With regards to the ai-tooling, the description is “The design, ux, development of AI/LLM-powered tools.” Use of AI tooling is not in scope.
The litmus test I’m using for this tag is if the body of work still makes sense if I were to replace any instance of LLM with consultant and it still hold up, it’s more likely than not outside of the bounds of topicality for the site.
With regards to the ai-tooling, the description is “The design, ux, development of AI/LLM-powered tools.” Use of AI tooling is not in scope.
Again, I don’t really care about my deleted story, just understanding the mod rules, but the agentic monitoring story was about how the author had built a LLM-powered tool to monitor logs. I don’t think the building v. use distinction is sufficiently clear to make in practice because for software development most tooling is stuff you build so you personally can use it. At the very least, I would rather err on the side of allowing use stories because at the moment, everyone is struggling to figure out how to use ai-tooling optimally, and some people contend that the optimal use of ai-tooling is none at all.
Look, I appreciate that navigating the conflicting demands here is difficult and I can see you are trying to figure out the best path. I’m not a huge user of Lobsters, so feel free to discount this, but:
At minimum I would suggest changing the tag to building-ai or something rather than ai-tools. The term ai-tools makes me think it is for discussion of the tools, including their use.
Stepping back: I don’t see how you can discuss the design, UX and development of a tool without talking about how it is used. I think it’s clear that “how will people use this?” is the primary consideration tool makers have when thinking about a tool’s design and UX, and it of course informs how the tool is developed.
Just to give my overall perspective here: I really value sites like Lobsters as a place to learn how to be a better software engineer. Since starting my career, sites like this have helped me deepen my understanding of all manner of topics: distributed databases, PL design, operating systems, testing methodology, type systems, tons of stuff like that. As a working engineer, it is obvious to me that the effective use of LLMs in software engineering is not only the next frontier in the evolution of our job, but it is the dominant and most important evolution. And as a working engineer, “how do I best use XYZ AI tool” is now, by far, the most important question for me to answer as I try to improve how I build software. Let’s talk about these tools! It’s really important!
To put it a bit more provocatively, I worry that any site that simply bans discussion of how to use these tools is consigning itself to irrelevance. I struggle to see how a community that is reflexively against even the discussion of the use of AI in software engineering will have anything useful to say about the actual practice of software engineering in the near future.
Anyway - I see myself as a visitor here, so feel free to disregard. I completely understand and endorse wanting to ban the endless existential career/social/blah blah dread type posts that are so frequent on other sites. But I really do not think banning the discussion of, for example, “how can we best use AI agents to monitor the site” is the right path for Lobsters.
At minimum I would suggest changing the tag to building-ai or something rather than ai-tools. The term ai-tools makes me think it is for discussion of the tools, including their use.
That’s a fair point. Would something like ai-tool-dev work? Part of the reason for the name is to make sure that ai and ai-tools show up near each other in the multi-select on the new submission page.
Stepping back: I don’t see how you can discuss the design, UX and development of a tool without talking about how it is used. I think it’s clear that “how will people use this?” is the primary consideration tool makers have when thinking about a tool’s design and UX, and it of course informs how the tool is developed.
Just to give my overall perspective here: I really value sites like Lobsters as a place to learn how to be a better software engineer. Since starting my career, sites like this have helped me deepen my understanding of all manner of topics: distributed databases, PL design, operating systems, testing methodology, type systems, tons of stuff like that. As a working engineer, it is obvious to me that the effective use of LLMs in software engineering is not only the next frontier in the evolution of our job, but it is the dominant and most important evolution. And as a working engineer, “how do I best use XYZ AI tool” is now, by far, the most important question for me to answer as I try to improve how I build software. Let’s talk about these tools! It’s really important!
From my perspective, if you’re talking about the user experience and the use of the tool in a cohesive body of work, that’s a topical article under the ai-tooling tag. I don’t think having a straight discussion of use is topical.
