There been Discussions here:
https://codeberg.org/ethical-foss/open-slopware/pulls/1482
https://www.reddit.com/r/emulation/comments/1x5eltg/retroarch_fork_being_discussed_due_to_heavy_llm/
Saying it’s not a Good Thing for Retroarch and saying when AI being Used should be Forked from Retroarch
Just like to know what People Say and Think about this
That’s why i stopped updating since July. I’m afraid of potential increased amount of regressions because of this and i don’t have the time to troubleshoot as much anymore.
For reference, here the commit history for RetroArch: https://github.com/libretro/RetroArch/commits/master/
I do update RetroArch from source, it is build automatically and installed. During the last couple of months there was only one game breaking issue, where the audio didn’t work anymore. Which has been solved quickly. But this could have been prevented if the devs would do it the old way, I believe. So I’m worried about the quality control here, but on the same time its my own fault to install a current development version. So fair enough I guess.
Not gonna lie, when I noticed this, it was a bit disappointing. My general stance on Ai has changed a little bit and I do not reject everything with Ai automatically anymore. But there is another issue here. See RetroArch had a few issues with some people who didn’t like it for various reasons. Now this Ai things will give them even more food and many will reject the project because of it. And this is a shame, because RetroArch is one of the best pieces of software.
And who is ready to fork and work on it? Without using any Ai? Will it stay compatible to the cores, configuration and user interface?
Ai is the future, people who reject it at whole will have barely any choices.
When they have there Stable/Offical Release they would have to make it sure it works very well without Issues?
I guess yes. But this becomes harder to do with so many Ai commits and changes. And the update cycles getting long, its been 11 months now. Looking at other Ai driven development, there are reasons for concern. But its really too early to make final judgements. I just feel uneasy about this at the moment, that’s all. I’ll update every 2 weeks or so from current development version, and it usually works.
There is no alternative, and even if there was a fork, I wouldn’t trust it for a long time. I have a trust issue and hope RetroArch won’t break my heart. Our relationship is on a stress test. And let’s be clear, I’m the one who stays at home and RetroArch devs do all the work. There aren’t many who help with coding, so I can somewhat understand using those helpers. My only hope is, it won’t eat them alive. OK enough of funny words.
I have one Folder with the Official Release as my Main one I use to play games
Then I have also one for the Nightly Version so I can test with the new updates that come out.
So maybe that something that people could try with the New AI code coming in?
On my Steam Deck I use the Steam version of RetroArch, which only updates with full releases. And I miss a few features or changes there. I could change the settings so it updates more frequently from current development, but I’m not sure about the consequences. On my Steam Deck I just want it to work without worrying, and it currently does. It’s like having your ex and new gf at the same time, because the new one is -potentially- unstable. Most of the time it works, but I still can’t sleep well at night.
I use PC and iOS Versions so I can’t really help with the Steam Deck Version
Isn’t the Nightly Version have a Warning that it can be Unstable?
I am not sure why this matters? The only thing that should matter is the final product, whether or not it is a good application, and so far, Retroarch is. How the code was made is irrelevant. At the end of the day it is the responsibility of the person submitting the code and it should be judge on its quality, not how it was made.
That is very True that as long as it works then does not matter where it has come from
Problem with AI I think people hate that it’s to open to Mistakes and has broken other Software and can take awhile to Fix
Really is a personal choice
I agree. If I need to hammer in a nail, I don’t care whether the hammer was handcrafted by someone over two days or produced by a machine in a factory in ten minutes. What matters is that it does the job. The faster, the better. However, the generative use of AI for artistic creations is a completely different matter, but that’s not what this post is about. RetroArch is still the best gaming ecosystem in existence today.
There are legitimate reasons why it matters from where the code comes from. First the code from an ai agent is often less readable, in some cases even completely unreadable, by humans. This is very important. Another reason why it matters is, we don’t know if there will be licensing issues in the future. Because right now its the wild west and it is possible that code from Ai has to be public domain, and cannot be licensed. Now, if that will become true, we don’t know yet, but that’s the issue; we don’t know. It is also a bit unethical, because it is trained on data and other software without their consent. There are debates to if this is even okay or not, I just list the reasons.
If the programmer would do most work and use Ai as a support, it is completely different story then letting it vibe code on its own. I don’t know to what degree this is on RetroArch. Have in mind, I use RetroArch (and still love it) and I have local Ai models installed on my system to experiment. So I’m not 100% against Ai. I just made the comment to explain why some people (and to some extend me too) have a few issues where the code comes from.
Human code can also be unreadable, even more so than AI. I work with legacy systems as part of my job and some of these codebases that I interact with make me want to rip my eyes out.
And you are basically saying what I said. It is up to the person submitting the code, they are responsible. If they just let the agent go wild without any oversight or understanding what it is doing, that’s not the AI’s fault, that’s the programmer’s fault.
it’s just a tool at the end of the day.
I guess my issue with this post is, it feels like people looking for drama. Have you had issue with Retroach? has anything broke? If you do not like AI for whatever personal reasons and you no longer want to use the application, that’s fine, but to start dooming over the mere use of it without any actual signs of regression is just ridiculous. You might as well not use windows or linux, or pretty much any OS at this point because they all each have AI assisted code in them.
I don’t think this post or any critique over Ai is about searching drama or looking for doom and gloom (however, I always look for DOOM). People are concerned, so its their right to discuss, question and getting informed. I don’t see an troll attempt anywhere, just curious discussion and sharing experience and opinion.
It is also not like that there weren’t issues, but it is hard to know exactly if this would have happened without Ai tools. One form of regression is also that code quality gets worse over time and basically the Ai company is programming the tool and have power over it, more than their actual contributors. At least in extreme cases, I don’t think (hope) this is the case in RetroArch. I’m talking in general.
Look at other projects that are heavily created with Ai tools. They are often broken. Not saying RetroArch is, but that’s one reason why people are worried. Right now RetroArch works pretty good and is probably my favorite application (or suite of applications, however you want say it) on my entire computer. I want to emphasize I’m not hating on RetroArch, just expressing my concern and other public concerns about heavily using Ai. And I brought up a few points before too, so it’s not completely unreasonable.
I understand what you are saying and I do not mean to make my post seem aggressive. I just find the entire discourse a little tiring. You are correct that there are many projects that are subpar, but i don’t blame AI for that, I blame the developers that do not know what they are doing.
To put it another way, until proven otherwise, I trust that the RA devs knows what they are doing and will continue to deliver a good application. If the entire thing catches on fire a year from now, my reaction wont be “AI ruined it”, it will be “Incompetent developers ruined it”.
Agreed and my stance is similar; I trust the RetroArch developers, otherwise I would have stopped using it. And its never to blame Ai, off course the user of the Ai is to blame. I don’t think anyone here is blaming Ai anyway, but are afraid that the developer changes to someone who starts taking shortcuts and gets worse.
I mean similar to someone doing stupid stuff when drunken, you don’t blame alcohol, but the “user” instead. Yet we have rules and laws around alcohol, and we are afraid of people being drunken.
There were a couple of decades of debates whether paged memory is a good compromise and had it’s determined opposers on the OS scene.Today it’s mainstream, integrated in relevant environments.
The downsides of AI coding contribution also exist, but it doesn’t come in the way that an experienced coder with years of environmental experience is using it. It’s rather that someone has an idea and five minutes of time and then the former has to clean up the incompatibilities, regressions etc.