Somewhere out there there is someone better than you at what you do. There’s a company with more resources to do what you do. There are both that can do it quicker, more efficiently, and for cheaper. I say this not to make you feel worse, but hopefully better as all of this was true way before AI and its never stopped anyone before. Think of AI as a tool to extend yourself, not a replacement.

In the end you will gain new knowledge from doing the project.

And you can try to use AI to speed up the project and see if that works for you, and gain a better insight on what are the current AI systems shortcomings are.

Why should i bother doing something “like this” if someone else has already done it? Because you will gain the experience to go to the next level.

Its never a waste of time but an improvement of yourself.

I guess it depends on your motivation for said project.

Do you enjoy the process of working on the project? Do you use it to learn new things and improve the skills required for it? Or all you care about is the end project?

If it’s the last one I can see why you could get discouraged but if it’s the other two then not really.

If I work on something specific I do it because I want to create this thing using my skills, experience and ideas. I also try to do as much of it as I can on my own, without using other’s people assets etc. For me, the work I put in is an important of the process and each step towards completion makes me a little proud and happy that I can create something.

At the same time I’m a weirdo who tends to use less efficient methods if they aren’t as fun to use when working on personal projects. I don’t really care about using AI in the first place so that might skew my view a bit.

There’s also the fact that AI isn’t omnipotent. It makes mistakes just like us and I’d rather fix my own mistakes since this way I know what I did, how I did it and where to look when things go wrong.

Sorry for the word salad. Your post gave me a reason to spew some unfiltered thoughts about an issue I never really thought about.

Everyone’s comments here are right but I do feel you OP, I’ve been wanting to learn Godot and react recently but seeing people just pull code straight from ChatGPT and it’s way better then I can code right now, does feel a bit like “should I really try this path”.

Tbf I do think you can work hand in hand with ChatGPT where it just makes you faster at development, but I still do feel a little sting. And with so much stuff I want to do, I’ve re-prioritised other things instead.

Quite the opposite, actually. Because of AI, I got back into web development for the first time in 20 years. I’d never be able to figure out the new stuff like HTTPS and DDoS protection on my own if it weren’t for ChatGPT’s help. It got me up to speed on HTML5, and helped me remember how PHP and Javascript work, all within a couple of days. Would have taken months otherwise.

You should be using AI as a tool to expand your creativity, not feel hindered by it.

What kind of project were you thinking about?

I only see AI do narrow tasks in the near future. Draw a picture, write a story with very forseeable plot twists, draft computer code… Certainly things that are very useful, especially if AI is used to assist a human. But I don’t see it cough up something really substantial and of value (on its own). Especially the complexity in a creative project. It can draw a nice astronaut on a horse. But not come up with a clever new approach to a subject, know the history and transfer something meaningful into the art piece. Same with the complexity and different roles in computer game development. I think thats too difficult for quite some time to come. But it’ll make you faster at generating assets and write all the boilerplate code.

But if you just do simple things or sell drawings of people in anime style, you probably don’t need to bother any more. Same with a few jobs.

Not even.

I’m shit on ansible because it’s really cheesy and weak. But that’s how the bills get paid some days.

Now I put the task into chatgpt and even 3.5 will give me a 70% solution I can fuck with and send back “hey will this work” and let it correct my indentation - fuck that - so it’ll probably run.

That’s like 90% of the work there, and it’s only an hour in. I’m not chuffed at ‘cheating’, as I know I could have gotten there without the paper to cheat off, but I would have wasted far too long. And Ansible will die either through its own brontosaurus layout or IBM will snshittify it to the point people need to churn away anyway. Then maybe it’ll be to something that brings me joy and I’ll want to do more and not less.

I quit my job to start the year and I’m currently doing a sabbatical year. I’m apathetic about the idea of eventually honing in on a specialty to learn when I re-enter the workforce because I’m unsure how sustainable the skills I learn will be in demand for.

The only thing I can think of is expanding my base level understanding of LLMs. My bet is that they will become the foundation with which future projects are launched in the same way that elementary school is the foundation for basic reading/writing/comprehension skills.

Read some existentialism, no joke. I don’t agree 100% but I read a bunch of Beuvoir over the weekend and one thing I did like was it made me internalize the idea that coming up with a project I care about and achieving it is worthwhile in and of itself regardless of if it “could” be done by someone/something else.

Think about it this way, there are mathematicians from 500 years ago who did a lot of stuff by hand for hours that I could work out with a calculator in seconds today. But does that mean all their work was worthless? If I create a fairly shitty drawing, but I’m proud of my having created it, am I wrong to be proud simply because my friend who is a great artist could make a better one in half the time?

It’s not just about the journey, but it’s not just about the destination either–its about the journey to the destination, and placing value only in one of those things will cause you to be at a loss for the rest of your life.

Definitely not. The more I use LLMs, the more I realize how error prone and unreliable they are. They’re good for generate flavor text, I would never rely on them for anything critical. Context size and nothing verifying correctness of the output are two big ones where they’ll never replace humans. They’re a useful tool to augment productivity, they’re no where near replacing humans yet.