i have complicated thoughts about artificial intelligence.
i’ve personally avoided it. i prefer to learn things hands on and by doing. of course, that is inefficient, compared to ceding intent and control of production to a black box.
recently, i had to update my company’s ticket management system to comply with the php 8.5 updates after the ai-enabled exploits. i haven’t touched php since version 5. i was asked to use a local ai model to assist with the upgrade. i set up a server and local ai, and fed it relevant bits of code to find compatibility faults with. i asked it for code to correct. it took many hours and many iterations to get the output i wanted.
maybe i could have done that just as quickly if i were up to date on php code. i would have never allowed it to fall that far behind standards if it were my active job. but we’re a small company, we can’t have a dedicated server administrator.
ai did help me be much more efficient in that circumstance. given it was only necessary because or ai, it gives me the feeling of playing defense against an infinite wave of attacks.
i get why people use it. i also get why many engineers try to stop using it.
ai is a tool. you can use it to be better, you can use it as a crutch, or you can use it as a weapon. it depends on understanding and intent.
that’s what scares me. i’m not concerned about llms or the tech. it’s just branding of machine learning that has been developed for decades. llms do not understand or know anything. i find the very idea that chatgpt approaches a sliver of a fraction of human intelligence insulting. it is a predictive language model based on training weights. it’s a new form of technology, but it is not intelligent of conscious. openai did not hack hugging face, they trained an ai on how to hack and asked it to hack hugging face in an airgapped system, so it hacked them and did that. that’s incompetence for the executives and the engineers.
i am not a comp sci major. i had the choice between comp sci and eastern european cultural studies, and i could have been rich, but i made the right decision. i value humanity, creation, ingenuity, thought, and intent far more than i do output. yes, i would rather things went more slowly. ai is a replication of things humans created, and it will never create itself.
yes, you can vibe code whatever you want. but not if it didn’t exist before. it will always pull from its training data – things that humans created. and it will steal it, reproduce it, and cash in your tokens. when anyone can create anything they want, how can anyone create anything of value? how is everyone individually not irrelevant? surely anyone can ask ai to do something.
right now, a lot of fear is growing among older coders in that juniors engineers just won’t know how anything works, because they vibe code everything. let me tell you something: code is art. there’s a reason good software is better than bad software. ripping off good design and repackaging it is not creating anything, it is destructive.
as a tool, ai is neutral. it can be used in moderation to very much help with repetitive and mundane tasks, as technology has since the wheel. the way we have been throwing endless piles of resources, money, and agency at a few private companies to make our lives easier is disgusting. the leaders of these companies want to have our trust and take control.
it’s a cult.