as a junior developer, i am nervous. what used to be earlier given to junior devs is now being crunched by LLMs, how do i future proof myself?
Don't fall into the anti-AI hype
891–900 of 1001 posts
Re: Don't fall into the anti-AI hype
#892Earlier quoted context omitted.
> non-trivial coding tasks I’ve come back to the idea LLMs are super search engines. If you ask it a narrow, specific question, with one answer, you may well get the answer. For the “non-trivial” questions, there always will be multiple answers, and you’ll get from the LLM all of these depending on the precise words you use to prompt it. You won’t get the best answer, and in a complex scenario requiring highly recurs…
I wish LLMs were good at search. I've tried to evaluate them many times for their quality at answering research questions for astrophysics (specifically numerical relativity). If they were good at answering questions, I'd use them in a heartbeat Without exception, every technical question I've ever asked an LLM that I know the answer to, has been substantially wrong in some fashion. This makes it just.. absolutely us…
LLMs got good at search last year. You need to use the right ones though - ChatGPT Thinking mode and Google AI mode (that's https://www.google.com/ai - which is NOT the same as regular Google's "AI overviews" which are still mostly trash) are both excellent.
I've been tracking advances in AI assisted search here - https://simonwillison.net/tags/ai-assisted-search/ - in particular:
- https://simonwillison.net/2025/Apr/21/ai-assisted-search/ - April is when they started getting good, with o3 and the various deep research tools
- https://simonwillison.net/2025/Sep/6/research-goblin/ - GPT-5 got excellent. This post includes several detailed examples, including "Starbucks in the UK don’t sell cake pops! Do a deep investigative dive".
- https://simonwillison.net/2025/Sep/7/ai-mode/ - AI mode from Google
Re: Don't fall into the anti-AI hype
#893Earlier quoted context omitted.
Your comment sounds like John Glenn's quote "Get the girl to check the numbers… If she says they’re good, I’m ready to go." about Katherine Johnson to double check the calculations done by the first computers used by NASA. At that time in history, it was probably accurate and the safest thing to do, but we all know how computer evolved from that time and now we don't have human calculators anymore but rather human ch…
But LLMs aren't advertised as some future thing. They're advertised as being almighty and replacing devs in great numbers. And that's simply not true. It's a fad like 3D movies
I know they are pumped and overhyped to death, indeed they are. But that does not mean that they already have some use today and that they can (or not) improve in the future.
I'm skeptical about LLMs as well but I also wanted to see what they are actually capable of doing and I vibe coded an Android app in Kotlin (from scratch) with Claude Code and Opus 4.5 and it basically worked. I'm pretty sure the code is horrible to the eyes of a Kotlin developer because I added so many feature by asking CC to do it over the last 2-3 weeks that it already desperately need a refactor.
But still, this is not something an autocomplete would be able to do for you.
Re: Don't fall into the anti-AI hype
#894Earlier quoted context omitted.
It's the second. Like, yes, prompting is a skill and you need to learn it for AI to do something useful but usefulness quickly falls down a cliff once you go past "greenfield implementation" or " basically example code" or "the thing done a lot so AI have a lot of reference to put from" it quickly gets into kinda sorta but not really working state. It can still be used effectively on smaller parts of the codebase (I…
People already say here that they don’t even look the code anymore. ”That is AIs job”. As long as there is a spec and tests pass, they are happy! I just can’t do that.
It's just the next rung on the enshittification ladder. So many steps in our "progress" to enlightenment as a society, as a technology community, is just abstracting away work with a "good enough" solution that is around an 80% solution
That's fine for the first iteration or two, because you think "oh man this is going to make me so productive, I'll be able to use this new productivity to wring 40% of progress out of that 20% gap"
But instead we just move on to the next thing, bring that 20% shittified gap along with us, and the next thing that gets built or paved over has a 20% gap, and eventually we're bankrupt from rolling over all that negative equity
Re: Don't fall into the anti-AI hype
#895Earlier quoted context omitted.
The biggest frustration with LLMs for me is people telling me I'm not promoting it in a good way. Just think about any product where they are selling a half baked product, and repeatedly telling the user you are not using it properly.
But that's not how most products work. If you buy a table saw and can't figure out how to cut a straight line in a piece of wood with it - or keep cutting your fingers off - but didn't take any time at all to learn how to use it, that's on you. Likewise a car, you have to take lessons and a test before you can use those! Why should LLMs be any different?
I'm glad you brought up the power tool analogy - I've bought a $40 soldering iron once, which looked just like the Weller that cost like 5x as much. There was nothing wrong with it on the surface, it was well built and heated up just fine.
