AI made me stupid in coding.
Posted by r2cyp@reddit | learnprogramming | View on Reddit | 103 comments
Two years ago I had an internship where I had to create a plugin for an existing WordPress website using PHP. I was the only programmer on the team. My supervisor only knew about WordPress styling and the others were working in a completely different sector. I had applied too late for internships and didn’t want to delay my studies, so this was my only option.
The supervisor told me to build a custom plugin for the checkout page and I was completely lost. I knew PHP but had no knowledge of the WordPress framework. I tried reading the documentation but it was hard to understand and other sources were often outdated. The only real resource I had was a small YouTube tutorial playlist with fewer than a thousand views per video. That became my lifeline. I followed along, learned the concepts, and eventually managed to complete the task. That experience helped me understand the WordPress core and I finally started to make sense of the official documentation. In the end I built a plugin for both the admin side and the user side of the website all by myself. My skills in programming tripled in size, but of course I gained no experience in testing, reviewing and stuff. When I checked recently I saw that my old supervisor is still using the plugin today.
Now I’m studying a higher level degree in the same field. It’s something like a master, though not exactly the same in my country. The big change is that I discovered AI. Whenever I get stuck I use it, but over time I have become too dependent on it. My skills became worse than ever. I still pass my exams, where AI is not allowed, but I can feel my knowledge fading. It feels like I have lost years of experience and become a beginner again.
There is a guy in my class who never uses AI and I am jealous. Around 90% of the students here rely on AI for assignments, and many fail the exams for this reason, which also feels like a sad reality, yet that guy still scores the highest.
AI can be good sometimes, but it's a virus on you. If you use it too much, you can't stop. I wish I had never discovered AI, that would be a time when I could at least show my skills and knowledge, but today I feel like a dumb ass who is no different from those who use AI in my class and suck at coding without it.
Long story, but it happened to me sadly. I decided to build some projects without AI and it’s been doing good. It’s like a memory refreshment. I plan to build a simple PHP framework soon, as my final internship is coming up to graduate fully. Don't rely on AI too much guys. The love of programming is building yourself. That's also why I chose this path.
frustratedsignup@reddit
AI can be helpful, but I use in rather sparingly. For instance, I was writing a python script to check the amount of free space on a Windows server system, but I didn't want to use some third party library/module for the project. Keeping it reduced the standard library would avoid a lot of potential complexity and make the solution portable to other servers.. AI solved this problem in a rather nice way. It basically used the standard library to call the Win32/64 API functions needed for the solution. That isn't something I would do on my own, but it was nice to see that it was possible to utilize my former knowledge of those API's in a completely different language.
I also did some work with curses and Tkinter recently and, for those projects, I avoided AI completely because I wanted to get to know how those features worked on my own. I wasn't under a rush to get it done, so I took my time to learn what I needed to know.
Sgg__@reddit
I use AI a lot for work and my productivity skyrockets. Now I have more free time that I can use to practice. Think that way 🤷♂️
zilyck@reddit
I think a big part of programming is gaining information to solve problems. I never studied programming but I needed to do things which required code, so I taught myself. This was shortly before those LLM AIs became popular. I did learn a lot through googling problems I ran into, reading documentations, looking at github for similar projects, etc. Now it's all in one place but worse, I'm not sure if I would have the same success today.
Funnily enough I think coders not using AI to learn will have some advantage rn, while those who can only work with AI might just be replaced by it completely.
AffectionateCode5339@reddit
am doing it without ai, completely understand u,
CircuitryWizard@reddit
But at the same time, you can learn the wrong thing while surfing the Internet and just get tired of information overload. And unlike people, AI does not learn (in the usual sense) and is essentially an eternal junior (the models that exist now) and is limited by a context window, the increase of which squarely increases the costs of working and training the model. Well, and do not forget that AI can and will often hallucinate and therefore a controller is needed who will check the output code.
We can also say that people who know how to use a card index and quickly read heaps of articles on physical media have some advantage over pathetic Internet-dependent users who need to use Google to search for information and dig around on different sites where they can use such pathetic tools as a search on the page.
zilyck@reddit
I still think there is a difference, but not sure what exactly it is. Maybe in my case AI would have helped, since I really only needed code for one project, so losing a lot of time learning something on a deep enough level for a shorter project might be a waste. But same as money, knowledge kind of compounds, no matter if you gain it through physical media or online. If AI solves all your problems without you knowing how, wouldn't you be stuck as an eternal junior as well?
