Three questions to answer before you open ChatGPT

The Most Common Mistake

Here is a scene that plays out thousands of times a day in offices, classrooms, and home studios around the world: someone opens ChatGPT, Claude, or Copilot, stares at the blank prompt box, and types something like "Help me with my presentation" or "Write something about our Q3 results."

Then they spend twenty minutes going back and forth, getting outputs that are not quite right, rephrasing their request, feeling vaguely frustrated, and eventually either giving up or accepting something mediocre.

The problem was never the AI. The problem was that they reached for the tool before they knew what they actually wanted.

This is the most common mistake in AI usage, and it is the one that costs the most time. It is the equivalent of walking into a hardware shop and saying "I need a tool" without knowing whether you are hanging a picture or building a shed. You might walk out with a hammer either way, but only one of those situations calls for it.

Before You Touch Any AI Tool, Answer Three Questions

The Recognise stage of the READY Method asks you to pause and map the problem before you think about solutions. Specifically, you need to answer three questions:

1. What exactly am I trying to achieve?

Not vaguely. Not "I need to write something." Be specific. "I need a 500-word summary of this 40-page report that highlights the three main risks for a non-technical audience." That level of clarity changes everything — not just for AI, but for any work you do. If you cannot articulate what "done" looks like, you are not ready to delegate to anyone, human or machine.

2. What does "done well" look like?

This is about quality criteria. A first draft that captures the key ideas is very different from a polished, publication-ready piece. A rough data analysis that identifies trends is very different from a rigorous statistical model. Knowing your quality bar helps you decide how much human oversight the task needs, and whether AI output will require light editing or heavy reworking.

3. What kind of thinking does this task require?

This is the crucial one. Not all work is the same, and not all of it is equally suited to AI. Let us break it down.

Four Types of Work

Think about the tasks you do in a typical week. They generally fall into one of four categories:

Routine and Repetitive Work These are tasks with clear rules and predictable patterns. Formatting documents, cleaning data, sending standard communications, converting information from one structure to another. The thinking required is procedural — follow the steps, apply the rules, produce the output. This is often where AI delivers the most immediate, obvious value.

Creative and Exploratory Work These are tasks where you are generating ideas, exploring possibilities, or producing something original. Writing a marketing campaign, brainstorming solutions to a problem, designing a new process. The thinking here is divergent — you want breadth, novelty, and unexpected connections. AI can be genuinely useful here, but in a very different way than with routine work. It is a sparring partner, not an executor.

Analytical and Judgement-Heavy Work These are tasks that require you to weigh evidence, make assessments, and form conclusions. Evaluating a business proposal, diagnosing a technical problem, deciding between competing strategies. The thinking is convergent and often relies on domain expertise, experience, and contextual understanding. AI can support this work — by gathering information, surfacing patterns, or stress-testing your reasoning — but the judgement itself remains yours.

Sensitive and High-Stakes Work These are tasks where the consequences of getting it wrong are significant, where confidentiality matters, or where human relationships are at the centre. Giving someone difficult feedback, making a hiring decision, handling a complaint, navigating a legal or ethical grey area. The thinking here is deeply human — it involves empathy, accountability, and an understanding of context that AI simply does not have.

Your Expertise Is the Foundation

There is a tempting narrative that AI can make anyone an expert at anything. It cannot. What AI can do is amplify the capability you already have.

If you are a skilled data analyst, AI can help you process data faster, spot patterns you might have missed, and automate your reporting. If you know nothing about data analysis, AI might produce output that looks impressive but contains fundamental errors you would never catch.

This is not a reason to avoid using AI in areas where you are still learning — quite the opposite. AI can be a superb learning tool. But you need to be honest with yourself about where you are on the expertise spectrum for any given task, because that determines how much you can trust the output and how much verification it needs.

The Recognise stage is about this honesty. It is about looking clearly at what you need to do, what kind of thinking it requires, and what you bring to the table — before you decide whether and how AI fits in.