Your AI Transformation

How’s your AI transformation going?

Not your company’s AI transformation, I already know how that’s going. Pretty similar to their agile transformation I imagine. Or even their digital transformation, my personal favourite wheeze, something so vague that you could transform into, well, anything.

Anyway, AI transformation within your company. Stop me if I’m wrong:

• Someone with an urge to try out new things introduces AI to your company, starts a slack channel maybe. Or a consultant has sold your executive the future of software development where it’s much cheaper.
• There is a piecemeal, disjointed attempt at using AI, usually with a side of naivety regarding where data is flowing to.
• Tools get agreed, usually whatever comes with whatever companies you already use for other tools, everyone gets access, very few people really know what they are doing.
• No one cares about usage, as you pay per seat, rather than by usage.
• AI companies realise they need to make money, so they change their business model to token consumption. Your company applies the brakes after a big bill.
• Sheep dip training on prompt engineering and model selection. None on evaluating which activities to actually use AI for.
• All the wrong activities are outsourced to AI, leaving a judgement gap.
• None of the things that actually slow your business down (usually outside of software development) are transformed, just like your agile transformation. See decision making, Figma dependency, compliance psychosis, pretty much all project management (hint, work on the risks people), distance from customer, release testing theatre, crystal ball activities like budgeting and estimation, I could go on.
• The software development function frets about token usage and begins malfunctioning.
• Profit. And by profit I mean fail.

I’m talking about something far more interesting. Your own personal AI transformation. Not into AI (although that might be what the tech bro’s have in mind for us) but how you use it in your own work. Now, I’m one of those weirdo’s who thinks that technology should assist me to do the boring stuff, while letting me express myself with the things I find interesting.

So, my own AI transformation:

• Writing a script for uploading test data, a webhook to enable testing and other boring tasks that help you to do the fun parts. I use AI for this.
• Understanding old code bases in languages I am unfamiliar with. I use AI for this, as the number of barriers to my understanding is higher, there is unlikely to be anyone around who was there when it was written, so I need a hand to get started.
• Writing unit tests. I don’t use AI for this. If you ask an AI to write the code and then write the tests, I can’t think of a more bankrupt way to develop software. Adding unit tests after writing the code was always inferior, but not writing the application code and then not writing the tests is a complete abdication. Write some unit tests then ask the AI assistant to generate application code is better. I would accept filling in any unit test coverage gaps with the assistance of the AI, after writing your own first.
• Writing test automation (distinct from coming up with what to test and designing those tests). Personally, I use my developers and tell them to use AI. If you as a quality engineer are sitting around using AI to write Playwright test automation, you’ve got it all the wrong way up. If I never have to write a Playwright test again, I’ll be delighted.
• Contributing to a story kick off, brainstorming test ideas, designing tests, reviewing an architecture for testability, I don’t use AI. This is the stuff of life, where you make a big difference as a quality engineer. Why outsource this stuff? Keep your skills sharp. If you don’t know how to do these things, find out, it will be a transformation in its own right.
• Performing the most valuable activity you can as a quality engineer, meaning exploratory testing. A well designed, thoughtful exploratory testing session which surfaces assumptions, bugs, tests claims and is communicated well to your team is a revolutionary act in my opinion. I do not use AI for this.
• Summarising some of the activities above to update Jira and other such information deep freezes demanded by your organisation. I use AI for this. In fact, if you can argue that it shouldn’t really be needed, use AI for it.
• Summarising your testing into a compelling narrative used to talk to your team and customers. I don’t use AI for this. It’s your big chance to get everyone on board. Everyone now has an AI detector, which will go off immediately and just watch those eyes glaze over.
• Writing blogs. I don’t use AI for this. I like writing, why would I deny myself what I enjoy doing for the sake of speed and convenience over expressing myself.

I could go on, but I won’t as you hopefully get the idea, practice the things that make you as a quality engineer special, consider AI use for the aspects of the role you enjoy less. New technologies often ask us to transform how we work and always will, just make sure you don’t transform the value out of your role.

Imagine if your role as a quality engineer was reduced to reviewing AI output? Does that sound like a future you want? I don’t.