CircuitryWizard@reddit
Well, first of all, AI is replacing juniors now, it can't replace middles at least because of the limited context window and hallucinations.
And as for the rest, it was a joke about how every time a new teaching method appears, people appear who start grumbling about how it used to be better.
Like Socrates, who believed that books suck because it's not like having a live dialogue with another philosopher.
Full_Advertising_438@reddit
I like to use AI at the beginning of a project as a sort of brainstorming session, for Unit tests. AI generates decent testing code, and even dummy data. And also if I get stuck over a really long period of time looking for the Bug, I copy paste the code describe the behavior and what I should expect and then see where the error was.
Historical_Copy_9004@reddit
TL;DR: Learned coding pre-AI, now use AI heavily across all dev lifecycles but maintain understanding of design decisions and intervene when AI gets stuck.
Hey, I really appreciate your honesty here! I learned to code before AI too and have been using Cursor/Claude Code heavily for two years. Here's the thing though - we never really coded "solo" anyway. Before AI, we had autocomplete, Stack Overflow, GitHub repos to reference. It's always been human + machine cooperation.
What's been working great for me: • Use AI for everything BUT stay in the loop - I incorporate AI into design, implementation, debugging, testing, ops. But I actively intervene when I see it heading the wrong direction • Maintain high-level understanding - Even if AI writes the implementation, I make sure I understand what the code does and why design decisions were made
• Document your thinking - I keep my thought process in the codebase (comments, docs, commit messages) so I don't lose context
Your "no AI" challenge is like saying "build without an IDE" - educational, but is it practical? The real skill isn't memorizing syntax anymore, it's knowing when AI is wrong and how to guide it. That classmate who never uses AI might ace exams, but in real work, efficiency matters too 😊
You're already on the right track by recognizing the issue and taking action! Maybe try a middle ground? Use AI but force yourself to explain every piece to a rubber duck first. You clearly have strong fundamentals from that WordPress plugin experience - don't sell yourself short!
g2i_support@reddit
You recognize the problem and you're already fixing it by building projects without AI. That's the solution - keep doing that. Your plan to build a PHP framework is smart since it forces you to understand fundamentals again.
The dependency issue is real but you haven't lost years of experience. Your WordPress internship skills are still there, just buried under AI habits. That classmate who scores highest without AI proves the skills matter more than the shortcuts.
Building without AI feels harder because you got used to instant answers, but the struggle is where actual learning happens. Your brain needs to work through problems to retain knowledge.
Keep the AI ban for personal projects. Use it sparingly for work if needed, but always try solving problems yourself first. The fact that you built a plugin that's still being used shows you have real programming ability - you just need to trust it again.
Tricky-Equivalent529@reddit
AI to learn = Good!
AI to give you direct answers == Bad!
Putnam3145@reddit
"AI to learn" seems to be killing peoples' brains pretty bad, too. "The AI is willing to give me an answer to a question [which had 5 correct answers on the first page of google] instead of tell me I could've googled it" tends to be the most common really positive take I see about AI in learning, which isn't a good sign.
Yetiani@reddit
naaaa they just don't know how to use it, you are supposed to grab it as a tutor, not to write code for you, create challenges for your specific skill set, reframe problems you already solved or suggest new paths to follow
CircuitryWizard@reddit
Well, learning on ready-made code is also learning, especially if you use several variations of code written by different models from which the best option was chosen and it was analyzed why this option is the best, or generally synthesizing all the received options into one piece taking the best of what was offered.
Yetiani@reddit
yeah but that advice without nuance can easily become what happened to OP. of course reading code can help but again you need to know when and what so it doesn't hinder your advancement
CircuitryWizard@reddit
So you have the superpower of analyzing code and writing code without knowing when and what is happening?
Yetiani@reddit
what are you talking about lol
No_Zookeepergame2532@reddit
Using AI to learn is literally no different than reading the info in a textbook.
Putnam3145@reddit
Except that it can randomly be completely wrong and this isn't even a bug, because nobody ever made a guarantee that the information it gives you would be correct!
