With generative AI in application testing, I see a repeated unhelpful agent message: insert specifications to generate test cases. It's now more important than ever to get this: test cases are not testing. Testing is the process of empirical research for information we might be missing. Test cases can be an input (as the traditional framing of it goes) or they can be an output (as the exploratory framing of it goes). I recommend they are code - and I demo this idea primarily in Python code.
In this session, we take a single page application and ask: how would you test this? With notes, programmatic tests in python, a listing of now known bugs, we will see how testing actually works, and why in age of AI the 'human in the loop' or 'alignment engineer' would do well with solid testing skills. And it does not hurt that those very same solid testing skills are valued in developers in the current market.
maaretp
sugarenia
chriscoyier
csswizardry
cassininazir
denniskardys
lara
eileencodes
deanohume
pauljervisheath
mojombo
morganepeng
