I feel really sorry for this person. This is purely my own view of the future, and I am no fortune teller. But the reality he is desperately clinging to is just gone. AI in software development is...
I feel really sorry for this person. This is purely my own view of the future, and I am no fortune teller. But the reality he is desperately clinging to is just gone.
AI in software development is here to stay. Denying this is genuine mental denial. I wont say that non-AI code will disappear, and hands-off vibecoding is still genuinely reckless, but I believe the amount of AI generated code is certainly already outpacing humans, and this is just the beginning.
As I mentioned in a cousin post: AI software development is not inevitable, it is not mandatory to accept it. If a project or community doesn't want it, the AI coders should shut up, fork, and...
As I mentioned in a cousin post:
AI software development is not inevitable, it is not mandatory to accept it. If a project or community doesn't want it, the AI coders should shut up, fork, and form their own community instead of trying to force their values into the community that is rejecting theirs.
I personally find a disclosure policy a good middle ground, allowing for that flexibility while still fostering a default stance of being human-first.
You are reversing the roles in the scenario. I also believe non-AI code will still exist. I mentioned as much above. It will however must likely become niche. Not worse, maybe better, but niche....
You are reversing the roles in the scenario.
I also believe non-AI code will still exist. I mentioned as much above. It will however must likely become niche. Not worse, maybe better, but niche.
Like an artisinaly crafted closet versus an ikea wardrobe. There are certainly more of the latter in existence, and the quality difference in the example is more profound in the furniture example than in the coding one.
In this case, the owners of the project decided to loosen their AI restrictions, and this contributor disagrees. Like you mention. He is free to fork it, "shut up about it" and run his own project.
But I explicitly wanted to frame this not as a pragmatic "get with the times" post, but as a genuine emphatic response to seeing someone lose something, a project, a community, that they were pasionate about. I genuinely feel the dread that goes with such a loss, and even if I believe it is mostly inevitable for the majority of software development, I think we should at least pause and aknowledge the human impact it has on those involved.
The issue is that you can't prove that something wasn't AI generated. Many have tried, and the results are not great. All anyone has to do is prompt the model some verbose version of 'don't sound...
The issue is that you can't prove that something wasn't AI generated. Many have tried, and the results are not great. All anyone has to do is prompt the model some verbose version of 'don't sound like an AI'
Honestly, you don't need to "prove" AI. You need to prove the PR author knows what's going on. If they can't defend their own code, it won't hold up to scrutiny in the main branch. That's all part...
Honestly, you don't need to "prove" AI. You need to prove the PR author knows what's going on. If they can't defend their own code, it won't hold up to scrutiny in the main branch. That's all part of a proper PR pre-AI.
But based on the author, moderation is thin as is, and there's rarely enough maintainers on such FOSS projects as is. I imagine this is easily subject to DARVO tactics on how "the Bevy community isn't welcoming", when it's simply the standard way FOSS projects have been run for decades.
Yes. But there will be space for them to co-exist, and there is no reason to try to force acceptance where it is not wanted. Which is the biggest problem a lot of projects and communities are facing.
Yes. But there will be space for them to co-exist, and there is no reason to try to force acceptance where it is not wanted. Which is the biggest problem a lot of projects and communities are facing.
I'm not familiar with Bevy or it's community, but it sounds like perhaps this is actually the inverse case? The above goes both ways.
If a project or community doesn't want it, the AI coders should shut up, fork, and form their own community instead of trying to force their values into the community that is rejecting theirs.
I'm not familiar with Bevy or it's community, but it sounds like perhaps this is actually the inverse case? The above goes both ways.
I don't disagree, but given the initial AI policy of 'hard ban' I think that is not the case. My read is that the culture has shifted, and that making life easier on the maintaners was the primary...
I don't disagree, but given the initial AI policy of 'hard ban' I think that is not the case.
My read is that the culture has shifted, and that making life easier on the maintaners was the primary goal. The author may well have lost their space already, but I think that is a worthy and justified bemoaning if the pro-AI crowd is newer and displaced the critics entirely.
Are cultures not allowed to shift or people change their minds? I think the nazi bar analogy is a bad one for obvious reasons. Cultures of bars shift all the time as people age out, kids...
Are cultures not allowed to shift or people change their minds?
I think the nazi bar analogy is a bad one for obvious reasons. Cultures of bars shift all the time as people age out, kids re-inventing underused watering holes is almost a national pastime here. Not every shift in culture is bad as in nazi bad.
They do. However, Tildes serves as the perfect example: The floodgates are intentionally closed to prevent this exact sort of seismic shift. Broadly speaking, we have an extremely low (if not 0)...
They do. However, Tildes serves as the perfect example: The floodgates are intentionally closed to prevent this exact sort of seismic shift.
Broadly speaking, we have an extremely low (if not 0) tolerance of bigotry here. The old guard does want that to remain, if elements of that trickle in, our best course of action is to stomp them out. And I'd like to think we do a better than average job of it.
And in this case, the metaphor is apt (again, presuming that OG AI policy was reflective of the community at that time); An undesired cultural shift crept in and allowed to foster.
The point of the Nazi bar metaphor is not that it's Nazi bad. It's that if there is a community goal, ousting/rejecting the attitude of newcomers who oppose that goal, no matter how trivial, is the appropriate response. Hence why men should be careful when chatting in women's spaces.
And a culture of a bar shifting is fine....if the bar owner agrees. But if the bar owner doesn't want the biker bar to become a dance club, he's not wrong for ousting people abusing the jukebox and complaining there isn't enough room to dance.
Might the bar go under because their culture is dying? Sure. But that's for them to decide.
Open spaces will always suffer from this potential flooding and destruction of the culture. And thus, it's probably prudent for spaces to close off and oust if a cultural shift happens too quickly.
It is reasonable to give a small window of forgiveness, but yea. There is a reason I stopped frequenting ~lgbt, and that's because I had a bad case of saying dumb things (even without ill...
It is reasonable to give a small window of forgiveness, but yea.
There is a reason I stopped frequenting ~lgbt, and that's because I had a bad case of saying dumb things (even without ill intentions) that hurt feelings and aren't terribly relevant in any reasonable public discourse. That even if there were tiny droppings of reasonable discussion to be had within a very narrow bounds, they did little beyond open the door for right wing talking points and pollute a space that really wasn't for that. That's probably why I'm at least more than a little blind to the current state of bigotry here (beyond typical straight white male privilege blinders).
Shame there are a few others that do not follow my example and do some combination of learn, shut up, or leave.
I agree people need to spend some time educating others. Not just LGBT stuff, but anything where someone misbehaves. People are so ready to just shut someone out. Give people a chance to improve,...
I agree people need to spend some time educating others. Not just LGBT stuff, but anything where someone misbehaves. People are so ready to just shut someone out. Give people a chance to improve, you might just get surprised when they do. There is an art to it. You have to keep a cool head and find a way to exploit their goodness against them. Twist their arm a bit and force them to look in the mirror.
