RE: https://mastodon.social/@tdpauw/116392584939788699
Thierry pulls no punches, as usual.
| I write | https://thinkinglabs.io/ |
| I code | https://github.com/tdpauw |
| pronouns | they/them |
RE: https://mastodon.social/@tdpauw/116392584939788699
Thierry pulls no punches, as usual.
Published: Don’t Let AI Invert The Testing Pyramid
Lately, I encountered a specific article titled "Quality Engineering with AI". I have seen it shared with a dangerous level of enthusiasm within organisations.
To the untrained eye, the article appears to validate a so-called "modern" testing strategy; to anyone who understands software economics and the mechanics of software delivery, it is plain disconcerting.
https://thinkinglabs.io/articles/2026/04/12/dont-let-ai-invert-the-testing-pyramid.html
Published: Don’t Let AI Invert The Testing Pyramid
Lately, I encountered a specific article titled "Quality Engineering with AI". I have seen it shared with a dangerous level of enthusiasm within organisations.
To the untrained eye, the article appears to validate a so-called "modern" testing strategy; to anyone who understands software economics and the mechanics of software delivery, it is plain disconcerting.
https://thinkinglabs.io/articles/2026/04/12/dont-let-ai-invert-the-testing-pyramid.html
I got a dm from someone who joined one of my open enrollment workshops in DC in the 00’s and then brought me in to work with his teams as he worked on different things. Anyhoo, he saw my 2008 Conway’s Law blog post quoted in a meeting the other day, and reached out to tell me :)
Thank you to @tdpauw and @matthewskelton and others for bringing my um “colorful” paraphrases and implications of Conway’s Law to more folk <smiles>
There is an art (and a science) to numerical precision that seems lost in software, writing and conversation. The trick to appropriate precision is understanding accuracy. This all falls under the banner of numeracy.
For example, I just received a confirmation of a cinema booking that gave the time of the film in HH:MM:SS format. The site lists programme times in HH:MM. They normally start trailers within a few minutes of the advertised time. To list seconds is an innumerate and false promise.
Which references another interesting article on the topic from Adam Tornhill
Skills rot at machine speed: AI is changing how developers learn and think
In reply, someone shared this interesting article:
Cog Debt Article: From Tech Debt to Cognitive and Intent Debt
https://arxiv.org/pdf/2603.22106
Comprehension Debt in the context of AI
notes from Denali Lumma via the DORA community
That was a bit of my worry, that people no longer make the effort to understand and internalise. There have been more posts about that: there is an upcoming reasoning deficit.
We have seen this with Stackoverflow to some extent (nothing bad about SO, great tool, it has been very helpful). Now it risks being exacerbated.
https://gist.github.com/rossdm/080959eb58a5f72bcc297925b0a5be36
An article was shared on one of my clients’ Slack about AI and testing: how about approaching testing when AI is changing the code all the time, kind of.
The people are enthusiastic about the article coz it kind of confirms their not-so-performant testing strategy.
I violently disagree with the article because it confirms bad habits.
So, want to write an article about it, but referencing the article feels like shaming the author. How to avoid that? How should I move forward?