AI is very good at taking friction out of thinking. That’s not an insult, it’s the point. But in my experience, if everything feels easy, you should probably be worried.
In Software Testing, it can be very easy to get by doing very little, but on the flip side, it can feel excruciatingly difficult to stand out. AI fits into this perfectly. It doesn’t just offer shortcuts; it industrialises them. Why wrestle with ambiguity when an AI can generate 20 test cases in seconds? Why sit with uncertainty when you can get an instant answer that sounds confident? The danger isn’t that AI is wrong. The danger is that it removes the discomfort that forces us to think.
Friction is Where Understanding Lives
The most valuable moments in my career haven’t been the “efficient” ones. They are the moments where a requirement feels “off”, but I can’t yet explain why, or a risk nags at me even though the tests are green. That friction is not wasted effort. It’s where understanding forms. When we struggle to articulate a test scenario, we’re often discovering a gap in the product. When we argue with ourselves about an edge case, we’re advocating for a user who isn’t in the room.
QA is an Advocacy Role
I feel like this phrase has become extremely buzzwordy to the point of irritation. But its something that i still strongly advocate for and its become more important in this new age of AI. At its core, QA is not about coverage or tooling. It is about advocacy. We represent the users who don’t behave “correctly,” the edge cases the business didn’t consider, and the failure modes no one wants to talk about. That advocacy requires judgment, context, and thinking, the exact things that are easiest to outsource. If we blindly accept AI-generated tests, we’re not just saving time; we’re handing over responsibility. And AI doesn’t carry responsibility. We do.
How to get the most out of AI (without losing your edge)
The goal isn’t to avoid AI; it’s to use it in a way that sharpens your thinking instead of replacing it. Here are the tactics I use:
- Think first, prompt second: Before asking AI for ideas, spend five minutes thinking through the problem yourself. Even rough ideas create a baseline you can compare against. If you don’t have a hypothesis before you prompt, you’re just spectating.
- Challenge the output: Treat AI like a junior team member who speaks with unearned confidence. It’s useful and fast, but it requires your review and your context.
- Use it for speed, not certainty: AI is excellent at accelerating repetitive thinking—generating test data, drafting API payloads, or creating variations of scenarios. But speed should never be confused with correctness.
Keep the user in the centre: The best QA work comes from empathy. AI can simulate scenarios, but it doesn’t feel frustration, confusion, or inconvenience the way users do.
Don’t remove the friction too soon
If everything feels easy, that should worry us. Good testing often feels uncomfortable, uncertain, slow, and mentally tiring. That’s not inefficiency; that’s responsibility. AI can help us move faster after we’ve thought. But if we let it think for us, we risk losing the very thing that makes QA valuable.
Friction isn’t something to eliminate. It’s something to learn from.
