AI literally can't replace you.

It's not good enough to do that yet, and maybe it never will be. If you're one of the people hiding under your bed in fear of AI, start listening to people other than "hype bros".

Plenty of perspectives on this will tell you AI is better as an assistant taking orders from you than the other way around.

You have goals.

AI serves those goals. It doesn't give you goals, it helps you chart a path to achieving them.

Just think for a moment, seriously. Why would you delegate the ability to decide what you want out of your own life?

Yeah, so how can you use AI to learn QA skills?

(a reader asked me to write about this, so here we go)

If I had to learn a new skill, here's how I'd use AI to do it:

1. Use AI to decide which skill to learn 📊 🗺️

Tools like Perplexity are great for researching the market to determine what to learn.

Let's say I want to learn API testing. That tells me literally nothing about what technologies to learn.

I could write API tests that use:

• Any of 5 common QA Automation programming languages (C#, Java, Python, Ruby, JS/TS)
• Manual testing tools like Postman or Insomnia to store and send requests
• CI tools like Newman or Azure DevOps Pipelines to run API tests as part of a triggered workflow
• Contract testing, Response Validation testing, or even comparisons between response data and the corresponding database records
• Headless E2E user workflows or isolated endpoint validation

So you tell me, which one do you start with?

Yeah, thought so. As always, it depends.

You can tell AI what your situation is, the state of API testing at your company right now, your skill set, etc., and it can give you the best recommendations for your specific situation.

It can even give you a roadmap:

2. Use AI as a rubber duck 🦆

As you're trudging down the path of learning a new skill, you'll want help.

You won't always have a Senior QA or a mentor around to ask questions to, or maybe you want to do this on your own without disturbing anyone.

Either way, AI is great as a tutor.

You can bother AI all day and it doesn't care. Use that to your advantage.

3. Use AI + Docs for maximum accuracy 🤖🤝📄

AI hallucinates. That's not news to any of us in 2025.

So what do you do about this?

When you ask a question about Playwright Locators, for example, look up the docs on Locators.

Copy/paste something that teaches AI what you're talking about so it makes fewer "educated guesses".

AI may be trained on the entire public internet of data, but when you ask it a question, any context you provide is prioritized over its existing data set.

So if it thought it knew what a Locator was before you gave it documentation, now it definitely knows.

This is especially helpful when you get trapped in a loop of AI trying unsuccessfully to solve your problem with the same solutions, like Whack-A-Mole.

When it seems to misunderstand something, feed it information like you're giving a student a critical piece of information to finish their homework assignment.

Speaking of API testing, I gave a live workshop on it yesterday.

It was exclusively for my Social QA Bootcamp students and it cost them nothing to attend.

I even recorded it so they could rewatch it later or people who couldn't make it could still participate.

We messed around with Playwright and TypeScript and they had an optional homework assignment that would help them become more familiar with API testing tools Playwright offers.

If you want to be in the next cohort when I relaunch in May, hop on the list.

​>>> Join the Cohort 2 Waitlist​

Just like last time, the waitlist gets first dibs and there will be limited seats. When they're gone, they're gone. No exceptions.

What skill can you use AI to learn by the end of Q1? I'd love to know in a reply.

Cheers,

Steven