My favorite thing to do on PA Nextdoor is to take Trump headlines and replace with Shapiro with a footer explaining what I did, and watch people foam at the mouth about Democrats being corrupt...
My favorite thing to do on PA Nextdoor is to take Trump headlines and replace with Shapiro with a footer explaining what I did, and watch people foam at the mouth about Democrats being corrupt because they didn't read the last line.
And then when you point this out, you either get no response, or a level of incoherent mental gymnastics that I didn't think a human brain was capable of performing.
People need shared physical spaces where they can gain each other’s trust and discuss politics. Honestly simulating such an environment in high school would be well worth allocating 20% of the...
People need shared physical spaces where they can gain each other’s trust and discuss politics. Honestly simulating such an environment in high school would be well worth allocating 20% of the students’ time to. I’ve seen plenty of Democrats fluster in the exact same way (not as often by my recollection, but we’re all human and fallible in the same ways regardless of who we vote for).
I mean it helps that usually Democrat bad behavior is not quite as blatantly hypocritical. It happens. But rarely as bad as a $5,000 check if you vote for my team.
I mean it helps that usually Democrat bad behavior is not quite as blatantly hypocritical. It happens. But rarely as bad as a $5,000 check if you vote for my team.
AI is here to stay the same way Reddit is here to stay. Maybe so, but that doesn't mean I need to use Reddit to function. Likewise, I imagine for someone talking as a volunteer contributor to a...
AI is here to stay the same way Reddit is here to stay. Maybe so, but that doesn't mean I need to use Reddit to function.
Likewise, I imagine for someone talking as a volunteer contributor to a large open source project that the author has every bit of choice in how to spend their time in such a community.
But a specialized game engine is the exact kind of niche I expect such supposed "artisan code" to remain valued. As the author mentioned, this is not just a tool programmers use. It's used to drive creatives, and creatives cannot simply function off of "well it works". I can point to plenty of games on Steam with such a descriptor.
I suppose things may be different professionally when it comes to being transformed into a duty of Slop Doula on the company dime. I still have rent to pay and will do what I need to do. But know I'll spend my off time preparing for a future where I am not reliant on such work to survive.
Can't remember who wrote it, but I agree with them. The basic gist of it was that plugins will become the norm in applications. It's risky to accept AI-generated code into projects and integrate...
Can't remember who wrote it, but I agree with them. The basic gist of it was that plugins will become the norm in applications.
It's risky to accept AI-generated code into projects and integrate them to the core. What more and more applications need to do is have a robust plugin system. Then people can AI-slop/engineer/author as many plugins as they want with as much oversight as they're comfortable with.
This way the core application can stay artisanally coded if that's what the community and/or devs want and people can add features without disrupting it - or needing Yet Another Fork.
I agree, and I think this is going to be one of the big generational divides going forward. It'll be very interesting to see how the 'AI-natives' turn out... I really hope that most of them can...
I agree, and I think this is going to be one of the big generational divides going forward. It'll be very interesting to see how the 'AI-natives' turn out... I really hope that most of them can find interesting and novel ways to make use of AI instead of just letting their skills atrophy (or never develop at all). I think there's all kinds of potential for them if they can reach it.
The author makes their disgust with AI very obvious, but disgust is not a good reason for prejudice. Apparently they want a "safe space" for their anti-AI prejudices? They object to a "no dunking"...
The author makes their disgust with AI very obvious, but disgust is not a good reason for prejudice.
Apparently they want a "safe space" for their anti-AI prejudices? They object to a "no dunking" rule that would prevent people from freely expressing their anti-AI prejudices using memes?
I wouldn't want to be in that kind of community.
I expect that it's only going to get worse though. Anti-AI populism is on the rise and uncritical boosters don't help either. They're going to feed each other.
Maybe we will need safe spaces for reasonable people who prefer evidence-based discussions?
That's not really what I took away from this at all. And your response does feel somewhat like you're validating this line: I think there a few key points: LLM-coded projects are not exactly great...
That's not really what I took away from this at all. And your response does feel somewhat like you're validating this line:
Meanwhile, almost all arguments against "AI" use were more or less thrown out or ignored.
I think there a few key points:
LLM-coded projects are not exactly great for the purposes of sharing and learning from each other. If the LLM spits something out, and a human asks me a question about it, I now basically have to answer that question for myself first. Handcoding it myself means I'm answering from myself. And I can get the frustration if the goal of a space was sharing lessons learned on the journey and not 'look at the end product.' I can't really speak to whteher that space really is/was that, but overall that sentiment is reasonable IMO.
Having done it for several months now, LLM coding is largely orthagonal to learning, especially when coming in to unknown territory. You don't really learn much until the LLM screws up. Then you're mostly bughunting rather than writing.
A no dunking rule does invite the Nazi Bar problem, particularly in an open space. If Bevey was an anti-AI (or at least AI-skeptical) community, placing the newer AI boosters on level ground with the old community is not a neutral move; it's practically an invitation to displace the old culture.
The bullet list at the end is the typical laundry list of a community getting hit with an Eternal September.
For my own views that echo the author's:
I think it's reasonable to allow AI-assisted contributions with a disclosure policy, while also allowing for Tildes-style "We're rejecting this because we decided to" PR policy. As well as not necessarily further welcoming AI boosrers into a space that previously rejected them.
In this era when any project can easily become flooded, project owners need to be given the grace to straight up reject anything without writing up a massive well-reasoned rejection note.
AI use is not inevitable. If a given community doesn't want it, and they ban it, then maybe the AI users should respect that, fork, and form their own community instead of trying to force their stance on the existing community. If their AI methods truely are superior, then they'll leave the old projects in the dust and both communities can live in peace apart from each other.
Reconciling this is more important on giant projects like the Linux kernel. Less so on a boutique project that I'm hearing about for the first time.
While we're here, what are your thoughts on the kernel? I can't decide, cause obviously people vibecoding in there would be terrible, but at the same time it lowers the entry level knowledge of...
While we're here, what are your thoughts on the kernel?
I can't decide, cause obviously people vibecoding in there would be terrible, but at the same time it lowers the entry level knowledge of contributing, which might end up being a good thing for the project as a whole.
I know it'll be an eternal September kinda thing if that happens and so the maintainers who really only had to converse amongst themselves will suddenly find an overwhelming amount of PRs that are not up to standard, but, if they can set up a system to auto reject PRs based on some failed tests or something there might be some gems in there fixing like 20 year old security bugs.
I think the resolution for the kernel is 'quality is quality' and accountability. A disclosure policy, mandates that it must be tracable to a real person. And that if a PR fails a vibe check, the...
I think the resolution for the kernel is 'quality is quality' and accountability. A disclosure policy, mandates that it must be tracable to a real person. And that if a PR fails a vibe check, the onus is on the submitter to pass that vibe check and not the maintainer. IIRC even before the rise of vibecoding the kernel had to push away a lot of sub-par cruft. CVEs that are not CVEs, that sort of thing.
