28
16

Learning and improving skills in age of AI

2d 3h ago by thelemmy.club/u/fake_tourist in fuck_ai

Hi, I want to understand how you guys learn new things in the age of AI? I am a software engineer relatively new to the industry ~5 YOE. I am trying to learn new things in my job but since we use so much AI for programming and stuff, I feel that I might not be learning as much as I would have if AI did not exist.

I cannot stop using AI since it does make few things easier, and also my company keeps watch if someone is not using enough tokens. On other hand I also dont want to fall behind on my task deadlines by doing things old school way.

Is there any way to get better at workplace in this situation? I just want to get really, really good in my domain.

On the token front. A few of my friends have protestes by tokenmaxxxing. They burnt through so many tokens in the first week that management cancelled that quick.

I'm learning the same way that's always worked for me. Identifying my biggest weakness and tackling it. For me that's the fact that I don't architect from the start. So I've started revisiting UML diagrams and various software architecture methodologies from various textbooks. I talk to other software engineers about their methods. And I try to implement a specific change (actually writing any sort of a plan now instead of writing the first function and sorting as I go)

The best thing to do is to stop using it entirely. It objectively makes you stupid, so not-using it helps keep the stupid away.

The second best would be to use it as you might use StackOverflow: ask it to help solve a toy problem that represents the crux of a larger problem you're dealing with. It will provide a toy answer, but that would represent a starting point for you solving the problem yourself. In order to actually solve your problem, you'll be forced to understand the toy solution and how its best applied to the larger problem.

But seriously, put the sparkly autocomplete down and read. There are likely ample docs to help you, as well as lots of forums with your answers. Learning how to research is part of learning too.

Well, specialist books are still around. We can always read a textbook on advanced digital signal processing. Or concepts in some programming language, written by the authors of XY themselves... Though... Practice makes perfect. It's gonna be hard to remember a book worth of information, without exercise. Or properly improve without trying to apply it and make a few mistakes yourself in the process.

thanks, that does make sense. Will try it.

cannot stop using AI since it does make few things easier,

Sounds like: I cannot ask the stranger at the gym to stop lifting the weights for me since it does make things easier

my company keeps watch if someone is not using enough tokens.

So does mine. It's called tokenmaxxing, where you just use the tokens in a bullshit way. Tune your AI config to use as many tokens as possible. Prompt it, put it in a loop, toss the answers. The only way they'll be able to tell that you're not actually using AI will be when you stop turning in AI slop. (Which they won't be because they're AI brain rotted.)

On other hand I also dont want to fall behind on my task deadlines by doing things old school way

You really won't. I mean, you may at the beginning because you don't have any skills. But, once you get skills you won't fall behind and you'll actually be stronger because you'll actually understand the code and will spend less time going back and fixing problems.

Unfortunately the only things that are guaranteed to be slop free are things published in the time before AI. Classics like SICP will always be relevant

super easy... DONT USE AI, Learn everything by hand, from books, guides, other humans.

when you can do the work without referencing a lot of outside things, then use AI as a TOOL to confirm things... not create it.

never relay on AI.. pretend it doesn't exist when working.

r relay on AI… pretend it doesn’t exist when workin

I have deadlines and token usages to meet. I just cannot stop it.

Then you're locked in by your employer.

I switched mine also in part of AI bullshit.

hey not everyone can switch like this and this particular job I really care about because I gain access and opportunity to work on things I would not at many other places. Actually all other companies in this domain are also same with the AI usage.

Just answering your question, truth hurts and isn't very convenient.

Also finding another employer that's perfect in thst regard might have other unexpected downsides.

Short, they're part of the problem in any case.

I have the same problem, and I decided to just work on a personal project in my spare time without AI.

This is my next step. I started doing project Euler for fun with a programming language I've never used before and it was fun. But now I wanted to tackle something at home that doesn't feel too much like a chore.

I would say just do some book reading and practice coding in all the time that all that AI is saving - but poeple who measure find that AI is not actually saving time at coding tasks.

Still, when I was locked into closed source bullshit frameworks by my job, I practiced with open tools and frameworks in every spare minute.

Many people have no time at all to invest in their future careers. I have found that eeking out a few minutes here and there got me ahead of the pack.

Of course, I'm obsolete now, because web frameworks were invented in 2003, and they stopped hiring programmers, entirely. Oh - I mean, I'm obsolete now because of AI vibe code tools.

The reason keeps changing, but I've been obsolete for a long time.

Edit: I look forward to being obsolete for a novel new reason in 3 to 5 years.

This is exactly what an obsolete programmer would say. Also, I had a good laugh, thanks

It's their problem in the end. Don't fight it. Learn fun analog skills in your free time.