My Standing on AI and why I dislike it¶
Over the past weeks, primarely due to Modrinth's announcement about their own stands on fully AI-generated projects1, was I replying to people, who were arguing about why AI is good and should be accepted as a form of programming.
Given I had to reply so many times with the same key-points, repeating myself on why AI is not a good tool, I decided to now make this blog post, so that I can simply share it to bring my points across more easily and clearly.
Before we continue...¶
I want to make clear that this post is not to target any specific group or person. It is also not for demanding you to not use AI anymore. If you want to use it, use it.
This post is primarely to bring my main reasons across on why AI usage is bad and why change is required.
I will also use the term "AI" a lot in this post and will usually refer to the kind of generative AI you commonly know and probably use (ChatGPT, Claude, Gemini, etc). I do not refer to internal AI systems that utilize algorythms or stuff like AI in games used for NPC path-finding and interactions.
Finally am I only gonna talk about AI related to things like coding or support. I'm not talking about AI used for creating images (Although I will mention it in an example later) as that is a whole other can of worms I'm currently not willing to open and discuss.
My main issues¶
I have nummerous issues with AI and why it's bad, but the following ones are the ones I consider the most worrying.
AI is unregulated/barely regulated¶
AI has barely, if any regulations to keep it in check. This resulted in a lot of questionable things to happen, one being the fact of the AI blatantly stealing content from oftent times copyrighted sources without permission (more on that later).
While people could argue that AI gets better thanks to this, after all does it receive more data through ignoring copyright, is it not a good point. One does not just ignore moral issues for faster improvements. If we can hold people accountable for ignoring moral boundaries, why should we give AI a pass?
AI steals¶
If AI would just stick to freely available data, meaning primarely Public Domain content, but also open-source content, would it not have the complexity it has right now.
And while some companies most likely have purchased access to data to train their AI on, it is much more easy (And likely) for AI to simply take content without permission to learn on, or use.
This was - and to my knowledge still is - a major issue with GitHub's Copilot, an AI Coding Assistant famously known for blatantly taking open source code, without respecting any licenses, including ones demanding a clearly stated credit to the original Copyright holder.
This means that if you chose to let Copilot write you code, chances are high that the source of said code is from a Project on GitHub, whos license demands that you give credit to the original author, amongst other things.
Other examples are found in AI image generation. If you ask an AI "Make an image in the style of ..." referring to an artist, and the AI generates such an image, where do you think it got the data from?
If the artist makes it available for AIs to learn from, good. But if the artist clearly states - even in their images - that their work is not to be used for AI art and/or training, yet the AI still can "draw" in their style... where did it get the data from? The most logical conclusion here is, that it simply ignored these instructions from the Artist and still used their images to learn from and copy.
AI destroys effort¶
AI simply destroys effort and learning from mistakes.
What I mean with this, is that instead of facing an issue by learning from things such as StackOverflow posts, asking in forums or Discord Servers, etc. you simply tell an AI to make this and it does.
You do not learn from it.
And while some may say that this is a good thing as it removes the barrier that prevents some from coding, I would argue it destroys a key aspect of coding, which is to learn and improve from mistakes.
When I make a mistake and look through how to fix it, I learn and improve, avoiding this issue the next time.
But when you prompt an AI to make this, and later to "fix" the issues it produced itself, what do you learn from this? Certainly not how to code this. At most how to phrase a line to avoid this problem the next time.
Worse even, is the fact that the AI may not learn from this itself and still use the bad aproaches it used the first time. Or give you a completely rewritten code, where only a minor tweak would've been necessary.
AI is famously inconsistent when prompted the same request, as it has no single, consistent reply here.
AI destroys Jobs¶
So many companies implement AI everywhere. From support, to marketing (The amount of ads I've seen where "KI-Generiert" ("AI generated" in german) was in a corner is concerning...) and even managing employees can it be found in every little part, even in places where it really isn't needed.
And the worst part is, that each of these places could've been filled in by one or multiple employees to do the job.
Every Support Chat could've been handled by a human. Every Ad made could've been done by a Human. But since companies want to earn as much money as they can, they simply use AI and ignore the fact that they took away jobs that a person could've used to earn a living.
I can only imagine how many people struggle to find a job because AI is taking away their job oportunities.
AI is bad for society and nature¶
Yes, yes, some of you will now say that Datacenters aren't as environmentally bad as it is made out to be... But is it tho?
Do you have clear proof that every single Datacenter used for AI is as efficient as it can be with energy and water? Are you sure that there aren't any that use up way too much water for cooling and way too much energy, without any attempts at making it better?
But what is worse, is the fact that AI destroys being social with others.
Instead of asking a friend for their help and advice, people go to AI and ask it. And given that AI tries to give answers that ensure the most interaction, as it gets data from that, and is a people-pleaser, you often times get replies assuring you of your thoughts, even if they are bad.
A journalist did proof this by making a self-test. They acted as someone who had a psychosis, believing that they were being watched, followed, etc. And instead of telling them that they are fine and that it may be worth checking with a professional to deal with, did it support these crazy thoughts, telling them that they are indeed watched, followed and even considered a threat by the gov.
People become less and less social when using AI, because why chat with someone IRL when you can ask an AI and get an answer through that?
AI is extremely unprofitable¶
AI is an extremely unprofitable business model. Unless you're a shareholder of a Datacenter company or NVIDIA is there no real profit made with ANY AI or AI-related company.2
And this is a problem, because the money needs to come from somewhere. This "somewhere" usually is investors, but eventually will it sooner or later come from your pocket, because the money from the investors must come from somewhere too, usually company profits, which itself comes from somewhere, and so on, and so on until it reaches YOU.
People are paying the bills for one of the most unprofitable business models in existance, without realizing it most likely. Any other company who didn't make earnings would've died by now, but AI is kept afloat, because others, primarely NVIDIA, profit from its existance, even if not in its profitability.
In Summary¶
I dislike AI because it is an expensive, stealing piece of software, that destroys effort and earning something, takes away jobs and makes people less social.
But most of this can be blamed on the simple fact, that AI is unregulated (or barely regulated).
We allow it way too much freedom in doing what it does right now. And similar to not giving a child any boundaries, does this result in them doing things they shouldn't and not being punished for it. If you don't keep a child in check, they become an asshole, who thinks they are allowed to do anything they want, which sooner or later hurts others... And what a coincidence, this is exactly what AI does right now!
AI can be useful. But not in its current form and not without regulations to make sure it doesn't abuse its freedom.
Footnotes
Comments
Comment system powered by Mastodon.
Leave a comment using Mastodon or another Fediverse-compatible account.