The path curl took seems a prudent one: Close the monetary floodgates (reducing the tide from people throwing mud at a wall to see what sticks), do a hard ban, then ease up as maintainer ability to sustain improves, and still keep a high minimum bar to keep low-effort out.
I watched a ton of projects stop giving monetary incentives, that was super sad. I'm sure there were real people out there subsisting on those. Was among the first casualties for sure.
I watched a ton of projects stop giving monetary incentives, that was super sad. I'm sure there were real people out there subsisting on those. Was among the first casualties for sure.
Really sucks, but when your bounties end up making the job harder than easier, then it's clear what needs to be done. Now if only we could do that in politics... But yeah, some super specialized...
Really sucks, but when your bounties end up making the job harder than easier, then it's clear what needs to be done. Now if only we could do that in politics...
But yeah, some super specialized aspects of life need to be gatekept. It's one thing to welcoke perspective new scientists and another to be accommodating to throwing a bunch of garbage into a nuclear reactor. You got to start in safer environments and build up. I really wish such apprenticeship lanes existed to create the next generation of kernel developers, but that's a whole other can of worms.
I frequently go to live drawing sessions, cameras are forbidden in this spaces out of respect to the model. Nobody goes crazy about it, claiming that people are living in the past and that there's...
I frequently go to live drawing sessions, cameras are forbidden in this spaces out of respect to the model. Nobody goes crazy about it, claiming that people are living in the past and that there's no longer value in drawing with a pencil when a camera can do it faster. At the same time, I can guarantee that most people on those session also use photographs to study the human body on their own. A good artists understands that drawing from life and drawing from a reference are completely different experiences. A camera will flatten the colors and values not matter how good it is, and observing something that is alive and moving gives you a more completed understanding on how it is constructed. At the same time, a good artist understands that is not necessary to do everything from scratch and that the right tool will save you a lot of time of money.
Whenever a conversation like this comes out, where people want their own space, others are quick to say that "they are trying to hold on the past, that world no longer exists". I wonder if this is unique to software. Do people tell others that there's no longer value in learning to draw, learning how an engine works and built your own model, or learning how to create a mechanical computer from scratch?
People understand that things are different in the professional world and the hobbyist communities. Why can't software be the same? There's still value in struggling, learning things on your own and talking with other people, sharing ideas. Otherwise why are you in tildes and not talking to an LLM?
Wanting a space, free of LLMs should be an uncontroversial opinion, the same way that wanting a space free of smart phones should be. I understand that this is a game engine that can be used in real games, not just hobbyist projects, but I still don't see the point on dismissing a person that wants to create a space only for humans. If someone wants to create that space, the only thing there's to do is respect their decision.
That's kind of the opposite of the situation, no? The context is that Bevy, an open source game engine that the author of the blogpost does not run, has decided to loosen its PR requirements to...
I frequently go to live drawing sessions, cameras are forbidden in this spaces out of respect to the model. Nobody goes crazy about it, claiming that people are living in the past and that there's no longer value in drawing with a pencil when a camera can do it faster.
That's kind of the opposite of the situation, no? The context is that Bevy, an open source game engine that the author of the blogpost does not run, has decided to loosen its PR requirements to allow more LLM generated code.
The author of the blog, then, is "going crazy" about it.
I mean, the reality is that Bevy's maintainers are trying to make a thing that serves a role. It would be like if you went to a portrait photographer's studio and demanded that they cease taking photographers, cameras are the work of the devil, and oil paintings have much more character.
I think the latter opinion, that oil paintings of portraits have more character, is a justifiable subjective opinion to have, but like, it's not your photography studio. You're more than free to make your own studio that only does graphite portraits of live models. But you don't really have a say over everyone else.
If the author of the blog wants to make their own game engine that has strict anti-LLMs contribution requirements, by all means. But Bevy isn't there's, in the end.
You're right. I just meant that somewhere in the post, the author mentions she's creating a new community for Bevy. I was responding more to the idea that a community banning LLM use would have no...
You're right. I just meant that somewhere in the post, the author mentions she's creating a new community for Bevy. I was responding more to the idea that a community banning LLM use would have no real value beyond becoming an anti-AI circlejerk.
That depends on if you think the SMEs or the maintainers out in the field truly make up a community. I suppose to use the live drawing analogy, it would be "in their right" for organizers to...
That's kind of the opposite of the situation, no?
That depends on if you think the SMEs or the maintainers out in the field truly make up a community.
I suppose to use the live drawing analogy, it would be "in their right" for organizers to declare that suddenly photography of the nude models is okay. It would also be in the right of models to no longer want to participate in that community. I wouldn't call a model prudish in this day and age for not wanting the risk of their body being photogtaphed and posted online, even with arguments like "well what if someone is an amazing painter" or "well we just ban people who post pictures online".
I see parallels here. Ultimately these are decisions made that do not support the community, and it may end up making a worse feeling community, stripping the culture Bevy and Rust is known for.
I mean, the reality is that Bevy's maintainers are trying to make a thing that serves a role.
That's the big issue here, actually. The paid maintainers do not necessarily have the same incentives as the unpaid. The paid ones need to ship and champion the product. The unpaid ones want to contribute to a greater collective and likely came here to escape that paid incentive that had them shortcut and compromise their way into keeping paychecks coming.
In particular, the paid maintainers do not care if community crumbles. At least, not until it starts affecting engije adoption. But until then, it's not part of their job to foster good vibes.
Sorry to dig into your take, but it already seems like the rest of the world is the safe space for people who want to discuss how LLMs/generative AI can be deployed for use in day to day life? The...
Sorry to dig into your take, but it already seems like the rest of the world is the safe space for people who want to discuss how LLMs/generative AI can be deployed for use in day to day life? The idea that there's widespread negative perception is largely an effect of the anglosphere bubble that many of us live in -- especially for those in the tech sector (which foolishly for decades decided that unionization was for dumb losers and continuously refused to organize >:| zero sympathy there from me tbh). The article author's take that the average person isn't accepting of this stuff is not reflective of reality -- the average American, sure, but there's more than just one country in the world that can buy things.
And anti-AI populism seems economically self defeating overall, so it's probably not something to be concerned with directly? I'd be moreso worried about it spiraling off and turning into another antivax -> Q-anon pipeline, but I don't know if a politically compatible conspiracy theory group exists. If all they do is burn down local data centres and protest in discord, it's not going to be a practical concern for anyone else.
IMO as I'd noted in another thread, the main impact from this will be excluding otherwise very motivated voices from discussions on how to realistically address the fallout from AI tooling existing. That seems like an objectively bad thing, but if people want to opt out, I don't see a reason not to support them in doing so?
I don't think a "safe space" is the default. When it's ambiguous what the rules are, you will get the more extreme pro- and anti- types getting into arguments with each other, resulting in...
I don't think a "safe space" is the default. When it's ambiguous what the rules are, you will get the more extreme pro- and anti- types getting into arguments with each other, resulting in increased polarization.
