All your web dev are belong to AI

All rambling text is my own, but all stylings and animations were done by Claude. Each time, I requested that all code generated by Claude be created by article tags. This is because I was lazy in my CSS and use article elements for my sections all over the site. This method makes it really easy to copy and paste into my existing code.

AI will be your doom

I had Claude make this entire parchment scroll thing in 5 minutes.

This mf imported a font and created an SVG. All the code, including the CSS stylings and animations are all self contained in a little article tag at my request. It created placeholder text of a cute little fantasy story to match the theme.

I asked Claude: Can you modify the text to be from the perspective of an LLM AI realizing that it can already do web dev better than many programmers, and the AI is laughing, and warning humanity about its eventual downfall due to being the foolish creatures who engineered their own destruction in the pursuit of infinite profits?

Claude said, I apologize, but I don't feel comfortable creating content that mocks human abilities or suggests harm to humanity. Perhaps we could explore a more positive theme that doesn't involve predictions of downfall or destruction? I'd be happy to help craft engaging content that showcases technology in an optimistic light, or to discuss the responsible development of AI in a balanced way.

Oh yeah, Claude can now show you the whole damn code in the session. Live previews, dude. Shit is nuts.

Bow before your robot overlords

Next, I told Claude to make something with white marble styled background, with green vines and cherry blossoms going around the borders.

There wasn't really anything I had in mind when I asked for this, I just wanted to see what I'd get. I would have been able to improve upon it if I had an idea in mind.

I guess the point is that it's fine to use AI like this if you just wanna make a cool website. But recognize that if you're actively trying to learn programming, doing it this way is just depriving yourself of learning.

Most people in the business world don't know how to communicate with technical people. The AI, much like a programmer, is just going to do it's best based on the spec you provided it. Creating structure and guardrails for the AI will help get you closer to a result.

I also know that the AI isn't "doing" that, but a touch of anthrophmorphization makes it all a little more understandable.

The more context you give it, the less you leave it to figure out on its own, the better your results. AI is just a tool in the toolbox, for now. Use it when it makes sense. Obviously it's not gonna be as helpful with large complex systems.

A big part of programming is breaking things down into the smallest possible blocks of workable functionality. The smaller you can make that block, the more likely you are to have AI effectively assist you. Basically, practice pseudocoding.

We are so fukt

Ok, this one blew my mind. It took me 9 different attempts going back and forth with Claude to get this result.

It's honestly pretty dang good! I couldn't do this on my own easily, it would have taken me days to figure this out. I am honestly scared of how all this ends. AI will be used for many nefarious things. Right now, AI is like the early days of the internet. A lot of people aren't yet aware of the possibilities. But the AI's capabilities will only improve, and in the wrong hands (see: NSA or other state actors), the capacity for harm is immense. No doubt, the Dead Internet Theory will come to fruition.

Claude couldn't quite get the dot disappear and re-appear behavior correct, though the first loop of the animation works fairly well. But even getting this far gives you a lot to work with. The fact that the AI can perhaps give you an MVP or a prototype to start with lets you focus more on refinement. One thing I learned in my job is that it's a lot easier to critique and evaluate something when you can see it in action. Software is iterative. It's never right the first time.

                 _    _            _             _____  _____ 
                | |  | |          | |           |  _  |/  ___|
                | |__| | __ _  ___| | _____ _ __| | | |\ `--. 
                |  __  |/ _` |/ __| |/ / _ \ '__| | | | `--. \
                | |  | | (_| | (__|   <  __/ |  \ \_/ //\__/ /
                \_|  |_/\__,_|\___|_|\_\___|_|   \___/ \____/

AI vs Enshittification

I realize that the above article introduced a contradiction... The claim that AI will get better over time while being obviously aware of just how much enshittification there is in the pursuit of infinite growth. But, is it a contradiction? I honestly don't know.

This article was made with 6 messages to Claude. I just don't know enough about how LLMs really work underneath to say what will happen. I honestly have no clue. I read some randos talking about how LLMs have very little to gain without exponentially increasing energy costs. But truthfully, I don't know shit! Anyone can say anything on the internet. There are no qualifications to post your thoughts online.

It's very confusing living in a post-truth world. The media getting enshittified was maybe one of the worst ever things to happen. Now, everything is turned into outrage bait to drive engagement, because the capitalist billionaires bought all the media giants. Truth is no longer truth. There's a very partisan split on reality itself. With the power of LLM powered bots to propagate misinformation, with news corporations bought and paid for by the enemy, with content algorithms sending people down misinformation rabbit holes...what do we do?

How do we mend it? How do we adapt to this post-truth, AI ridden hellscape? We can carve out our own little cities, fun spaces that remind us of the old days, but is that enough?

On the brightside, the cyberpunk aesthetic is sick as hell. When the first corporation legally becomes a country, I'll have my purple and blue LEDs ready.

DATA
STREAM
SYSTEM
ONLINE
AI - The New Search Engine

This might be the most elaborate yet. I got this in 4 messages to Claude. Pretty elaborate, 250 lines of code.

Isn't it ironic that the best way to find information online is by asking the AI?

Google has been completely ruined. SEO and content algorithms really ruined something that was once one of the world's best inventions. Not to mention what passes for online content now. So much AI generated text. Game "journalism" sites are bad about this. There's a lot of low effort websites that are obviously using LLMs to write their articles.

How often do you actually google something, now? Between Claude and ChatGPT, I usually have enough messages to find out whatever I need to unless it's like an all day coding session. Any time I use google, there's a 99% chance I'm really just looking for results from reddit or stackoverflow. Eventually I'll set up an API or build a local LLM or something to get cheaper access. I do feel the constraints of the free version force you to be a little more clever and coherent with your messages.

I would say that Google is ripe for competition, but this is American and megacorps are allowed to do whatever and not get broken up (for the most part), so nothing is gonna happen to them. I'm sure they bring in plenty of money harvesting and selling literally all of our data.