There's a bit of drama going on with the popular game manager Lutris right now, with users pointing out the developer using AI generated code via Claude.
No reason to hide the Claude contributions if his reasoning isn’t flawed. Honestly, my biggest beef is using AI as a subscription service, there are plenty of local LLM alternatives, and that just feeds the incentive for the web crawlers currently assaulting the Internet, for the companies also tucking in surveillance and training on your use of their AI.
Honestly, it can save you having to search through wordy API documentation, as long as you bother to make sure to make sure you end up knowing the hows and whys of what it is presenting you and whether it is good programming methodology. In a lot of ways it is no different and even faster than having to search for the answers through support forums and stack overflow. It might be built upon IP theft, but unfortunately in practical terms, you will be at a disadvantage against people that use it, so you might as well use it in a way that does not give them any way to profit off of it (Local LLMs). I’d argue that the case against these applications of generative AI is way different than those for image and video generation.
The biggest problem is when developers begin to depend on it too much without learning the nuance, but it would be a lie if there aren’t a lot of developers who contribute to Open Source without really bothering to familiarize with it already and who are more interested in the end goal than best practices. Not sure if this will make the problem better or worse, but devs who use AI without bothering to learn will have a hell of a time providing proper maintenance for their code.
Not disagreeing with you, but Anthropic believes code is the path to AGI.
I want to be clear so somebody doesn’t have a fit - I do not personally believe LLMs are capable of AGI. But this isn’t about what I believe.
They believe that coding is the path because it’s verifiable and a generatable. Frontier AI companies aren’t training on the global internet anymore, it’s poisoned with AI slop. Non-frontier AI companies do, we’ve all seen it. But it’s my opinion that non frontier AI companies are basically all but irrelevant (I’m not talking about open source/hugging face). Anthropic knows this, and their idea (again, not mine, don’t get mad at me please!) is that by training on code their AI will get better at non-coding activities as well, and if they make it good enough at coding it’ll become truly intelligent in all ways.
What I’m getting at is, there’s lots of good reasons to avoid using LLMs/AIs/Companies that shove ai down my throat (looking at you Microsoft- I don’t fucking want copilot in my fucking notepad - if anybody from MS is reading this fuck your AI in everything and fuck your AI ridden operating system), but local LLMs are not a replacement for Opus and Anthropic isn’t scraping the open internet anymore. I’m sure they did at first though.
The biggest problem is when developers begin to depend on it too much without learning the nuance
I couldn’t agree more. The brain is like a muscle, if you use it, it gets stronger. If you don’t, it gets weaker. “Vibe” coding is using your brain at a minimum, and if all you do is vibe out slop you’re not really learning much.
It’s only slop if you don’t know what you’re doing and/or are using low quality tools. But I have over 30 years of programming experience and use the best tool currently available. It was tremendously helpful in helping me catch up with everything I wasn’t able to do last year because of health issues / depression.
It sounds like they thought it through and decided it’s the best way to do the work. Removing the attributions seems like a little bit of a petty “fuck you”, but so is opening a github issue just to whine about AI. Someone who is volunteering their time to make free software shouldn’t have to put up with people with an ideological bone to pick who feel entitled to tell them how.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: [email protected]
No game suggestions, friend requests, surveys, or begging.
No Let’s Plays, streams, highlight reels/montages, random videos or shorts.
No off-topic posts/comments, within reason.
Use the original source, no clickbait titles, no duplicates.
(Submissions should be from the original source if possible, unless from paywalled or non-english sources.
If the title is clickbait or lacks context you may lightly edit the title.)
No reason to hide the Claude contributions if his reasoning isn’t flawed. Honestly, my biggest beef is using AI as a subscription service, there are plenty of local LLM alternatives, and that just feeds the incentive for the web crawlers currently assaulting the Internet, for the companies also tucking in surveillance and training on your use of their AI.
Honestly, it can save you having to search through wordy API documentation, as long as you bother to make sure to make sure you end up knowing the hows and whys of what it is presenting you and whether it is good programming methodology. In a lot of ways it is no different and even faster than having to search for the answers through support forums and stack overflow. It might be built upon IP theft, but unfortunately in practical terms, you will be at a disadvantage against people that use it, so you might as well use it in a way that does not give them any way to profit off of it (Local LLMs). I’d argue that the case against these applications of generative AI is way different than those for image and video generation.
The biggest problem is when developers begin to depend on it too much without learning the nuance, but it would be a lie if there aren’t a lot of developers who contribute to Open Source without really bothering to familiarize with it already and who are more interested in the end goal than best practices. Not sure if this will make the problem better or worse, but devs who use AI without bothering to learn will have a hell of a time providing proper maintenance for their code.
Not disagreeing with you, but Anthropic believes code is the path to AGI.
I want to be clear so somebody doesn’t have a fit - I do not personally believe LLMs are capable of AGI. But this isn’t about what I believe.
They believe that coding is the path because it’s verifiable and a generatable. Frontier AI companies aren’t training on the global internet anymore, it’s poisoned with AI slop. Non-frontier AI companies do, we’ve all seen it. But it’s my opinion that non frontier AI companies are basically all but irrelevant (I’m not talking about open source/hugging face). Anthropic knows this, and their idea (again, not mine, don’t get mad at me please!) is that by training on code their AI will get better at non-coding activities as well, and if they make it good enough at coding it’ll become truly intelligent in all ways.
What I’m getting at is, there’s lots of good reasons to avoid using LLMs/AIs/Companies that shove ai down my throat (looking at you Microsoft- I don’t fucking want copilot in my fucking notepad - if anybody from MS is reading this fuck your AI in everything and fuck your AI ridden operating system), but local LLMs are not a replacement for Opus and Anthropic isn’t scraping the open internet anymore. I’m sure they did at first though.
It sounds like they thought it through and decided it’s the best way to do the work. Removing the attributions seems like a little bit of a petty “fuck you”, but so is opening a github issue just to whine about AI. Someone who is volunteering their time to make free software shouldn’t have to put up with people with an ideological bone to pick who feel entitled to tell them how.