For example, recently on Bluesky a bunch of anti-AI types were dogpiling an open source developer who put out a new, experimental Linux port, because they used AI to do it. Someone from the Bluesky company had to say that this isn't acceptable.
Bluesky is not a space that has any official policy about AI being good or bad. They did add a way to label posts as AI generated. Also, there is an experimental Bluesky product that uses AI, so clearly there are people who work there who are into it.
I think it ought to be possible to have a "big tent" policy where both pro- and anti- AI opinions are acceptable? We are still figuring this stuff out.
I'm particularly aggravated by the folks who assume that being anti-AI is the only ethical position, that people who agree with them are ethical and people who disagree must be unethical. To me, that’s fighting dirty. Another way of fighting dirty would be assuming that there’s an anti-AI consensus in a community when actually, there is no consensus because they never really talked about it. I don’t know of that’s the case for the Bevy community, but it sort of sounds like it?
But I don't want to let that polarize me, because there are bad takes on both sides, and it doesn't mean people with more nuanced positions don't exist.
Its an interesting chart, but I think this interpretation is too focused on the "concerned over excited" and not enough on "excited over concerned". There is indeed a concern that correlates to...
The idea that there's widespread negative perception is largely an effect of the anglosphere bubble that many of us live in
Its an interesting chart, but I think this interpretation is too focused on the "concerned over excited" and not enough on "excited over concerned". There is indeed a concern that correlates to "how much danager do I feel my futute is in from AI". But I find it equally interesting that the "excited over concerned" is staggeringly consistent distribution: 10-20%, with Israel being to only real outlier with 29%. A much much smaller spread than 16-50% on concern.
It seems overall, even the most "excited" countries are ultimately still in a "wait-and-see" mentality.
To use the author's words: To use the obvious analog to Tildes: I'm sure it's hard to "definitively prove" Reddit has gotten worse over the years in all but a few extremely obvious cases for a few...
To use the author's words:
You can't moderate away how people feel about a topic, only what they post in official channels. People feel worse about sharing their work as it's sandwiched between low-effort, low-engagement-with-bevy work. They feel worse about looking in showcase because there is very little high-effort work or social engagement.
To use the obvious analog to Tildes: I'm sure it's hard to "definitively prove" Reddit has gotten worse over the years in all but a few extremely obvious cases for a few truly abandoned communities. But despite that, many of us did come seeking this community because we knew the vibes were off and Reddit no longer offered whatever we initially came there for. Maybe some of us use Tildes in tandem with Reddit, maybe some of us completely left the Reddit bar. But there's clearly reasons beyond objective facts that brought us here.
I can see how people spamming an open source project with low-effort demos could be a problem. And also, spamming the forums with low-effort memes would also be a problem. Aren't both of those...
I can see how people spamming an open source project with low-effort demos could be a problem. And also, spamming the forums with low-effort memes would also be a problem. Aren't both of those solved with better moderation? In neither case is moderation about trying to change how people feel. Instead it's intended to get across what sort of behavior is ok in that forum.
Yes, they are. I don't think passive agreseive memes have ever solved any conflict, and even on my pseudanonymous accounts I try to avoid such low effort swipes (let alone my real name which would...
Aren't both of those solved with better moderation?
Yes, they are. I don't think passive agreseive memes have ever solved any conflict, and even on my pseudanonymous accounts I try to avoid such low effort swipes (let alone my real name which would be how I enter a community like Bevy). But I guess some people don't have such restraint.
I do feel a bit iffy about banning emoji reactions, though. It's a core feature of the platform and if people can't express themselves that way, we end up with content much harder/time consuming to moderate. It's like every time you try to ban downvotes in Reddit; it never ends well, even with good intentions.
Instead it's intended to get across what sort of behavior is ok in that forum.
Moderation can be passive aggressive too. At some point, a bar needs to determine if it is a nazi bar or not. Trying to dance around the issue just leads to conflict. What I see here with little actions like banning reactions is "we want to welcome AI generated projects, but we don't want to say it out loud". Which is cowardly no matter your stance. At some point, the community will answer that question for you.
An observation of how a change in AI policy has affected the Bevy game engine community in the months since a new AI policy came into effect. As can be inferred from the title, the author, a...
An observation of how a change in AI policy has affected the Bevy game engine community in the months since a new AI policy came into effect.
As can be inferred from the title, the author, a long-time Bevy contributor, has had a very negative experience with the changes in the community. Their post describes the new policy's effects on the behavior and health of the development community, and how the changes in the community have directly decreased the author's willingness to continue participating in the Bevy project.
It seems like we've got two groups of insecure people fighting each other online and they make everything worse for the rest of us: The LLM-assisted idiot who thinks they can now finally do things...
It seems like we've got two groups of insecure people fighting each other online and they make everything worse for the rest of us:
The LLM-assisted idiot who thinks they can now finally do things they couldn't before. They always felt a bit lesser for having a programming language between them and their goals. Trying to attack LLMs or the work they produce with them makes them lash out. This type of person shouldn't be tolerated.
The LLM-avoidant graybeard that has not figured out how a new and versatile tool could help them in any capacity. They feel insecure because something that is a source of income, a source of pride, and a source of enjoyment is flipping upside down while they're standing on it. Avoiding self-education on LLM use leaves them with less information - it's strictly a disadvantage.
I work on a semi-professional project together with LLM-using and LLM-avoiding developers alike. We never fight about it. Everyone is doing their best work using a different set of tools. It is in fact possible to co-exist while using different tech stacks. This kind of harmony is not something the internet has been good at achieving... ever. But I suppose in the past you'd mostly stick to your Gentoo forum and the Debian people stuck to theirs. Now you have every forum splitting in a novel way.
Edit: My current philosophy regarding LLMs and software development is that such a tremendous change to the industry provides opportunity to figure out new avenues to wealth and success as much as they fuck up the old avenues. If you’re a smart person who was able to figure out software engineering you’re also smart enough to be one of the first in whatever new frontiers emerge. Until computers obsolete the human mind the smarter humans should be able to use chaotic changes to great self benefit.
I work in games, so my attitude on a personal perspective is "landmine; avoid until defused". Just being speulated that you're using AI in your game is OR suicide, and the potential benefits seem...
I work in games, so my attitude on a personal perspective is "landmine; avoid until defused".
Just being speulated that you're using AI in your game is OR suicide, and the potential benefits seem minimal anyway. So there's no point not simply staying course and keeping an eye out. My usage is pretty much limited to "ive googled this for 30 minutes with no good answer, let's see if Claude can give me new avenues to google". (P.S. I will never forgive Google for sabatoging its own search just to be more profitable than Sin)
Besides, there's already so many moving parts you have to manage in game development as is. Being asked to retool in yet another workflow that maybe potentially can kind of speed up development is a story you hear every other week, even pre-AI. I don't want to subscribe or worry about token usage for frontier models and trying to setup an AI workflow with open weight models (which I tried for a non-gaming experiment) was taking days as is when I could have simply plumbed it myself. I'm already vulnerable to automation hypnosis, so I don't want this to be yet another perpetrator.