No_Zookeepergame2532@reddit
You can even ask it to cite its sources so you can check the information yourself
-CJF-@reddit
The problem is that by the time you do all of that you could've just used the textbook and not had to manually verify everything, which can be extremely time-consuming for more complex tasks.
The reason it can be time consuming is because the AI often generates its answers by wrangling the data from web searches with multiple sources (Stack Overflow, Reddit Posts, etc.) and it fails often on nuance.
No_Zookeepergame2532@reddit
You can tailor it to only use accredited sources though
-CJF-@reddit
It can still be wrong very often because of the data it was natively trained on and its inherit pattern-matching nature, and that's especially dangerous to people learning because they presumably don't know if what it's telling them is correct—remember, they're using it to learn.
Note, I'm not saying not to use it. I sometimes use it myself, even for learning, but it's not a replacement for traditional methods (textbooks, documentation, stackoverflow, etc.), it's just... different. Kind of hard to explain, but I find it useful for fast-tracking a new language or discussing/reinforcing concepts I've already learned through traditional means. It definitely should not be a first choice for exposure to new topics and you can't trust anything it says.
No_Zookeepergame2532@reddit
You should be verifying with multiple sources anyway, no matter if it's AI or textbook.
-CJF-@reddit
You don't need to verify human-vetted sources of information like a textbook or official documentation. It can still be useful to use multiple sources because it helps to learn things in different ways, but textbooks are already vetted by multiple layers of skilled people and are almost never wrong in the same way that an AI often is.
roboticfoxdeer@reddit
and if you're lucky they might even be real!
No_Zookeepergame2532@reddit
They've always been real for me
laveshnk@reddit
20 years ago people said the same thing about google. How getting information at the click of a button was worse for the brain than wading through books of information.
Its the same thing now. Google also provides tons of incorrect information, Ive had docs that straight up lied about api structure and functions format that AI was able to help me debug and find out quite well
haidere36@reddit
The fundamental issue in all cases is that when you listen to an expert on a topic, they're presumed to have researched to that point that anything they say about it with confidence is probably true. When google provides incorrect information, it's almost always someone who isn't actually an expert in the topic.
The problem is that people treat AI as though it itself is the expert now. It's not "intelligent" and it doesn't actually "know" anything but people treat it as though it's spent years of rigorous study learning complex topics because it's designed to appear that way in order to entice people to use it.
There are countless examples at this point of AI making stupid mistakes that a human being simply wouldn't make, and that's because the human would know the information is incorrect and not spread it. AI doesn't know anything, so incorrect and correct information aren't distinguished.
Reliable information needs to come from reliable sources. "Google" is not a reliable source in and of itself, but it can lead you to such sources. AI can't.
Potential_Egg_69@reddit
AI certainly can if you ask it to explain the output with sources
That way you can see where it gets the code idea from and validate the piece of information explaining it
I think you'll find that most of the time, it's right
A larger and more complex problem means a worse answer. You have to break it up into smaller problems, and keep each chat focused on one topic.
And sure, maybe some people are at the cutting edge where the problems are novel - AI won't help you there. And tbh, neither will researching via Google
As others have said, it is a tool that requires human input. If you use it wrong; it will be shit. And if you don't need it - it seems pointless
scandii@reddit
that does not differ from any other source of information, sadly.
-CJF-@reddit
Actually, it's very different from human-vetted sources of information such as textbooks or even posts on Stack Overflow or Reddit. Of course, humans can still be wrong, but information and code that's been manually-vetted by a human is way less likely to be wrong. AI is wrong a lot.
That doesn't mean it's not useful for learning, but you can't just take what it spits out at face value. You have to run the code, test it, ensure it does what it should. Then you can take notes and relate back to the info. It's a useful and unique way of learning, but it's not necessary faster or more efficient.
scandii@reddit
we employ techniques like unit tests, code reviews and testers and yet we end up with critical bugs in production.
my issue is really just this emphasis on AI's "tendency to be wrong", when in reality it is seldom wrong in scenarios where it has a lot of data to draw on, e.g. simple learning examples - where it is borderline unusable when you are querying about niche things or issues in complex layered environments, which learning scenarios seldom are.
smadgerano@reddit
Humanity is in trouble.
