AI is here whether we are ready or not. AI is increasingly used to create test plans, perform simple test automation, and even pass ISTQB tests. These AI tools will shine when experienced testers can instill their testing expertise into the AI.
Trained on Public Data
Popular and general-purpose AI tools, like ChatGPT, are trained on public data. They have read every ISTQB training manual, blog posts by Michael Bolton, and even Tariq King’s training classes on AI for testing. They have also watched Angie Jones’ Java for Selenium presentations — maybe even appreciating her wit. However, the bots have also encountered anti-automation rants, beginner-level selenium tests on GitHub, test plan templates with only sample tests, and even tweets and comments from AI skeptics — which won’t age well. Most of that public testing content is general and often ‘basic’ — meaning the bots are not learning from the best of us. Most excellent test cases, code, and testers are hidden behind corporate firewalls and inaccessible to AI, which limits their ability to learn from the very best of us.
While AI can improve the average tester’s performance and speed, its learning is limited to the average of us. The impact of AI on testing will largely depend on what it learns.
Learning from the Best Testers
AI should learn from the best testers in the world. Imagine a tester specializing in address fields — an AI trained on their data, techniques, and opinionated oracles would likely outperform the average fresher. If the AI learned from an address-field tester at FedEx, it would know the longest valid address in the US, the shortest in Singapore, and the addresses most likely to crash the system — you probably don’t know this yourself.
We could also generate API tests using the Tariq bot, search box tests using the Arbon bot (I’ve worked on two search engines <blush>), generate Java for Selenium tests by Angie Jones’ bot, or even a set of thought-provoking questions from a Bolton bot.
If we could digitize the best testers and unleash them on all our apps, how much better might our software be? The odds of encountering basic bugs in apps would decrease dramatically, ultimately leading to better user experiences and a more efficient world.
Humans Still Wanted (and Needed)
Although AI technology has many advantages, there is still a role for human testers. Many developers and testers struggle to provide basic test coverage, and it's expensive, but with AI bots handling these tasks, humans can focus on the app-specific features that make their products unique. The humans may have time to learn more about the business and code or take a more extended lunch with their development team. For developers just starting a project or those who cannot afford to hire a human, having access to virtualized testers at their disposal through AI technology would be awesome.
Furthermore, if an AI version of Tariq files a bug, your project manager will likely prioritize it for immediate resolution. Additionally, what if an AI bot could suggest fixes to the code? Or explain in detail why a particular security issue is a concern and explains how it could be exploited. Or why an accessibility bug should be fixed? Imagine an AI based on Michael Larson that not only finds such issues but also writes an empathetic argument for why the bug should be fixed. With AI’s help, more bugs will likely be fixed with newfound speed and urgency.
If you are an expert in a particular testing specialization, like SAP or Salesforce applications, etc.. You could train an AI bot to test like you and share the same testing perspective with the world. You could become famous in the testing world as your AI doppelgänger tests apps worldwide just like you would.
“Bring Your AI Expert Testers to Work” Day
What if you could select which bots to use and bring them into your team? You could save the company money, improve the software, and ship faster while making yourself look “AI-clever” and indispensable. Test managers will appreciate the additional coverage when they can’t get a new position open. Individual testers will find managing virtual testers like Tariq fun and empowering. Developers can spin up a quick virtual expert test team on the fly, and Directors/Management-overhead will love the cost savings and knowing the product couldn’t have been tested better.
I’ve started this journey of digitizing testers and hope you join me — we need your expertise and opportunity to learn from testing your app! More posts to come on this topic, with more details.
If you’re interested in being part of this revolution in testing, please signup for the beta or to be digitized.
— Jason Arbon, Test Nerd