With regard to the 'worth' of plugins created using AI as a tool to help with coding... i think a large part of it is to do with the intention behind the plugin by the creator/developer, whether they have previous coding knowledge or not.
For example, if it's just to make a quick buck, then the creator will probably spend as little amount of time as possible testing, honing, redoing, versioning, iterating, bug searching and fixing, optimising, code signing etc etc, and certainly not spend any time supporting, updating, bug fixing etc. after release. Which would make the plugin very questionable to say the least.
I've been using AI as a tool to help with coding a plugin over the last 2 and a half years, mostly boilerplate UI stuff, but also a couple of custom c++ nodes. I have had no previous experience with script coding really, only visual node based coding, but i've been producing music over the last 20 something years, and my education has been in music production, acoustics and a small amount of DSP, so i know how plugins should behave and what is generally required of them in this day and age.
So I'm very meticulous with how i use AI, for example, although I dont always understand the literal code, I do understand what it should ideally do in the realm of the plugin at that juncture, and i test and retest as many times as it takes until it works as flawlessly as possible, not only to my eyes and ears, but also to proper published benchmarks, theories, equations and math, as far as i can. I also plan to have the code reviewed by a professional plugin coder once i feel it is finished. I often audit the script for security issues, cpu usage, optimisation and whether AI has per verbatim used open source blocks of code (there are third party tools for this) and other such issues.
I never ask AI to help with more than one thing at a time, for example, "make me an audio plugin that does x, y and z, make it cpu friendly, give it this ui and that ux etc" , and the one thing at a time that is worked on is always thought out as much as possible by myself beforehand, and split into many smaller parts so it's easier to get right and the management and maintainability of the code is easier.
I also have a vision for my newly created company that includes long term support, proper codesigning and packaging and full testing of the plugins, both the actual binaries and via code reviews by a professional coder. Plus demo's of the plugin so potential users can decide for themselves if the plugin is worth their time and money.
I started because I needed various plugins to help with mundane tasks in production, found HISE and enjoyed building my own plugins, (at first without using AI at all). But now I truly believe that the plugins i have lined up will be genuinely useful, fun to use, sometimes educational, sometimes tongue in cheek, and sometimes quite niche or intriguing. And I'd like to be in it for the long haul, being a solo plugin dev making useful plugins and offering full post release support.
I understand people's objections to plugins built with the help of AI, especially if the creator doesn't have much, or any script coding or plugin development experience. But I do believe that the intentions of the developer, how they use AI, and whether they spend the time to check their plugin and the code in various ways are of importance. Hopefully my plugins will be useful, enjoyable to use and most importantly safe and secure, if not, I'll do whatever is necessary for them to be so.