No_Zookeepergame2532@reddit
Its been in trouble and AI isn't the reason
smadgerano@reddit
I use AI daily. It wasn't the use of AI that I was referring to. It's your understanding and justification of what it is that concerns me.
No_Zookeepergame2532@reddit
Whatever you say dude
roboticfoxdeer@reddit
Yes it fucking is different
No_Zookeepergame2532@reddit
You're right, AI can answer questions while textbooks cant
roboticfoxdeer@reddit
It's called having a brain and using it and being able to answer your own questions
No_Zookeepergame2532@reddit
Asking questions is a part of HOW you learn in ANY setting lmfao
Sheepherder-Optimal@reddit
I would say a textbook is a more reliable source BUT AI is an amazing way to learn something complex. I used AI to learn optometry and conduct my own vision exam. You can use it like a teacher and learn. Like with optometry, its not memorization. Its a scientific process. It involves optics and an understanding of how light works and what a lense does. Its fascinating.
awkreddit@reddit
Yes it is: it can be wrong, unlike a textbook. Also if you Google you can compare 5 textbooks snippets and get a better overall range of info.
LoL_is_pepega_BIA@reddit
What's the difference?
The AI will wade through bullshit search results to give me decent answers (often enough)
TheAxodoxian@reddit
I have learned a lot of useful things from reading search results which did not address my question.
I think that AI might be able to answer questions, you also need to be able to ask the correct questions. And need to be able to judge the answer.
Just to be clear asking the correct questions and being able to judge answers were already important before AI, and probably even more worthwhile now. Since it allows to utilize AI more efficiently.
For complex decisions I usually still start with normal searches, and if I cannot find an answer then I ask AI. Based on the AI answer I use more searches, as to understand the background and context the AI will not always tell (and to separate truth from hallucinations).
I sometimes use AI to generate code, but I only use it this way as to speed up coding of features which I could very easily develop (being limited by my typing speed). I still read every line of code, and only use code I fully understand and agree with. I work in a complex field, so even small mistakes can bite me later, so rather be fully confident that the code is correct, and also - maybe even more importantly - it does exactly what I think it does.
Putnam3145@reddit
Wading through the search results is an incredibly important skill that you need to know if you ever do anything non-trivial.
Illustrious_Matter_8@reddit
I prefer ai searches over google
LoL_is_pepega_BIA@reddit
At that point, naturally, AI is not very useful
It's really only good for getting to the useful stuff quickly when your level of knowledge is low
roboticfoxdeer@reddit
literally!! thank you for saying this
PsychologicalGur26@reddit
You probably already were
iShivamz@reddit
please provide the link for the tutorial playlist you used to understand Wordpress
Timanious@reddit
If you really want to understand Wordpress my advice is to download it from Wordpress.org and install it on a Raspberry Pi running a LAMP (Linux Apache MySQL PHP) webserver stack. Wordpress is just a website with a visitor and admin side to write posts etcetera (so basically two websites) that uses html, css, php, JavaScript and a database, like most other websites do so you can just study standard web dev tutorials. The default bare WP installation comes with maybe one or two standard themes and a few plugins so it’s easy to understand the code just by messing around with it once it’s installed it’s actually quite elegantly designed. By having it running on a Raspberry Pi (or other ‘other’ computer) you can figure out both the front and backend side of things relatively easily. I’m not a web dev but a game designer, so definitely not a web app pro but I never had any real problems figuring out WP and using it for my own blogs.
ryosen@reddit
Or just install it into a VM and don’t incur the cost of purchasing additional dedicated hardware to start the learning process.
Septem_151@reddit
It’s always useful to have a pi as a developer though. Not like you can only use it for that one thing.
JivanP@reddit
Absolutely not. The vast majority of free video content about WordPress is extremely poor, because it's such a popular platform and attracts people of all skill levels. The maxim that "those who can't do, teach" rings true here.
Refer to the official documentation and tutorials, they're really very good. Start here: https://learn.wordpress.org/
morakoshka@reddit
swanna check it out too
zedpapa@reddit
Seconding that
Fun-Helicopter-2257@reddit
how AI could make you even more stupid if you cant figure WordPress by yourself?
84_110_105_97@reddit
tkt man this happened to me, the thing is a real drug, and very hard to stop, (I stopped AI completely a few years ago now I am progressing again in the field in C, C++, php, rust, ect...)