(Use Kagi for search, no AI easy controls to filter crap you don't want) I work in mobile games and I have first hand knowledge of household name mobile game production. They use AI for...
(Use Kagi for search, no AI easy controls to filter crap you don't want)
I work in mobile games and I have first hand knowledge of household name mobile game production. They use AI for everything. Most used AI and ML years before ChatGPT was a thing, both for data and art. They had (and still have) custom systems for generating seasonal content variants, backgrounds etc. This gives artists more time to focus on "hero art" instead of drawing trees and rocks and castles.
Only PC gamers care even a little about AI in their games and IMO most of it is just baseless hysteria. Like Expedition 33 having one background texture AI-generated causing a massive rage storm.
I'm no game developer. I'm not a developer of any sort, but as a user, if I find that an artistic project I was interested in was made with AI, I'm not going to partake in said project. I won't...
I'm no game developer. I'm not a developer of any sort, but as a user, if I find that an artistic project I was interested in was made with AI, I'm not going to partake in said project. I won't play games made with AI, I won't read AI books, and I won't use an AI recipe. I do my best to avoid LLMs as much as possible because I just don't want them in my life. I get that the author is upset, and I understand why as someone who is just so tired of having AI/LLM use shoved down my throat every day for literally everything I do, when I just want human-created art. I want humans to code and do the art for games, I want humans to write books, and I want humans to create things. Especially because this use is so harsh and dangerous to our environment, and climate change is a massive issue that we just aren't addressing, but it is making worse with every new data center that is deployed.
And I think you bring up a good point the author touched on: If you are casting a wide net, courting artists to be a part of your community, you need to tread very carefully with your attitudes...
And I think you bring up a good point the author touched on: If you are casting a wide net, courting artists to be a part of your community, you need to tread very carefully with your attitudes toward AI contributions.
A lot of artists aren't going to take well to making some contributions, then have somebody take that hard effort and say 'well I took this and used it as a foundation to have the AI generate an entire asset library in that style.'
At the risk of breaking the "no dunking" rule, I think this entire post comes from a very bad faith position. That doesn't mean the position isn't valid, I think Fallible isn't entirely incorrect...
At the risk of breaking the "no dunking" rule, I think this entire post comes from a very bad faith position. That doesn't mean the position isn't valid, I think Fallible isn't entirely incorrect or out to lunch here, but what I will say is that they clearly are tacitly aware they're standing in moving water. Whether you like it or not, and I'll personally generally err on the side of "not", this isn't going away. If you're on the side that doesn't want anything to do with the topic of AI-assisted coding, and this hasn't been a point of contention before because it, well, didn't exist, then I can see how you feel a community that you used to love slowly filled up with the worst kind of vocal assholes. This is nothing new, and I do feel sorry for them, and I've seen it before in different communities with different issues.
I just don't really see, on this particular issue, how you're meant to go forward. For what it's worth, I think the AI policy that Bevy has currently is very measured and valid, and drawing comparisons to the Nazi bar problem are weirdly pointed and unhelpful. But, and I say this without trying to be unhelpful, you can't make everyone happy.
I'm willing to believe that, despite the assertion that the technology is inextricable from its negative aspects (which I don't believe, by the way), there's some pretty significant or at least important parts of Bevy that are wholly written by AI. I have no reason to believe this, true, but I also have no reason to doubt it. This just happens now. It's table stakes, see my comment on the platform Cara being scraped for more details. And maybe it's bad, maybe it's good, sure, but it doesn't change the reality that it's only going to get worse, or better, depending on what side of the aisle you're on.
What I can not understand is either side making a weirdly emotional argument over the state of any given project. Last time I checked, Bevy is open source, and it can be forked. If you'd like to do that, you can, and again, I don't say this to dunk on anyone. I've forked projects before because they didn't go the way I wanted them, and I will continue to do so, because sometimes getting the thing done how I want it to be done is easier than fighting with a community who doesn't see it your way.
This is not me saying "if you don't like it, you can leave", it's me saying "sometimes it's better to curate your experience yourself". I mean, they clearly said that they don't much enjoy the discussions or the community how they used to. Seems like holding on will only make it worse, something readers of my contributions on here will know I personally had to learn myself.
I feel really sorry for this person. This is purely my own view of the future, and I am no fortune teller. But the reality he is desperately clinging to is just gone.
AI in software development is here to stay. Denying this is genuine mental denial. I wont say that non-AI code will disappear, and hands-off vibecoding is still genuinely reckless, but I believe the amount of AI generated code is certainly already outpacing humans, and this is just the beginning.
As I mentioned in a cousin post:
AI software development is not inevitable, it is not mandatory to accept it. If a project or community doesn't want it, the AI coders should shut up, fork, and form their own community instead of trying to force their values into the community that is rejecting theirs.
I personally find a disclosure policy a good middle ground, allowing for that flexibility while still fostering a default stance of being human-first.
You are reversing the roles in the scenario.
I also believe non-AI code will still exist. I mentioned as much above. It will however must likely become niche. Not worse, maybe better, but niche.
Like an artisinaly crafted closet versus an ikea wardrobe. There are certainly more of the latter in existence, and the quality difference in the example is more profound in the furniture example than in the coding one.
In this case, the owners of the project decided to loosen their AI restrictions, and this contributor disagrees. Like you mention. He is free to fork it, "shut up about it" and run his own project.
But I explicitly wanted to frame this not as a pragmatic "get with the times" post, but as a genuine emphatic response to seeing someone lose something, a project, a community, that they were pasionate about. I genuinely feel the dread that goes with such a loss, and even if I believe it is mostly inevitable for the majority of software development, I think we should at least pause and aknowledge the human impact it has on those involved.
The issue is that you can't prove that something wasn't AI generated. Many have tried, and the results are not great. All anyone has to do is prompt the model some verbose version of 'don't sound like an AI'
Honestly, you don't need to "prove" AI. You need to prove the PR author knows what's going on. If they can't defend their own code, it won't hold up to scrutiny in the main branch. That's all part of a proper PR pre-AI.
But based on the author, moderation is thin as is, and there's rarely enough maintainers on such FOSS projects as is. I imagine this is easily subject to DARVO tactics on how "the Bevy community isn't welcoming", when it's simply the standard way FOSS projects have been run for decades.
Sounds like it would be the other way around if the author thinks the admins are compromised.
In a project, not per se, in general, it is inevitable
Yes. But there will be space for them to co-exist, and there is no reason to try to force acceptance where it is not wanted. Which is the biggest problem a lot of projects and communities are facing.
I'm not familiar with Bevy or it's community, but it sounds like perhaps this is actually the inverse case? The above goes both ways.
I don't disagree, but given the initial AI policy of 'hard ban' I think that is not the case.
My read is that the culture has shifted, and that making life easier on the maintaners was the primary goal. The author may well have lost their space already, but I think that is a worthy and justified bemoaning if the pro-AI crowd is newer and displaced the critics entirely.
