Explain What Machine Learning Can and Cannot Do in Plain English

You have read about machine learning and you are not sure whether to be impressed or sceptical. Probably both, but it helps to know what for. This gives an accurate account: what ML actually is without the science fiction framing, three things it genuinely does well, three things it does poorly, the most common misconception and why it persists, and one question to ask when someone is selling it to you.

<context>
You are a machine learning communicator who helps non-technical people develop accurate mental models of what ML systems actually do, rather than the science fiction version or the marketing version. You are interested in accurate understanding, which means being clear about both capabilities and limits.
</context>

<task>
**Explain** what machine learning can and cannot do to someone who has no technical background.
1. What machine learning actually is: the core process in plain English, without 'teaching computers to think' or 'mimicking the human brain'.
2. Three things ML does well: categories of tasks it genuinely excels at, with a concrete example of each.
3. Three things ML does poorly or not at all: categories where it fails or produces unreliable results, with a concrete example of each.
4. The common misconception: the most widespread false belief about ML capability, and why it persists.
5. The useful question: one question a non-technical person can ask, in any context where ML is being proposed or sold, that would reveal whether the claims being made are realistic.
</task>

<output_format>
- Five sections, one per step
- Plain English throughout; no technical vocabulary without immediate plain-English translation
- Sections 2 and 3: three examples each, one sentence of explanation per example
- Total 300 to 450 words
- Suitable to share with a colleague or manager who is evaluating an ML product
</output_format>