I find this idea that humans all of a sudden write beautiful code very funny. Most code produced by hand is dirty and filled with ugly hacks. The argument might work against AI art but falls flat for programming.
If I saw that badge on someones github I would think it had something to do with lorem ipsum text generation, rather than anything to do with AI.
Literally all my code has been ”ghostwritten” for the past 18 months. Does not sound like something enterprise customers would like to hear and try to understand what it means.
I think we really need hard liability for software engineering.
You may have noticed the absolutely vast array of AI development tools and assistants and IDEs and integrations - this is a reasonable indicator that developers are actually doing AI/LLM development.
To have any chance of adoption you have to be at least a little strategic. You may think AI is pure evil but you have to make some concessions to AI users to incentivise participation. Try making it sound neutral through out the spectrum, use neutral colour scheme. Yes, you’re not telegraphing your position on AI so obviously any more but you might get some useful information out of others.
(2) The code I write with AI doesn’t fit on the scale.
To me, the metaphor doesn't really work, especially at level 5. Lorum Ipsum is literal placeholder text which is basically the same everywhere it's used; I don't see what that has to do with vibe code. (Also the verse/prose thing seems pretty wanky to me, but I admit that's just a matter of taste.)
"AIX®" is also a registered trademark of International Business Machines Corporation (IBM) and used for the AIX® operating system that is still in use today.
I would be careful to use that name.
Red should meant manual human review without automated tools nor AI.
Green for proper AI review and tests verifying the expected input/outputs.
This is perhaps overly generous to pure-human authorship. These days, when I write code I like to think I know what it does. I still wouldn't call most of it "crafted like poetry". When I was just learning though, I wrote plenty of code 100% without AI (in fairness, it didn't exist) that I had little understanding of, and it was only "deliberate" in that I deliberately cajoled it into passing the tests.
Or put differently: don't conflate human authorship with quality; people can write garbage without needing AI help.
If you're thinking "crafted like poetry" implies any kind of existential "quality", I'd like to introduce you to William McGonagall[0] and you will swiftly and powerfully be disabused of any "poetry -> quality" confusions.
The key insight was to not just handwave or guess at how much is automated, but make evaluation and review part of the continuous development loop. I first implemented in https://github.com/Entrpi/autoresearch-everywhere where I used it to deliberately automate more, in the spirit of Karpathy's upstream (and to very good effect. I have some of the best autoresearch results anywhere, and the platform is far more robust than it started).
