Using AI Without Being an Expert

Get on the train

The line that stuck with me: the AI train is coming. Either get on board or get run over.

It's here. It isn't going away. If you've been in one place for fifteen years and never opened one of these things, that's fine. Open one this week. It's a tool in your toolbox, and right now it's the one everyone's going to ask you about.

But here's the other saying: when all you have is a hammer, everything looks like a nail. AI is not your only tool. And the most predictable thing about it is that it will make a mistake, and it will sound confident while it does.

The first two questions

Open whichever one you like. They're all fine for this. Type:

"What can you do for me?" Let it tell you. You'll be surprised how much of a job search it can help with: rewriting a bullet point, explaining a term from a job posting, quizzing you on interview questions, checking your grammar.

Then, if you're in tech, the second question: "When was your training data last updated?" or "What's your knowledge cutoff?"

That one matters more than it sounds. Here's why.

The trap

Interviewers know people are using AI to write their take-home projects. So some of them set a trap. They'll include something in the assignment that changed after the AI's training date: a library update, a bug fix, a new way of doing something. If your project was written by the AI, it'll do it the old way, confidently, and you won't know. You'll turn it in with the tell sitting right there, and they'll know your code wasn't yours.

I've watched it from the other side. A candidate can't make a two-line change to a project he supposedly wrote. That's the end.

So: do the work yourself. If you get stuck, ask the AI a question, the way you'd ask a coworker. Don't paste in the assignment and paste out the answer.

Use it as a teacher, not a ghostwriter

Here's the pattern that actually works. Write the thing. Then hand it to the AI and say: review this. Find the mistakes. Tell me what a reviewer would flag.

I did that with my résumé, years into using the same one, and it found tense errors in every job description. Sentences that started in the present and ended in the past. I'd read that résumé a hundred times and never seen them. It found every one in thirty seconds.

Same with a take-home project. Build it, then screenshot it or paste it in and ask for a code review. Same with your answer to "tell me about a time you dealt with a difficult customer." Write the answer, then ask what's weak about it.

It's a teacher grading your homework. It is not a classmate you copy from. The difference shows up in the interview, every time.

What it's not for

It's not for writing your résumé from scratch. What comes out will sound like everyone else's, because it is everyone else's. That's why the AI in Bridani only works on small pieces: it reviews what you wrote, suggests where it's weak, and hands it back. You still have to make it yours.

It's not for applying to a hundred jobs a day. See the last post.

And it's not for answering interview questions in real time with the chat window open next to the call. People are doing that. It's obvious. Don't.

One more thing

Some of the bigger companies have started forbidding their junior developers from using AI at all, because those developers weren't learning to code. They were learning to write prompts.

You, with twenty years, learned it the hard way, out of books. That turns out to be an advantage. More on that next.