I’ve wanted to compile my thoughts surrounding AI and its implications for a while. I’ve worded and reworded my ideas many times, but I think this is the best place to document it all.
I want to preface this by recognizing that AI is revolutionary. I know that. I see it shaping the future of technology in ways we could have only dreamed of a few decades ago. I myself now make use of it all the time, whether to quickly debug or sketch out snippets of code, or to help answer questions when a Google search fails to satisfy. It even has some really great uses in the medical field when it comes to mass data analysis. Models are being trained to detect early signs of cancer that no human would be capable of detecting due to the sheer number of factors involved. AI is clearly capable of being used for great good and improving our lives. However, regardless of all the positives it may bring, I think there are some dangerous side effects that are evident in the kinds of things we are seeing online and, regrettably, in every aspect of life.
With every technology that makes our lives easier, unfortunately, many see it as just an easy way out. A way to cut out the need to invest more time and effort, or even to entrust tasks to other, much more qualified humans. “Enshittification,” as it has been called, is a very real phenomenon and far more than just a meme. AI has empowered those who see the world as nothing more than a means to an end (that end being profit, of course) with a superweapon.
I’m no expert in machine learning or how this technology works, but I have a basic understanding as an IT nerd and plenty of experience prompting it. I can tell you this: AI is useless without a human giving it thought. To put this in simpler words, AI cannot think for you. It does not replace that task, and if you let it, you will fail at whatever you are trying to achieve. People will always be necessary to steer things in the right direction, to actually have ideas. Think about AI art. Why does it look so uninteresting and bland? Because there wasn’t an idea behind it. It’s effectively the result of averaging a bunch of data that the AI reasoned was relevant based on some text input. There were no choices made; nothing was put where it was for any reason. This works for generic things, sure, but for what we would consider art as humans, it literally fails at the very definition.
Programming is, believe it or not, exactly the same. Can it make code? Yes, absolutely. Can it have ideas and reason why a specific way of doing things may be better than another? Absolutely not. Writing the code is half the battle. The other half is thinking about how you are going to structure what you are coding and that, my friend, only you can do.
AI cannot invent information it does not have. It will try, and the result will be specifically designed to trick you into thinking it’s correct, but much like you do not have clairvoyance, neither does the AI. By this principle, think of the following scenario:
You are in charge of programming something that requires very specific knowledge of a proprietary framework that any AI would simply be unable to work with, since it lacks context. Even if it had context, the framework would be so vast and complex that it would almost certainly fail to generate a satisfactory solution. So, you, the human, work your ass off writing a system with the framework by investigating and reading many lines of cryptic code. By the end, you have a working backend that is able to successfully call the framework and get the values you need. Now, someone comes along with an AI and asks it to build an entire frontend on top of your backend, which it would have never been able to do on its own. Of course, because it now has neatly set-up code that handles all the nasty stuff, it has no issue at all. Then this app is presented to the world. Do you think people realize the core of this app was the backend, which literally enabled the AI to do anything at all? Of course not. People see funny buttons and graphs and think, “Wow, how cool.”
My point here is that AI can’t solve serious problems on its own. But it can solve trivial tasks around those problems. So it’s extremely easy to effectively profit off someone else’s work by simply feeding it to an AI that will use it to put a nice coat of paint over it. I think this is, firstly, extremely unethical and, secondly, the complete antithesis of the philosophy of open source. That being of honesty and altruism.
If you want to use AI to trivialize building your frontend, go ahead. It’s fast and often gives decent first results (but not great; I’ll get to that). But remember that if you wrap AI code over someone else’s hard work you are actively overshadowing that developer.
Speaking of results, that’s another downside of AI I want to touch upon. AI gives nice foundations, sure. But that’s it. In my experience, there is no such thing as “one-shotting”, as I’ve heard it called. There is caring about your craft, and there is conforming to pure mediocrity. AI doesn’t know how you want to scale your project. It doesn’t know what the best user experience for your project is. It cannot have those eureka moments that make programming so beautiful and fun. It cannot go, “Oh! I had an idea! What if I rewrite X and Y such that they become more modular and allow for blah blah…” You get the idea. I see time and time again people simply conforming to what AI gives them, as though suddenly the concept of quality, or even the will to put in effort, has evaporated from their brains. If, when you are programming, whatever it is, your constant thought isn’t “How can I make this better?”, what are you doing? Of course, people have been making mediocre solutions for decades before AI was a thing, but my point is that AI makes it far too easy to be lazy. Before it, even if you were trying to do the least amount of work possible, the nature of programming would still force you to think about how to design things to make your life easier. Do you want instant gratification, or do you want a damn good project? Personally, I choose the latter.
So, it is this that concerns me the most:
Society is slowly leaning toward pure mediocrity, and AI is accelerating this process exponentially.
Look no further than AI art and video in the public sphere. Is what it generates better than what a human artist or video editor can make? In the vast majority of cases, no. By a landslide. But companies don’t care. They don’t care because it was never about quality; it is about money. AI is cheaper. It cuts the human, whom you have to pay so they can eat and pay rent, out of the equation. I can understand using AI for generating assets if you are a single individual who has no resources and needs it for something like a logo or a personal business or project, but the megacorps that are using it extensively now have billions in their pockets to hire real, talented people to make quality work.
So then that begs the question: why do we let them? Because surely, if the lower quality had an effect on profits due to people disliking it, I assure you companies would quit AI in the blink of an eye. Unfortunately, people eat it up. The average person really doesn’t care. I find that very sad.
“So Noel, essentially you are saying that anyone who uses AI is basically cyber-Satan?”
No. As stated earlier, I use it myself. The website you see right now was made much faster than I could have on my own thanks to editing and fixing existing CSS with it.
I’m saying that people with the power to put food on people’s plates are demonstrably using this technology to cheap out and save a few more bucks in their immense wallets.
And I’m saying that people, in general, who give up the concept of doing their best in favour of conformism are fundamentally misusing AI.
AI is a multiplier on top of your skills and effort. If your effort is 0… well, what’s anything multiplied by 0?
I’ll leave you with that to think about.