Knowledge
towards zero.
Where value moves.

A warning worth taking seriously.

In the interview below, Raoul Pal argues that abundant intelligence undermines the scarcity that once made specialist knowledge expensive. Lawyers are his example of people paid for what they know. It is an uncomfortable idea, especially if you have spent years becoming good at your work.

I want my colleagues to take that discomfort seriously. We should be asking where our value comes from before familiar tasks become cheap, fast and widely available. My purpose as an AI Lead is to help people make that transition: understand what is changing, experiment with it and create value in the work that follows.

I read “knowledge towards zero” as a warning about the price of reproducible answers and routine outputs. It is not a measured prediction that every kind of knowledge will become worthless. AI still costs money, produces mistakes and needs context. But if our contribution is mainly supplying an answer someone else can generate, the pressure on that contribution is real.

The lawyer and the leaking pipe.

Imagine a service built around reading documents, finding information and producing a standard first draft. AI changes the economics of that work. A client may need far fewer billable hours to get the same initial output. The professional whose offer depends heavily on those hours has a reason to rethink it.

That is different from saying lawyers are no longer needed. A legal matter can involve incomplete facts, competing interests, negotiation and consequences for a person’s life. The quality of an answer has to be assessed. A Stanford study of legal research tools found that specialised systems could still generate incorrect information. That study concerns the tools tested in 2024; it is not a current score for every system. ¹

Now consider someone repairing a leaking pipe in an unfamiliar house. Describing the repair is one part. Reaching the fitting, dealing with what is actually there and leaving a working installation are other parts. A text answer does not carry out those physical actions. For now, that is a meaningful difference between generating knowledge and delivering a hands-on result.

This is about the shape of a task, rather than a ranking of people by education. Practical trades contain knowledge, diagnosis and skill too. Routine physical work can also be automated, and robotics will keep developing. I would not promise any profession permanent protection. I would look closely at which parts of its work technology can actually perform.

Tasks change before job titles do.

The ILO’s 2025 assessment found that one in four jobs worldwide had some potential exposure to generative AI. It stresses that exposure is not a count of jobs lost, and that transformation is more likely than complete replacement. Jobs contain different tasks, many of which still require human involvement. ²

That is useful context for a warning like this one. A profession can continue to exist while its economics, entry-level work and daily routines change significantly. We need to examine the work underneath the title, and give people ways to move with those changes.

For me, the question is practical: if the first answer becomes easier to obtain, where can I contribute to a result that someone actually needs? That is where I would put my attention, both in my own practice and when helping a team.

Where I would build value.

In design, I can ask AI for options, wording or visual directions. My contribution begins with understanding the problem: who needs this, what should change for them and what constraints matter? It continues in deciding which direction deserves to be developed and recognising when an attractive result misses the point.

I want to strengthen that ability to frame, judge and finish work. It means understanding a brand, checking a result against real requirements, involving the right people and owning the choices we make. These abilities need practice. Calling them “human skills” does not automatically make us good at them.

The OECD’s 2026 skills report highlights critical thinking, creativity and collaboration alongside foundational and digital skills. It also distinguishes exposure to AI from the likelihood of a whole job being automated. That supports investing in a broader contribution than operating one tool. ³

I also see value in making a repeatable way of working: clear inputs, meaningful review points and a dependable handover. If I can help colleagues reach a better result with fewer unnecessary steps, I have contributed something we can use again. The value is in the working result and how reliably we reach it.

A first step for colleagues.

I would start with one recurring task from an ordinary working week. Write down the input, the output, who uses it and what makes the result good. Then separate the parts: gathering information, generating a draft, judging it and turning it into something useful.

Test one suitable part with AI using approved tools and appropriate material. Compare it with the current approach. Include the time needed to check, correct and finish the output. A quick draft is only progress if the complete task improves.

Use what you learn to choose one capability to develop: writing a clearer brief, recognising errors, understanding the end user, connecting steps in a workflow or explaining a difficult decision. Make that learning part of actual work, and share the examples with a colleague.

I want to help people do this through practical experiments and honest conversations. Start small enough to learn, make the result visible and decide together what to change next. There is room for people who are curious, people who are worried and people who need more time to understand the tools.

Give people a way forward.

Leaders have a responsibility here too. Colleagues need time to learn, useful access to tools and clarity about how their roles may change. If a task disappears, someone has to help work out the next contribution. “Become more valuable” means very little without the opportunity to practise new work.

I am enthusiastic about what AI makes possible, and I take the disruption seriously. Both are part of why I want to be involved. I want to warn my colleagues early enough that we can act, and help them find a place in the transition.

If knowledge becomes abundant, we have to become better at turning it into something that matters. Understand the need. Make the choice. Deliver the result. Help someone else do the same. That is the direction in which I want to move.