Are cultures not allowed to shift or people change their minds?
I think the nazi bar analogy is a bad one for obvious reasons. Cultures of bars shift all the time as people age out, kids re-inventing underused watering holes is almost a national pastime here. Not every shift in culture is bad as in nazi bad.
They do. However, Tildes serves as the perfect example: The floodgates are intentionally closed to prevent this exact sort of seismic shift.
Broadly speaking, we have an extremely low (if not 0) tolerance of bigotry here. The old guard does want that to remain, if elements of that trickle in, our best course of action is to stomp them out. And I'd like to think we do a better than average job of it.
And in this case, the metaphor is apt (again, presuming that OG AI policy was reflective of the community at that time); An undesired cultural shift crept in and allowed to foster.
The point of the Nazi bar metaphor is not that it's Nazi bad. It's that if there is a community goal, ousting/rejecting the attitude of newcomers who oppose that goal, no matter how trivial, is the appropriate response. Hence why men should be careful when chatting in women's spaces.
And a culture of a bar shifting is fine....if the bar owner agrees. But if the bar owner doesn't want the biker bar to become a dance club, he's not wrong for ousting people abusing the jukebox and complaining there isn't enough room to dance.
Might the bar go under because their culture is dying? Sure. But that's for them to decide.
Open spaces will always suffer from this potential flooding and destruction of the culture. And thus, it's probably prudent for spaces to close off and oust if a cultural shift happens too quickly.
It is in fact not 0...
It is reasonable to give a small window of forgiveness, but yea.
There is a reason I stopped frequenting ~lgbt, and that's because I had a bad case of saying dumb things (even without ill intentions) that hurt feelings and aren't terribly relevant in any reasonable public discourse. That even if there were tiny droppings of reasonable discussion to be had within a very narrow bounds, they did little beyond open the door for right wing talking points and pollute a space that really wasn't for that. That's probably why I'm at least more than a little blind to the current state of bigotry here (beyond typical straight white male privilege blinders).
Shame there are a few others that do not follow my example and do some combination of learn, shut up, or leave.
I agree people need to spend some time educating others. Not just LGBT stuff, but anything where someone misbehaves. People are so ready to just shut someone out. Give people a chance to improve, you might just get surprised when they do. There is an art to it. You have to keep a cool head and find a way to exploit their goodness against them. Twist their arm a bit and force them to look in the mirror.
My favorite thing to do on PA Nextdoor is to take Trump headlines and replace with Shapiro with a footer explaining what I did, and watch people foam at the mouth about Democrats being corrupt because they didn't read the last line.
And then when you point this out, you either get no response, or a level of incoherent mental gymnastics that I didn't think a human brain was capable of performing.
People need shared physical spaces where they can gain each other’s trust and discuss politics. Honestly simulating such an environment in high school would be well worth allocating 20% of the students’ time to. I’ve seen plenty of Democrats fluster in the exact same way (not as often by my recollection, but we’re all human and fallible in the same ways regardless of who we vote for).
I mean it helps that usually Democrat bad behavior is not quite as blatantly hypocritical. It happens. But rarely as bad as a $5,000 check if you vote for my team.
AI is here to stay the same way Reddit is here to stay. Maybe so, but that doesn't mean I need to use Reddit to function.
Likewise, I imagine for someone talking as a volunteer contributor to a large open source project that the author has every bit of choice in how to spend their time in such a community.
But a specialized game engine is the exact kind of niche I expect such supposed "artisan code" to remain valued. As the author mentioned, this is not just a tool programmers use. It's used to drive creatives, and creatives cannot simply function off of "well it works". I can point to plenty of games on Steam with such a descriptor.
I suppose things may be different professionally when it comes to being transformed into a duty of Slop Doula on the company dime. I still have rent to pay and will do what I need to do. But know I'll spend my off time preparing for a future where I am not reliant on such work to survive.
Can't remember who wrote it, but I agree with them. The basic gist of it was that plugins will become the norm in applications.
It's risky to accept AI-generated code into projects and integrate them to the core. What more and more applications need to do is have a robust plugin system. Then people can AI-slop/engineer/author as many plugins as they want with as much oversight as they're comfortable with.
This way the core application can stay artisanally coded if that's what the community and/or devs want and people can add features without disrupting it - or needing Yet Another Fork.
I agree, and I think this is going to be one of the big generational divides going forward. It'll be very interesting to see how the 'AI-natives' turn out... I really hope that most of them can find interesting and novel ways to make use of AI instead of just letting their skills atrophy (or never develop at all). I think there's all kinds of potential for them if they can reach it.
Truly unprecedented times we live in.
The author makes their disgust with AI very obvious, but disgust is not a good reason for prejudice.
Apparently they want a "safe space" for their anti-AI prejudices? They object to a "no dunking" rule that would prevent people from freely expressing their anti-AI prejudices using memes?
I wouldn't want to be in that kind of community.
I expect that it's only going to get worse though. Anti-AI populism is on the rise and uncritical boosters don't help either. They're going to feed each other.
Maybe we will need safe spaces for reasonable people who prefer evidence-based discussions?
That's not really what I took away from this at all. And your response does feel somewhat like you're validating this line:
I think there a few key points:
For my own views that echo the author's:
I think it's reasonable to allow AI-assisted contributions with a disclosure policy, while also allowing for Tildes-style "We're rejecting this because we decided to" PR policy. As well as not necessarily further welcoming AI boosrers into a space that previously rejected them.
In this era when any project can easily become flooded, project owners need to be given the grace to straight up reject anything without writing up a massive well-reasoned rejection note.
AI use is not inevitable. If a given community doesn't want it, and they ban it, then maybe the AI users should respect that, fork, and form their own community instead of trying to force their stance on the existing community. If their AI methods truely are superior, then they'll leave the old projects in the dust and both communities can live in peace apart from each other.
Reconciling this is more important on giant projects like the Linux kernel. Less so on a boutique project that I'm hearing about for the first time.
While we're here, what are your thoughts on the kernel?
I can't decide, cause obviously people vibecoding in there would be terrible, but at the same time it lowers the entry level knowledge of contributing, which might end up being a good thing for the project as a whole.
I know it'll be an eternal September kinda thing if that happens and so the maintainers who really only had to converse amongst themselves will suddenly find an overwhelming amount of PRs that are not up to standard, but, if they can set up a system to auto reject PRs based on some failed tests or something there might be some gems in there fixing like 20 year old security bugs.
I think the resolution for the kernel is 'quality is quality' and accountability. A disclosure policy, mandates that it must be tracable to a real person. And that if a PR fails a vibe check, the onus is on the submitter to pass that vibe check and not the maintainer. IIRC even before the rise of vibecoding the kernel had to push away a lot of sub-par cruft. CVEs that are not CVEs, that sort of thing.
The path curl took seems a prudent one: Close the monetary floodgates (reducing the tide from people throwing mud at a wall to see what sticks), do a hard ban, then ease up as maintainer ability to sustain improves, and still keep a high minimum bar to keep low-effort out.
