats throughout history. It repeats throughout our lives, our communities, our workday. The most successful people are the most curious. They have questions everywhere, because answers can be anywhere. The costs of curiosity are lower than ever. We don’t need to travel long distances to access books anymore. The ideas of billions of others are at our fingertips.

What I like about answers is they give us new questions. David Deutsch has argued that curiosity is a fundamentally human trait. What separates our minds from the idle curiosity of, say, a cat inspecting something new, is we are the only species that seeks to explain things which are already familiar. How does this change when each of us has a second brain? What about a second brain that is constantly part of and deeply familiar with our lives and all of the world's knowledge? The simple answer is we would use it to ask more questions.

This is why we built Comet, an AI-powered browser. Smarter critics have noticed it is more of an AI-powered assistant. In reality, it is a second brain. And right now, a browser just happens to be the best first place to deploy it. The technology of curiosity will remove barriers that suppress our natural desire to explore. In an age where AI holds nearly limitless information, the people who thrive will be those who never run out of questions. The great industrialists, the leading capitalists, the winners of this new era will all be the most curious.

Can there ever be an algorithm that knows the entire universe?

If so, it would know everything. It would know more than just the details that have been printed into factual records or listed on the Internet. It would know tacit interpersonal knowledge. It would know what’s happening right now, with who, what will happen next, and why. In one view, to ask this is to imagine building or being a God. In another, it is to ask what we are capable of. The truth is any view of this question is proof of Deutsch's point: we can't help but try to explain why someone would ask it. That’s curiosity.

I like to ask it because questions (and their answers, the secrets of the universe) are limitless. The saying that it is better to pursue wisdom than wealth was written many centuries ago in the Rigveda, and in the Book of Proverbs, attributed to King Solomon. The Rigveda explains the superiority of wisdom and righteous conduct (Dharma) over the mere pursuit of wealth (Artha). Regardless which text you prefer, they are proof that learning, and curiosity are so deeply ingrained in our history that they will always be central to our future.

Regardless which text you prefer, they are both proof that questions, learning, and curiosity are so deeply ingrained in our history that they will always be central to our future. As we accelerate into it, both AI and humans will need precisely what Perplexity is building right now. Answers, everywhere, to more questions.