← AI is Easy · Learn
AI is Easy

What AI cannot do (yet): five honest limits

Hype makes it sound like AI can do everything. Knowing where it stops saves you from the bad surprises.

1. It can be confidently wrong

It produces fluent text whether or not it is true. Check anything that matters.

2. It does not know your life

It only knows what you tell it in the conversation. Leave out a key detail and the advice may not fit you.

Still reading? Get the opening chapter free, no charge.

3. It has no real judgement

It can list pros and cons. It cannot tell what matters most to you or take the consequences of a choice.

4. It does not take responsibility

If you send an email or sign a document because of what it said, the responsibility is yours.

5. It may be out of date

Unless it can search the web, its knowledge stops at a point in time. Ask for sources and check dates.

The good news

Within these limits it is a patient helper for drafting, learning and planning, and the limits are easy to work around once you know them.

Click here to make this site a preferred source in your Google AI search results — you will see our plain-language AI guides prioritised when you search for topics we cover.

Want AI explained like this, all the way through?

Ten short books, one evening each, in the same plain words as this article. No jargon, no tech background needed.

Get the whole set · $14.99Or one book · $1.99

Not ready? Get the opening chapter free →

More plain-language guides →