I'm sharing a video with you that will help me: https://youtu.be/mZDhRjzV1Hk?si=btY1O5qMQUdfz5eQ
ThatCrankyGuy@reddit
It's a tool, like any other tool you need to know how to leverage it to your advantage.
SynthDude555@reddit
I'd argue you're the tool
ThatCrankyGuy@reddit
my wife thinks so
SynthDude555@reddit
I upvites that one because it honestly made me laugh.
Sheepherder-Optimal@reddit
Right. Like an IDE is a tool and it usually has AI built in. Are you also jealous of people who code without an IDE cuz they're soooo smart?
Twig@reddit
I think you're missing multiple points lol
ItzDubzmeister@reddit
I like to use AI for two things: 1. If I figure out a solution for some problem, I’ll paste the code and ask if there are any ways to improve, solving the problem myself then learning what I can improve on.
Treat AI like you would a professor, if you ask your teacher for the answers for a test or hw assignment, they wouldn’t give it to you.
BananaComfortable114@reddit
Me too, but if I get stuck for many hours/day, I’ll ask for the answer lol.
Caravaggio91@reddit
Exactly how I currently use it
BohemianJack@reddit
This is exactly the right use case.
I was having trouble unpacking a crazy yaml file structure and tossed it what I currently had, what the goal was, and it spit out a result that worked. Turns out there was a built in Terraform function that I hadn’t used yet that ended up solving my problem.
Use it as an assistant and you’ll get good results. Rely on it to do everything for you then you’re in big trouble
DensityInfinite@reddit
Good rule of thumb is to never ask AI to write code for you.
Instead of "Solve this problem," ask, "How would you approach this?"
One can ask for hints to a problem. One can also ask how it may approach/solve a problem if they get really desperate. But they should NEVER copy paste code from chat. Understand the logic behind the bot's response and use that to code up a solution yourself.
BlackMarketUpgrade@reddit
Yep this is the best way to go about it. I am a student and have given specific instructions to only write in pseudo code to explain concepts and only give me solutions when I explicitly ask for them. Also, all of my notes I write get pasted into gpt to make me flashcards based on those notes that I run everyday. A good tip is to basically just integrate AI into old school study techniques.
The way you have to think about AI is that it is a force multiplier. If you want to be lazy it’s a force multiplier for your laziness. If you want to really learn something it can be a force multiplier for that as well but it’s up to you.
DotAtom67@reddit
this is best possible advise for people learning. As they are struggling with keywords and syntax, adding design patterns makes things too hard for beginners, so letting the AI "help" with that is okay at that level.
DotAtom67@reddit
thing is, overtime you wont be just learning but meeting deadlines and focusing on delivering a product instead of only learning. So you won't query the AI for every bit of code it churns out, nor you will review it line by line, as if you were learning.
Add to that you will be getting the hang of doing stuff, ans that will also make you less prone to interrogate the AI about everything, as you will feel you are already at a good level.
Using AI is like using drugs to lift more weight.
Wonderful-Habit-139@reddit
This is the most realistic comment. People like to say “if you use it AI correctly” but 95% of the time they end up blindly copying and pasting back and forth between the codebase and the AI.
teslah3@reddit
I do this and even ask it to grade my work from 1-10 (not just for coding but sometimes just essays or whatever I want feedback on). I take what they say and rarely use the exact suggestions they give, but rather modify or adjust what parts could be better using my own logic and reasoning. AI is a great tool if you know how to use it.
Historical-Loan6076@reddit
AI is good for deep dive. It’s as powerful as having personal tutor in your pocket who is ready to answer you any doubt you have without getting angry for asking stupid questions and someone who doesn’t judge. I always wanted something like that.
gmdtrn@reddit
It does that. LLMs are very useful, but use them wisely.
ElectroEsper@reddit
I get you. AI can turn into a drug if not careful, I had grown addicted to it for a bit and my coding skills sank as a result.
Now, I've recovered, and settled on using it as a "Thinking tool", to "talk to myself" if you want.
Kwith@reddit
The only thing I use AI for is generating ideas for my D&D campaign. Even then it basically just takes my ideas, organized, expands, and makes it easier to read.
Pmart213@reddit
That’s your own choice to use it and hinder yourself. Just literally don’t open it when you have a problem and instead figure it out, and you will find yourself retaining a lot more knowledge.