But every time i tried to solder with it, the results sucked. I couldn't articulate why, and assumed I was doing something wrong (I probably was).
Then at my friends house, I got to try the real thing, and it worked like a dream. Again I can't pin down why, but everything just worked.
This is how I felt with LLMs (and image generation) - sometimes it just doesn't feel right, and I can't put my finger on what should I fix, but I come away often with the feeling that I needed to do way more tweaking than necessary and the results were just still mediocre.
Re: Don't fall into the anti-AI hype
#896Why is this article getting so many upvotes? It follows the same pattern -- I was able to use [AI codong agent] to achieve [task], [task] and [task] within [time]. It would not be possible to do that without it. [My thoughts about this] Which is the same as dozens if not hundreds of similar articles already posted here, and the comments in the discussion don't explore any new perspectives either. I honestly don't und…
I'm doubting [My thoughts about this] is actually author's, why write blog manually, automate that too
He didn't use an LLM to write this.
Re: Don't fall into the anti-AI hype
#897I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…
I'm exactly on the same boat. To anybody who want to try, a concrete example, that I have tested in all available LLMs: Make a prompt to get a common lisp application which makes a "hello triangle" in open gl, without using SDL or any framework, only OpenGL and GLFW bindings. None of the replies even compiled. I kept asking at least 5 times, with error feedback, to see if AI can do it. It did't work. Never. The best…
It sounds like you're using LLMs directly, instead of a coding agent. Agents are capable of testing their own code and using that to fix issues, which is what makes them so powerful.
Using Claude Code, I was able to successfully produce the Hello Triangle you asked for (note that I have never used CL before): https://github.com/philpax/hello-triangle-cl
For reference, here is the transcript of the entire interaction I had with CC (produced with simonw's excellent claude-code-transcripts): https://gisthost.github.io/?7924519b32addbf794c17f4dc7106bc2...
Edit: To better contextualise what it's doing, the detailed transcript page may be useful: https://gisthost.github.io/?7924519b32addbf794c17f4dc7106bc2...
Re: Don't fall into the anti-AI hype
#898I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…
I'm exactly on the same boat. To anybody who want to try, a concrete example, that I have tested in all available LLMs: Make a prompt to get a common lisp application which makes a "hello triangle" in open gl, without using SDL or any framework, only OpenGL and GLFW bindings. None of the replies even compiled. I kept asking at least 5 times, with error feedback, to see if AI can do it. It did't work. Never. The best…
"Please write me a program in Common LISP (SBCL is installed) which will render a simple "hello world" triangle in OpenGL. You should use only OpenGL and GLFW (using sbcl's FFI) for this, not any other existing 3D graphics framework."
This worked in codex-cli, albeit it took three rounds of passing back the errors. https://gist.github.com/jamesacraig/9ae0e5ed8ebae3e7fe157f67... has the resulting code.
Re: Don't fall into the anti-AI hype
#899I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…
That's been pretty much exactly my experience too. For what it's worth, multiple times in my career, I've worked at shops that once thought they could do it quick and cheap and it would be good enough, and then had to hire someone 'picky' like me to sort out the inevitable money-losing mess. From what I've seen even Opus 4.5 spit, the 'picky' are going to remain in demand for a little while longer still. Will that la…
You can be picky with Opus, just yell at it to refactor a few times. To reduce refactor cycles, give it correct and enough context before you start along with expected code style, etc. These things aren't one shot magic machines.
Re: Don't fall into the anti-AI hype
#900Earlier quoted context omitted.
This is a pretty common position: "I don't worry about getting left behind - it will only take a few weeks to catch up again". I don't think that's true. I'm really good at getting great results out of coding agents and LLMs. I've also been using LLMs for code on an almost daily basis since ChatGPT's release on November 30th 2022. That's more than three years ago now. Meanwhile I see a constant flow of complaints fro…
> I don't think you can just catch up in a few weeks, and I do think that the risk of falling behind isn't being taken seriously enough by much of the developer population. This is nonsense. This field moves so fast the things you did more than a year ago aren't relevant anymore. Claude code came out last year . Anyone using random shit from before that is not using it any more. It is completely obsolete in all but a…
You're accusing me of spreading harmful advice here, when you're the one telling people that they don't need to worry about not investing in their skills because "This field moves so fast the things you did more than a year ago aren't relevant anymore."
One of us is right here. I hope for your sake and the people that listen to you that it's you. I don't think it is.