I watched a ton of projects stop giving monetary incentives, that was super sad. I'm sure there were real people out there subsisting on those. Was among the first casualties for sure.
Really sucks, but when your bounties end up making the job harder than easier, then it's clear what needs to be done. Now if only we could do that in politics...
But yeah, some super specialized aspects of life need to be gatekept. It's one thing to welcoke perspective new scientists and another to be accommodating to throwing a bunch of garbage into a nuclear reactor. You got to start in safer environments and build up. I really wish such apprenticeship lanes existed to create the next generation of kernel developers, but that's a whole other can of worms.
I frequently go to live drawing sessions, cameras are forbidden in this spaces out of respect to the model. Nobody goes crazy about it, claiming that people are living in the past and that there's no longer value in drawing with a pencil when a camera can do it faster. At the same time, I can guarantee that most people on those session also use photographs to study the human body on their own. A good artists understands that drawing from life and drawing from a reference are completely different experiences. A camera will flatten the colors and values not matter how good it is, and observing something that is alive and moving gives you a more completed understanding on how it is constructed. At the same time, a good artist understands that is not necessary to do everything from scratch and that the right tool will save you a lot of time of money.
Whenever a conversation like this comes out, where people want their own space, others are quick to say that "they are trying to hold on the past, that world no longer exists". I wonder if this is unique to software. Do people tell others that there's no longer value in learning to draw, learning how an engine works and built your own model, or learning how to create a mechanical computer from scratch?
People understand that things are different in the professional world and the hobbyist communities. Why can't software be the same? There's still value in struggling, learning things on your own and talking with other people, sharing ideas. Otherwise why are you in tildes and not talking to an LLM?
Wanting a space, free of LLMs should be an uncontroversial opinion, the same way that wanting a space free of smart phones should be. I understand that this is a game engine that can be used in real games, not just hobbyist projects, but I still don't see the point on dismissing a person that wants to create a space only for humans. If someone wants to create that space, the only thing there's to do is respect their decision.
Edit: Fixed spelling, grammar, and some mistakes.
That's kind of the opposite of the situation, no? The context is that Bevy, an open source game engine that the author of the blogpost does not run, has decided to loosen its PR requirements to allow more LLM generated code.
The author of the blog, then, is "going crazy" about it.
I mean, the reality is that Bevy's maintainers are trying to make a thing that serves a role. It would be like if you went to a portrait photographer's studio and demanded that they cease taking photographers, cameras are the work of the devil, and oil paintings have much more character.
I think the latter opinion, that oil paintings of portraits have more character, is a justifiable subjective opinion to have, but like, it's not your photography studio. You're more than free to make your own studio that only does graphite portraits of live models. But you don't really have a say over everyone else.
If the author of the blog wants to make their own game engine that has strict anti-LLMs contribution requirements, by all means. But Bevy isn't there's, in the end.
You're right. I just meant that somewhere in the post, the author mentions she's creating a new community for Bevy. I was responding more to the idea that a community banning LLM use would have no real value beyond becoming an anti-AI circlejerk.
That depends on if you think the SMEs or the maintainers out in the field truly make up a community.
I suppose to use the live drawing analogy, it would be "in their right" for organizers to declare that suddenly photography of the nude models is okay. It would also be in the right of models to no longer want to participate in that community. I wouldn't call a model prudish in this day and age for not wanting the risk of their body being photogtaphed and posted online, even with arguments like "well what if someone is an amazing painter" or "well we just ban people who post pictures online".
I see parallels here. Ultimately these are decisions made that do not support the community, and it may end up making a worse feeling community, stripping the culture Bevy and Rust is known for.
That's the big issue here, actually. The paid maintainers do not necessarily have the same incentives as the unpaid. The paid ones need to ship and champion the product. The unpaid ones want to contribute to a greater collective and likely came here to escape that paid incentive that had them shortcut and compromise their way into keeping paychecks coming.
In particular, the paid maintainers do not care if community crumbles. At least, not until it starts affecting engije adoption. But until then, it's not part of their job to foster good vibes.
Sorry to dig into your take, but it already seems like the rest of the world is the safe space for people who want to discuss how LLMs/generative AI can be deployed for use in day to day life? The idea that there's widespread negative perception is largely an effect of the anglosphere bubble that many of us live in -- especially for those in the tech sector (which foolishly for decades decided that unionization was for dumb losers and continuously refused to organize >:| zero sympathy there from me tbh). The article author's take that the average person isn't accepting of this stuff is not reflective of reality -- the average American, sure, but there's more than just one country in the world that can buy things.
And anti-AI populism seems economically self defeating overall, so it's probably not something to be concerned with directly? I'd be moreso worried about it spiraling off and turning into another antivax -> Q-anon pipeline, but I don't know if a politically compatible conspiracy theory group exists. If all they do is burn down local data centres and protest in discord, it's not going to be a practical concern for anyone else.
IMO as I'd noted in another thread, the main impact from this will be excluding otherwise very motivated voices from discussions on how to realistically address the fallout from AI tooling existing. That seems like an objectively bad thing, but if people want to opt out, I don't see a reason not to support them in doing so?
I don't think a "safe space" is the default. When it's ambiguous what the rules are, you will get the more extreme pro- and anti- types getting into arguments with each other, resulting in increased polarization.
For example, recently on Bluesky a bunch of anti-AI types were dogpiling an open source developer who put out a new, experimental Linux port, because they used AI to do it. Someone from the Bluesky company had to say that this isn't acceptable.
Bluesky is not a space that has any official policy about AI being good or bad. They did add a way to label posts as AI generated. Also, there is an experimental Bluesky product that uses AI, so clearly there are people who work there who are into it.
I think it ought to be possible to have a "big tent" policy where both pro- and anti- AI opinions are acceptable? We are still figuring this stuff out.
I'm particularly aggravated by the folks who assume that being anti-AI is the only ethical position, that people who agree with them are ethical and people who disagree must be unethical. To me, that’s fighting dirty. Another way of fighting dirty would be assuming that there’s an anti-AI consensus in a community when actually, there is no consensus because they never really talked about it. I don’t know of that’s the case for the Bevy community, but it sort of sounds like it?
But I don't want to let that polarize me, because there are bad takes on both sides, and it doesn't mean people with more nuanced positions don't exist.
Its an interesting chart, but I think this interpretation is too focused on the "concerned over excited" and not enough on "excited over concerned". There is indeed a concern that correlates to "how much danager do I feel my futute is in from AI". But I find it equally interesting that the "excited over concerned" is staggeringly consistent distribution: 10-20%, with Israel being to only real outlier with 29%. A much much smaller spread than 16-50% on concern.
It seems overall, even the most "excited" countries are ultimately still in a "wait-and-see" mentality.