As someone who has been working in the career for a long time now, it’s extremely obvious the devs that use it and rely on it, and everyone loses all respect for them because real applications and businesses that make millions have a lot of very proprietary issues and challenges that arise that AI or even google cannot help much with, and the people that use AI lose the ability to intelligently figure them out themselves. They also submit code that they genuinely don’t fully understand and often does not fit perfectly with the existing codebase.
Don’t shoot yourself in the foot brother, because in a career at a large company it will 100% be noticeable to others and also cause you to not progress as much as you can.
Spaaze@reddit
Will probably sound like an advertisement, but promise it’s not. I recently found CodeCrafters which helped me a lot with this. It’s basically Codecademy for advanced developers. They make you rebuild popular tools like Git and Redis in a language of your choice, leading you through the process step by step without spoonfeeding solutions. I’m about halfway through Redis and had a lot of fun with it — obviously, with all AI features in my IDE disabled. Highly recommend trying it out.
Calm-Safety4029@reddit
Hi personally I use AI just to help me if I have a bug and I ask it for explanations otherwise I don't ask it to code for me otherwise I don't see the point in coding
Puzzled-Ad8231@reddit
I think learning from AI is like a DFS, you get an option and completely exhaust it till the end.On the other hand, The conventional learning process is like a BFS, where you explore each option and go deeper in them eventually. Here by option I mean, all the possible solutions that we have for a problem. Either of them is good, if you are willing to parse the entire graph, you will do it eventually. Obv DFS looks like a better approach cuz it's fast but might not give a bigger overview of the entire problem
CircuitryWizard@reddit
Do you know that you can ask AI to solve a specific problem with several dozen possible methods of solving this problem and AI will do it? And you can ask it to redo each possible option, ask it to explain, etc.?
Puzzled-Ad8231@reddit
If the objective is to just get the work done, then no issues. You will have your solution but won't know anything about what's happening. For long term benefits, only learning from AI works. I have seen some of my batchmates making really impressive projects but during the interview when asked bout the internals and basic concepts used, they were very weak. Knowing something doesn't only mean knowing how it works. It also means knowing and realising the shortcomings and being able to foresee what will be better for future scope of your project/product.
I was reading an article I don't remember the name of, Some great ceo was quoting in an interview that, the best engineers that he knows are not just the people who are on top of AI. The best engineers he met are people who are really senior and have great understanding of the underlying architecture of the particular software and then are also on top of AI and using it as an amazing assistant for works.
CircuitryWizard@reddit
I meant that if you ask the AI for one solution, it will give it. But at the same time, it can provide a bunch of possible different solutions upon request (besides, for example, I usually work in several different models in parallel) that are created specifically to solve a given problem. And in fact, if earlier training on other people's examples was more focused on finding the most suitable option while filtering out a bunch of unnecessary solutions, then here there is an immediate transition to the next stage - studying possible options and choosing the best one with a study of the reasons why this is the best option. From my point of view, such tactics will lead to better code in the future because earlier one possible solution was chosen that they tried to optimize for solving the problem and they switched to another method of solving the problem only if the existing solution did not satisfy.
And I agree that standard juniors are no longer particularly needed, because there are already AI "juniors" who, although they do not learn, work faster. And regarding training, AI can easily create a full-fledged personal training plan, almost instantly explaining unclear points, unlike self-study on the Internet, when you can wait weeks for an answer to a question, studying at a university, when you may not get an answer, and it is cheaper than personal training. Just keep in mind that AI, like a person, can lie and not admit it.
You just need to separate low-quality vibe coding from training with the help of AI.
Puzzled-Ad8231@reddit
Really well said... Intelligent use of AI is the key. Now "intelligence" is what we have to discuss
0011001100111000@reddit
I tend to use LLMs as a glorified Google search.
I tend to ask it theoretical questions without giving it code and ask it to explain its answers more if required.
Often, it'll give you some pseudocode which gives you the idea, but you still implement it yourself so you understand what the code is doing.
I tend to only use it for things I'd usually Google.
I tend to only use code generation for things that are boring and easy, like unit tests, or building a model class from a SQL CREATE TABLE statement.
