aselimov3

I disagree. A good developer that heavily uses AI just becomes average without realizing it. To remain a good programmer you have to keep actually programming.

Has an issue where the last fishes fight shows up during the wait for the current fishes fight. More importantly how can you afford this? Seems like every call goes to an LLM? Seems like it would be pretty expensive for a free game.

I definitely support maintainers defending themselves but this seemed just like a petty slap at LLM users. From my understanding, any user who used this testing library was vulnerable to the prompt injection.

Overall LLMs are certainly a net negative on humanity, but I don’t think being mad about it or their users is the best response. I really respect Andrew Kelly in this regard. He doesn’t accept LLM input into Zig and is generally anti-LLM but his approach is positive. “LLMs aren’t that good and it’s boring to use them. Check out how cool/fun/high quality real coding is”

Yea good explanation of what I meant. I understand the moral/ethical arguments against LLMs and they are compelling but the actions of the maintainer above seemed just like a petty swing at LLMs just for the sake of it. Not very rational in my opinion.

I’m trying to minimize my use of LLMs because I think they are harmful personally but I don’t get mad at people that use them (unless they are just spamming slop but then I just ignore it)

I don’t think so but maybe? I do use them in daily work so I might be compromised. But I also generally dislike their impact on humanity and try to limit my use where feasible for my own brain’s sake.

Personally I think Andrew Kelly’s take is the best. Basically not interested in LLMs but if someone uses them to do something cool then cool I guess?

I can’t remember where I read or watched this, but the argument was that the main problem with a lot of software is simply code base size. Once your code base is large enough , it can no longer really be understood as a whole which leads to performance issues/bugs/vulnerabilties. I think that’s just an unavoidable part of large software.

Agentic engineering is probably not helping since it tends to accelerate growth of LOC. I also personally think quality is an issue as well but it could be a skill issue on my end idk.

I definitely see where he is coming from, but his response was pretty bad. Seems like he has anti-ai psychosis that went way too far.

This gives similar energy to that guys npm package that deleted Russian users computers. Overall not a great look and any difficulty with job searching/conferences is probably well deserved.

Easy answer is large unmaintainable code bases + “agentic engineering”

Man this is pretty draconian. I suppose it’s not surprising coming from Britain with their significant anti freedom efforts

I’ve always thought that this is the argument people that say AI is conscious are making without realizing.

It’s also utter hogwash. Writing an article to argue against your own consciousness/free will is such an incoherent action that it doesn’t deserve engagement.

The most offensive part is that the author doesn’t live as though he doesn’t have free will because no one actually believes that.

Very curious to see how efficient this will be in practice

Is this much different than ziglings?

People stop creating music, stop writing blogs, stop doing experiments or any other cool stuffs and share it on the internet because LLM companies can just pirate the shit out of it without paying anything back.

This isn’t necessarily true. I still write my own blog by hand and write code by hand. Sure it gets fed into the LLM meat grinder but I think a lot of people are going to fight against the LLM slop and consume real human content

Even if it was forever is run by community really that great? The code is open source so a fork can happen at any time if it goes down the wrong path. Once you get a wide community involved I feel like decision making slows down. And the product gets worse.

I’m a big open source fan but I’m starting to lean to projects that limit contributions/have a strong leader since they tend to be higher quality in my experience.

Why doesn’t this link the project?

This is pretty cool but how strong is the overlap between developers and people writing recipes?

I don’t receive that much personal communication that’s AI copy paste. It is somewhat easy to tell when someone is writing with AI. I usually don’t read articles that have strong AI tells for example

I don’t use ai for any human to human communications. I don’t even really use it for grammar anymore. I’d rather post my own authentic style (blunders and all) then get my thoughts blended by the language grinder into sludge

This is definitely the way. I don’t think long term that full agentic engineering without writing code is sustainable. I think it definitely leads to atrophy

This is quality

This is such a bad take it’s crazy. Assuming a higher IQ person has better long horizon thinking and levels of abstraction seems blatantly false.

Recent example is SBF (and crypto buddies) who probably had a high IQ but had pretty short term thinking I would say.

Still trying to use IQ as a measure of intelligence is also pretty wild. Plenty of things high IQ people can’t do (like relate to others as evidenced by the post).

(Note: I don’t know my IQ so this might just be low IQ cope)

I like Hugo for my personal site but getting exposure is a little hard. I don’t want to constantly blast my blog everywhere

Good reasonable take. I think the tradeoff on AI automation is trading skill development, brain muscle growth, and technical understanding for speed. You have to balance it so you are net faster but also growing. Otherwise you are just indistinguishable from any other prompter.

I’m unconvinced by people specializing in “agentic engineering techniques” like managing agent swarms/spec writing/harness tuning. Regardless of how complex your set up is I don’t think you are actually growing your skillset/problem solving abilities if your only interaction with code is prompting. I also think those “skills” have a finite lifetime as each generation of model and harness requires different ways to hold them.

Eventually if the AI promise is meant to happen (which I’m skeptical), LLMs will be able to fully manage their own swarms/write their own specs without requiring our input. Only spending time learning how to get better results from today’s AI is basically a waste for your long term growth.

Missing the link

Site broken for me

lol paid 1k to play doom with worse performance probably

Man that’s a hard one because personal browsing for things like banking would require a GUI. But I could use a separate machine for that. This is a pretty decent experiment to run.

I was mainly just speaking to the point of the article with regards to ability to think about code/solve technical problems. I think most EMs I speak to would say their technical skill set is degraded from years not writing code. I think if you just review code without writing it, your technical skills degrade regardless of whether you are reviewing junior or LLM code. In both cases you maybe gain architecture design and soft skills but you are definitely not improving your technical skill set

Sometimes I get the "you're right!" response, but often it will also explain why it made the decision it did, and it's rational enough that I accept the new approach.

I think this is kind of what I’m worried about. Referencing Karpathy, an LLM can basically convince any one of anything. Doing this enough times and your opinions just become the LLM opinion. Same with problem solving approaches. I see this for myself so I reserve time to formulate my own solutions. Maybe my solutions are worse but at least I’m training the muscle that may lead to me outperforming the LLM in specific spaces

But the specs can’t be fully detailed or else they would just be code which is the point the article above makes. You can certainly limit where the LLMs have freedom by writing more detail but obviously that defeats the purpose of LLMs. What I mean more broadly though is more along the lines of baked in solutions. The LLM will generally prefer certain architectures/libraries/patterns. If you develop a spec alongside an LLM (which I think most people do) you will get pushed towards those. You kind of close yourself off to other ways of solving a problem.

You can stop this by detailing exactly what solution you want but I think continuous leaning on the LLM will lead you into a confined space.