To use the author's words:
To use the obvious analog to Tildes: I'm sure it's hard to "definitively prove" Reddit has gotten worse over the years in all but a few extremely obvious cases for a few truly abandoned communities. But despite that, many of us did come seeking this community because we knew the vibes were off and Reddit no longer offered whatever we initially came there for. Maybe some of us use Tildes in tandem with Reddit, maybe some of us completely left the Reddit bar. But there's clearly reasons beyond objective facts that brought us here.
I can see how people spamming an open source project with low-effort demos could be a problem. And also, spamming the forums with low-effort memes would also be a problem. Aren't both of those solved with better moderation? In neither case is moderation about trying to change how people feel. Instead it's intended to get across what sort of behavior is ok in that forum.
Yes, they are. I don't think passive agreseive memes have ever solved any conflict, and even on my pseudanonymous accounts I try to avoid such low effort swipes (let alone my real name which would be how I enter a community like Bevy). But I guess some people don't have such restraint.
I do feel a bit iffy about banning emoji reactions, though. It's a core feature of the platform and if people can't express themselves that way, we end up with content much harder/time consuming to moderate. It's like every time you try to ban downvotes in Reddit; it never ends well, even with good intentions.
Moderation can be passive aggressive too. At some point, a bar needs to determine if it is a nazi bar or not. Trying to dance around the issue just leads to conflict. What I see here with little actions like banning reactions is "we want to welcome AI generated projects, but we don't want to say it out loud". Which is cowardly no matter your stance. At some point, the community will answer that question for you.
An observation of how a change in AI policy has affected the Bevy game engine community in the months since a new AI policy came into effect.
As can be inferred from the title, the author, a long-time Bevy contributor, has had a very negative experience with the changes in the community. Their post describes the new policy's effects on the behavior and health of the development community, and how the changes in the community have directly decreased the author's willingness to continue participating in the Bevy project.
It seems like we've got two groups of insecure people fighting each other online and they make everything worse for the rest of us:
I work on a semi-professional project together with LLM-using and LLM-avoiding developers alike. We never fight about it. Everyone is doing their best work using a different set of tools. It is in fact possible to co-exist while using different tech stacks. This kind of harmony is not something the internet has been good at achieving... ever. But I suppose in the past you'd mostly stick to your Gentoo forum and the Debian people stuck to theirs. Now you have every forum splitting in a novel way.
Edit: My current philosophy regarding LLMs and software development is that such a tremendous change to the industry provides opportunity to figure out new avenues to wealth and success as much as they fuck up the old avenues. If you’re a smart person who was able to figure out software engineering you’re also smart enough to be one of the first in whatever new frontiers emerge. Until computers obsolete the human mind the smarter humans should be able to use chaotic changes to great self benefit.
I work in games, so my attitude on a personal perspective is "landmine; avoid until defused".
Just being speulated that you're using AI in your game is OR suicide, and the potential benefits seem minimal anyway. So there's no point not simply staying course and keeping an eye out. My usage is pretty much limited to "ive googled this for 30 minutes with no good answer, let's see if Claude can give me new avenues to google". (P.S. I will never forgive Google for sabatoging its own search just to be more profitable than Sin)
Besides, there's already so many moving parts you have to manage in game development as is. Being asked to retool in yet another workflow that maybe potentially can kind of speed up development is a story you hear every other week, even pre-AI. I don't want to subscribe or worry about token usage for frontier models and trying to setup an AI workflow with open weight models (which I tried for a non-gaming experiment) was taking days as is when I could have simply plumbed it myself. I'm already vulnerable to automation hypnosis, so I don't want this to be yet another perpetrator.
(Use Kagi for search, no AI easy controls to filter crap you don't want)
I work in mobile games and I have first hand knowledge of household name mobile game production. They use AI for everything. Most used AI and ML years before ChatGPT was a thing, both for data and art. They had (and still have) custom systems for generating seasonal content variants, backgrounds etc. This gives artists more time to focus on "hero art" instead of drawing trees and rocks and castles.
Only PC gamers care even a little about AI in their games and IMO most of it is just baseless hysteria. Like Expedition 33 having one background texture AI-generated causing a massive rage storm.
I'm no game developer. I'm not a developer of any sort, but as a user, if I find that an artistic project I was interested in was made with AI, I'm not going to partake in said project. I won't play games made with AI, I won't read AI books, and I won't use an AI recipe. I do my best to avoid LLMs as much as possible because I just don't want them in my life. I get that the author is upset, and I understand why as someone who is just so tired of having AI/LLM use shoved down my throat every day for literally everything I do, when I just want human-created art. I want humans to code and do the art for games, I want humans to write books, and I want humans to create things. Especially because this use is so harsh and dangerous to our environment, and climate change is a massive issue that we just aren't addressing, but it is making worse with every new data center that is deployed.
And I think you bring up a good point the author touched on: If you are casting a wide net, courting artists to be a part of your community, you need to tread very carefully with your attitudes toward AI contributions.
A lot of artists aren't going to take well to making some contributions, then have somebody take that hard effort and say 'well I took this and used it as a foundation to have the AI generate an entire asset library in that style.'
At the risk of breaking the "no dunking" rule, I think this entire post comes from a very bad faith position. That doesn't mean the position isn't valid, I think Fallible isn't entirely incorrect or out to lunch here, but what I will say is that they clearly are tacitly aware they're standing in moving water. Whether you like it or not, and I'll personally generally err on the side of "not", this isn't going away. If you're on the side that doesn't want anything to do with the topic of AI-assisted coding, and this hasn't been a point of contention before because it, well, didn't exist, then I can see how you feel a community that you used to love slowly filled up with the worst kind of vocal assholes. This is nothing new, and I do feel sorry for them, and I've seen it before in different communities with different issues.
I just don't really see, on this particular issue, how you're meant to go forward. For what it's worth, I think the AI policy that Bevy has currently is very measured and valid, and drawing comparisons to the Nazi bar problem are weirdly pointed and unhelpful. But, and I say this without trying to be unhelpful, you can't make everyone happy.
I'm willing to believe that, despite the assertion that the technology is inextricable from its negative aspects (which I don't believe, by the way), there's some pretty significant or at least important parts of Bevy that are wholly written by AI. I have no reason to believe this, true, but I also have no reason to doubt it. This just happens now. It's table stakes, see my comment on the platform Cara being scraped for more details. And maybe it's bad, maybe it's good, sure, but it doesn't change the reality that it's only going to get worse, or better, depending on what side of the aisle you're on.
What I can not understand is either side making a weirdly emotional argument over the state of any given project. Last time I checked, Bevy is open source, and it can be forked. If you'd like to do that, you can, and again, I don't say this to dunk on anyone. I've forked projects before because they didn't go the way I wanted them, and I will continue to do so, because sometimes getting the thing done how I want it to be done is easier than fighting with a community who doesn't see it your way.
This is not me saying "if you don't like it, you can leave", it's me saying "sometimes it's better to curate your experience yourself". I mean, they clearly said that they don't much enjoy the discussions or the community how they used to. Seems like holding on will only make it worse, something readers of my contributions on here will know I personally had to learn myself.