Day 3: What AI Is Good At (And Where It Fails)

The Concept

One of the most useful things you can do before relying on any tool is understand where it performs well and where it does not.

That sounds obvious, but with AI it is easy to skip this step. The tool is so flexible that it feels like it should be useful for almost anything. You can ask it to write an email, explain a policy, brainstorm a birthday gift, compare software, debug a spreadsheet formula, review a resume, plan a trip, or summarise a report. Because the interface is always the same, every task can start to look equally reasonable.

They are not equally reasonable.

People who get the most from AI are not the ones who trust it for everything. They are the ones who have a clear, honest picture of when AI adds genuine value and when it is the wrong tool, or only a partial tool, for the job.

Today's goal: build your personal map of AI's strength zone and caution zone.

By the end of this lesson, you should be able to look at a task and ask: Is this a good AI task, a cautious AI task, or a task where I should use another source entirely?

The strength zone

AI is strongest when the work involves language, structure, explanation, or momentum. These are the places where it can save real time and improve the quality of your thinking.

1. Drafting and rewriting

AI is very useful when you need a first draft or a better version of something you have already written.

That might mean:

• a difficult email,
• a short announcement,
• a project update,
• a social post,
• a meeting agenda,
• a reply to a customer,
• or a clearer version of a messy paragraph.

The value is not that AI writes the final version for you. The value is that it gets you out of the blank-page stage quickly. You can then edit the draft with your judgment, tone, and context.

Why this works: AI is trained on patterns of language. If you give it the audience, purpose, tone, and constraints, it can produce a useful starting point much faster than most people can begin from scratch.

2. Summarising and extracting

AI is strong at turning long or messy information into something shorter and more usable.

You can paste notes from a meeting and ask for action items. You can paste a long article and ask for the five most important ideas. You can paste rough thoughts and ask for themes. You can paste a customer message and ask what the person seems to need.

This is one of the most practical everyday uses because the output is easy to review. You already have the source material, so you can compare the summary against it.

Good use case: "Summarise this into five bullets, then list any decisions, deadlines, and open questions."

3. Explaining and simplifying

AI is excellent at explaining complex material in plain language.

If you receive a dense policy, a technical article, a medical leaflet, a finance document, or a confusing internal memo, AI can help you make sense of it. You can ask for an explanation at your level. You can ask for analogies. You can ask it to explain the same idea three ways until one version clicks.

The key is to treat the explanation as a learning aid, not as the final authority. If the topic has consequences, use AI to understand the issue and then verify important details from a reliable source.

Plain English: AI is very good at making hard material easier to approach. It is not automatically responsible for deciding what is true.

4. Organising scattered information

AI is especially helpful when you have information, but it is not yet organised.

This might look like:

• turning notes into an outline,
• grouping ideas into categories,
• converting a conversation into next steps,
• creating a checklist from a paragraph,
• identifying themes across several comments,
• or turning a messy plan into a sequence.

This kind of work is valuable because the human bottleneck is often not intelligence. It is the effort required to impose structure. AI can do the first structuring pass quickly, and you can decide what to keep.

5. Brainstorming options

AI can generate many ideas quickly. That makes it useful when you need range before judgment.

For example, you might ask for:

• ten possible titles,
• five ways to explain an idea,
• three project approaches,
• customer objections,
• risks in a plan,
• or alternative ways to solve a small workflow problem.

AI brainstorming is best when you do not accept the first list as final. Ask it to improve the list, remove obvious ideas, make the options more practical, or organise them by effort and impact.

Better prompt: "Give me 12 ideas, then separate them into safe, ambitious, and unusual options."

6. Comparing and clarifying trade-offs

AI can help you compare options by making trade-offs visible.

It can create a table of pros and cons. It can identify what might go wrong. It can show which option is faster, cheaper, simpler, or more flexible. It can ask questions you should answer before deciding.

This is useful because decisions often feel foggy when everything is still in your head. AI can turn the decision into something you can inspect.

Important boundary: AI can clarify a decision. It cannot fully make the decision for you, because it does not live with the consequences.

The caution zone

AI's weaknesses matter most when the output feels polished. A weak answer that looks weak is easy to reject. A weak answer that sounds confident is more dangerous.

1. Specific facts

AI can get facts wrong. It can misstate names, dates, statistics, policies, citations, and definitions. It may also blend true and false details in the same answer, which makes the mistake harder to spot.

This does not mean you should never ask factual questions. It means you should decide whether the fact matters.

If you are asking, "What is the rough idea behind this concept?" AI may be a helpful starting point. If you are asking, "What exact rule applies to this contract, medical issue, tax question, legal deadline, or public claim?" you need a reliable source.

Use this rule: If a fact will be acted on, published, paid for, or used to advise someone else, verify it.

2. Recent or changing information

AI may not know what changed recently. People change jobs. Product prices change. Laws change. Tool features change. Companies launch, merge, rename, or shut down. A model may describe the world as it existed during training, not as it exists now.

For current information, use AI carefully. It can help you make a checklist of what to look up, compare sources after you provide them, or draft questions to ask. But do not assume it knows the latest state of the world.

3. Numbers and calculations

Language models are not reliable calculators by default.

Some AI tools are connected to calculators, code interpreters, spreadsheets, or other systems that improve numerical reliability. But if you are simply chatting with a model, do not assume arithmetic, percentages, totals, or projections are correct.

This matters in budgets, prices, dates, schedules, measurements, forecasts, and anything involving money.

Simple habit: Let AI help set up the calculation, then check the numbers independently.

4. Personal judgment

AI can help you think through a personal decision, but it cannot understand your full life.

It does not know your relationships, emotional history, values, risk tolerance, financial situation, health context, family dynamics, workplace politics, or long-term priorities unless you explain them. Even then, it only has the version you provide.

That makes it useful as a mirror, not a replacement for judgment.

Ask it to surface trade-offs, generate options, or help you prepare for a conversation. Do not let it decide what matters most to you.

5. High-stakes advice

Medical, legal, financial, employment, safety, and compliance topics need extra care.

AI can be helpful in these areas when used correctly. It can explain terms, prepare questions, summarise documents you provide, or help you understand a process. But it should not be the final authority.

Better use: "Help me understand this letter and list the questions I should ask a qualified professional."

How to classify a task

Before reaching for AI, ask three quick questions:

1. Is this mainly a language, structure, or ideas task? If yes, AI is probably worth trying.
2. Does the answer contain facts, numbers, or current information that matter? If yes, use AI as a starting point and verify.
3. Does the decision depend on personal context, values, risk, or consequences? If yes, use AI to clarify, not decide.

This simple filter will prevent most common mistakes.

Use this today

Choose one real task from your week. It should not be hypothetical. Pick something you actually need to do.

Then classify it:

• Strength zone: drafting, summarising, explaining, organising, brainstorming, or comparing.
• Caution zone: facts, numbers, current information, professional advice, or personal judgment.
• Mixed task: useful for AI, but with at least one part you need to verify.

If it is in the strength zone, use AI today. If it is in the caution zone, use AI only to prepare, understand, or organise your next step. If it is mixed, write down exactly what you will verify.

Remember this

If you remember nothing else from Day 3, remember these three ideas:

• Use AI freely for language, structure, and momentum.
• Verify facts, numbers, and current information before acting on them.
• Let AI clarify personal decisions, but do not outsource your judgment.

That balance is the difference between mature use and blind trust. It is also the difference between under-using a genuinely helpful tool and expecting it to do work it was never built to do.