DetectandDestroy@reddit
I feel like this is the century old argument of “guns are bad because it makes killing easier”. If you don’t put the work into learning how the things work sure AI can do the work for you as shitty or good as your prompt is. Eventually if you wanted to get good at shooting and wonder why certain guns jam when you’re shooting it like an idiot then you’re gonna have to learn how it works. Same with coding. If you’re producing garbage you need to understand why. If you’re not reading documentation and actually learning the language then you’re just shooting yourself in the foot. I argue that AI could get curious people motivated enough to learn more about the topic but on the flip side, if you’re cave manning everything and not learning that’s more a you issue. As a society we need to stop thinking of AI as autonomous and think of it as a tool.
ZelphirKalt@reddit
Actually the things you are describing ... make me think you have a pretty good introspection and reflection upon your skills/knowledge. It is not uncommon for people in the WP ecosystem to think that the basic (and often horrible) things they do in their plugin development amount to all there is to programming, or that they don't need to learn more "because it works". I applaud you for recognizing that there is much more to learn. If you keep it up, you will go far! What I am referring to is:
So. All great insight. The good news: You can quickly get your "lost" experience back up, once you force yourself to think more for yourself, or change the way you use AI tooling. If you are going to use it, don't let it write code for you in your source code. Rather ask it questions and analyze the approaches it comes up with. Then carefully consider, whether the approach fits your project.
Xatraxalian@reddit
In the past, if you didn't know something, you could try to ask your teacher. You might get a pointer here or there if something was really difficult, but often, the answer was: "You have to find out by yourself or you won't learn how to do it."
AI is like the teacher that always gives you the answer you ask for... with the side effect that sometimes, it's actually wrong. Only you're not learning how to do it yourself, you may actually be learning the wrong thing.
AI is not a good tool for learning; it is a good tool for supporting you and finding things faster than a search engine, but you HAVE to know stuff already or you will be unable to verify what the AI is telling you.
Djblackberry64@reddit
Here is a resource repo I made with an AI tutor prompt so AI helps you learn and not save what to ask it. Link here: https://github.com/djblackberry64/Lempire-resource-bunker
Calm_Woodpecker_9433@reddit
Have you ever tried to "learn" WordPress framework from AI, instead of asking AI to "write" the code for you?
adnanMOV@reddit
I feel ya same thing happened to me, i used to be that guy (overachiever) Topper in every sub but since i was so much into tech stuff i started using AI too early and iv seen all the versions and updates of those chatbots coming and me trying new models in 2022 but eventually it made me dependent..
Jebble@reddit
If your "skills" degredaded so quickly after the tiny bit of AI we've had, then your skills were never really there.
TimedogGAF@reddit
AI did not "make" you bad at coding. You made yourself bad. AI didn't force you to do anything.
The attitude that "AI forced you" to do something is what needs to be fixed. Somehow who does not harbor such attitudes, where they shift the onus of responsibility away from themselves, is far less likely to have the problems you have with AI.
Stopping the use of AI won't solve the fundamental problem.
JohnnyOmmm@reddit
Duh
BrilliantRaisin915@reddit
What all my Professors have told me: AI as a coach ✅ AI as a decision maker ❌
Illustrious_Matter_8@reddit
So ai is the professor Eventually it's always you to decide
recontitter@reddit
There is this new option in chat gpt, teach me or something, I tried to use it and is good. Instead of giving you solutions, it is using Socratic method and challenges you. Tried it with infamous JavaScript promises and I kind of like it. It’s very much proven learning method but neuroscience and in pedagogy in general.
DotAtom67@reddit
Why it's always like that hahaha, have been there too
PineapplePiazzas@reddit
I bet you made this shitpost with ai, Aye!
cronixi4@reddit
I use AI as a tool to learn, I’m mainly interested in Web analytics, SEO and automations. I’ve let AI write a course for me tailored to this. I broke it down in to days, a total of 12 weeks. Just small concepts every day.
Besides providing enough exercises to really grasp what you learn each day and adding small projects in between to connect it all, it is also doing a amazing job to explain harder topic in a easy to understand way.
I will always ask it to review the solution I came up with and ask if there a better ways and what is the best practice as there are usually many solutions for 1 issue.
Use AI as a tool, a extension to your knowledge. Using it to write it all for you is waiting for a disaster to happen. Only use it for repetitive tasks that you know and can double check.