You do not need to learn anything technical. You type, it answers. What follows is the fastest way to find out whether it is any use to you.
The single most useful thing you can do first is give it a real piece of your own work whose answer you already know. Not because you need the answer — because that is how you learn how far to trust it.
Try three or four of these:
Here is a paragraph from a manuscript. Make the English clearer without changing the meaning.
Summarise this document in five bullet points.
Draft a short, friendly email to an editor asking whether they can take a job this week.
I need to explain to a customer why their manuscript needs a second round of editing. Help me say it without sounding critical of their writing.
What are three different ways I could phrase this sentence?
Catching it being right is how trust forms. Catching it being wrong is how you learn its limits. Both are useful; neither happens if you only ask it things you cannot check.
The biggest difference between people who find AI useful and people who do not is that the first group keeps talking to it. A first answer is a draft, not a verdict.
Shorter. Too formal — make it sound like a person wrote it. Keep the second paragraph, rewrite the rest. Why did you change that word?
It knows nothing about you unless you say. A request with context gets a far better answer than one without.
Weak: "Write an email about the deadline."
Strong: "I'm a project manager at an academic editing company. Write a short, warm email to a customer explaining that their manuscript will be ready on Friday instead of Wednesday because the editor asked for more time. Apologise once, not three times. Don't offer a discount."
The second one takes fifteen seconds longer to type and saves you a rewrite.
It responds well to constraints:
This is a trial. "I tried it twice and went back to doing it the old way" is a genuinely useful report — more useful than polite silence. Tell Daniel, in whichever language you prefer.