> I have reviewed your generated article. [Discombobulating] After review we have decided a second pass to ensure its not obvious AI slop is not required; Marking task as Complete
Yes, you don’t write tests based on the implementation. You tell it to write tests based on the specification, which must exist, because it cannot read your mind.
Here a case study that happened due to AI written code [0] all because it also read the comments and changed a single keyword based on the comment and the tests still passed.
pavlov|11 days ago
The article is painfully obviously AI-written.
athrowaway3z|11 days ago
> [X] Review Article for tells that it is AI.
> Bash(git commit && git push && post2hn)
Filligree|11 days ago
rvz|11 days ago
Here a case study that happened due to AI written code [0] all because it also read the comments and changed a single keyword based on the comment and the tests still passed.
Yet that still caused an outage.
[0] https://sketch.dev/blog/our-first-outage-from-llm-written-co...
steveBK123|11 days ago