Remember your first visit to a large library?

There were thousands — perhaps millions — of books surrounding you. Somewhere in that enormous collection was the information you needed, but you could not possibly read every book or search every shelf.

At first, the library may have felt overwhelming. Then you learnt how it worked.

You used the catalogue, found the right section and selected the books that mattered. You carried those books to a table and created a working space containing the information you needed.

That is a useful way to understand artificial intelligence.

The world of AI contains more information than any person could read, remember or examine in a lifetime. But AI does not consider all of that information every time you ask it a question.

Like someone working in a library, AI identifies information related to the problem and brings it into a limited working space. Your instructions help to determine which shelves it visits, which books it selects and what it puts on the table.

Once you understand that, AI becomes much less intimidating.

You do not need to know everything in the library. You need to know what you are trying to accomplish, provide the right information and ask useful questions.

With practice, you stop merely operating AI. You begin to be AI — working with it naturally, like working with books spread across a library table.

The three P’s of AI

To work effectively with AI, use three steps:

Prime. Prompt. Perfect.

AI without the three P’s is like hiring the world’s fastest researcher, giving them no background, asking a vague question and publishing their first answer without checking it.

Without priming, the researcher does not understand the situation.

Without prompting, the researcher does not know what you need.

Without perfecting, mistakes can pass straight into the finished work.

AI without the three P’s is extraordinary capability working without direction or supervision.

Prime

Priming means giving AI the information it needs before asking it to solve the problem.

This is where many people go wrong. They type one vague question, receive a vague answer and decide that AI is useless.

Imagine walking into a library and asking the librarian:

Do you have a book that can help me?

The librarian would naturally ask: help you with what?

AI has the same problem. It cannot know the parts of your situation that you have not shared.

Priming means putting the right books on the table before beginning the work. Give AI the background, facts, documents, objectives and limitations that define the problem.

Suppose I am involved in a landlord-and-tenant dispute. I might give AI the lease, relevant e-mails, payment records and a simple timeline.

These are the records I have for this case. Organise the facts, build a timeline and tell me what information appears to be missing.

That is priming.

I have brought the relevant books into the reading room and placed them on the table. AI can now see the dispute as a connected body of information rather than as one isolated question.

Good priming does not mean dumping everything you own into AI. Just as in a library, the material on the table should relate to the work being done.

More relevant context usually produces a better answer. More irrelevant context can bury it.

Prompt

Once AI has been primed, you begin working through the problem with it.

What are the strongest and weakest parts of my position?

What argument is the other side likely to make?

Each prompt moves the work forward.

You do not need to create one enormous, perfectly worded question. There is no magical sentence that suddenly unlocks AI.

Have a conversation. Examine each answer, correct misunderstandings, add information and ask the next question. Let each answer help reveal what should be asked next.

This is where AI begins to feel less like operating software and more like thinking with another intelligence.

You provide the purpose, experience and judgment. AI helps organise information, identify patterns, expose contradictions and explore possibilities.

You are not asking AI to think instead of you.

You are thinking together.

Perfect

The first answer is not the finished answer.

AI can misunderstand evidence, overlook contradictions and confidently state something that is wrong. In legal work, an invented court decision or an incorrect interpretation of the law could completely change the apparent strength of a case.

The legal example deliberately takes us into AI’s grey zone.

AI can organise the evidence, construct a timeline and suggest possible arguments. It may also help locate relevant laws and court decisions.

But should you walk into court and rely completely upon its answer?

Absolutely not.

This is where the third P — Perfect — becomes essential.

Ask AI to examine its work:

Which statements come directly from my evidence, which are your interpretation, and what might you have wrong?

Then ask it for real sources and check those sources yourself.

You can ask AI how confident it is, but confidence is not proof. AI can be confidently wrong. Important claims must be checked against the original documents and reliable sources.

There is also an important limit: AI examining its own work is still AI examining AI.

Where a mistake could cause real harm, a qualified professional should examine the result.

I use a little sniglet I made up to remember this: Progentic

Progentic means AI performs the work, but a human professional reviews the results.

In the landlord-and-tenant example, AI may sort the records, build the chronology, find contradictions and suggest arguments. The lawyer does not need to begin with a disorganised pile of documents. The lawyer can examine the organised work, verify the law and apply professional judgment.

The same principle applies to medical, financial and engineering work. The greater the consequence of an error, the more important professional review becomes.

AI does not eliminate expertise. It allows expertise to work across a much larger body of information.

Be AI

The first time you entered a large library, you may not have known where to begin.

Eventually, you learnt the system. The size of the library stopped being intimidating because you did not need to read everything. You only needed to locate the right material and create the right working space.

AI is much the same.

The goal is not to surrender your thinking to AI. The goal is to combine your purpose, knowledge and judgment with its ability to work across enormous amounts of information.

With practice, the process becomes natural — like working with books spread across a library table.

Prime the information.

Prompt the thinking.

Perfect the result.

That is how you stop merely using AI — and begin to Be AI.

• Jamie Thain is an information technology professional and inventor and uses artificial intelligence every day