To slightly expand on that point, I would like to expand on my litmus test. The reason why I’m evaluating these articles by replacing the “LLM” in them with “consultant”, is try and see if the article could also work as an article discussing the management of skills and abilities of a human versus the skills and abilities of a product. If this is the case, I’m leaning towards that the article is more of an article which covers the management side of the industry and out of scope for the site. We are also, explicitly, trying to avoid the “I made a thing an LLM” style postings. Handing off a unfamiliar repository with a thin announcement associated with it doesn’t provide the pedagogical experience we are going for.
That point being said, I’m fully aware that there’s going to be an amount of which does not fit in ai-tooling and fails the “consulting” test. My suspicion is that you are after a subset of that content for your “continuing education” needs. To that end, I’m not entirely sure what a definition of that content looks like that could be used to form the basis of a potential future tag. As demonstrated with ai-tooling, the less ambiguous those definitions can be, the better the tag will function. If you (or anyone reading this) wants to make an attempt at it though, I’m more than happy to be a sounding board.
Anyway - I see myself as a visitor here, so feel free to disregard. I completely understand and endorse wanting to ban the endless existential career/social/blah blah dread type posts that are so frequent on other sites. But I really do not think banning the discussion of, for example, “how can we best use AI agents to monitor the site” is the right path for Lobsters.
Targeted, constructive, feedback is always useful. Don’t feel like participation is a gate to providing it. We (the moderation team), do read it all, even if we’re not responding to it.
IMO that doesn’t really solve the original problem. Both ai-tools and ai-tool-dev rather imply “built with / using AI tooling” than “building AI tooling”. I’ve been raking my brain for a while, but I don’t think you can really imply otherwise while having AI be the first word. building-ai or developing-ai would both imply what the tag is actually meant to imply.
Part of the reason for the name is to make sure that ai and ai-tools show up near each other in the multi-select on the new submission page.
The submit page narrows by search anyway so someone looking for AI will see both ai and developing/building-ai and I’m pretty sure if someone wants to filter the tag, then they won’t be bothered by having to scroll a tiny bit more on the Filter page.
With regards to the ai-tooling, the description is “The design, ux, development of AI/LLM-powered tools.” Use of AI tooling is not in scope.
The litmus test I’m using for this tag is if the body of work still makes sense if I were to replace any instance of LLM with consultant and it still hold up, it’s more likely than not outside of the bounds of topicality for the site.
score 1 · canonical
Again, I don’t really care about my deleted story, just understanding the mod rules, but the agentic monitoring story was about how the author had built a LLM-powered tool to monitor logs. I don’t think the building v. use distinction is sufficiently clear to make in practice because for software development most tooling is stuff you build so you personally can use it. At the very least, I would rather err on the side of allowing use stories because at the moment, everyone is struggling to figure out how to use ai-tooling optimally, and some people contend that the optimal use of ai-tooling is none at all.
score 2 · canonical
Look, I appreciate that navigating the conflicting demands here is difficult and I can see you are trying to figure out the best path. I’m not a huge user of Lobsters, so feel free to discount this, but:
At minimum I would suggest changing the tag to
building-aior something rather thanai-tools. The termai-toolsmakes me think it is for discussion of the tools, including their use.Stepping back: I don’t see how you can discuss the design, UX and development of a tool without talking about how it is used. I think it’s clear that “how will people use this?” is the primary consideration tool makers have when thinking about a tool’s design and UX, and it of course informs how the tool is developed.
Just to give my overall perspective here: I really value sites like Lobsters as a place to learn how to be a better software engineer. Since starting my career, sites like this have helped me deepen my understanding of all manner of topics: distributed databases, PL design, operating systems, testing methodology, type systems, tons of stuff like that. As a working engineer, it is obvious to me that the effective use of LLMs in software engineering is not only the next frontier in the evolution of our job, but it is the dominant and most important evolution. And as a working engineer, “how do I best use XYZ AI tool” is now, by far, the most important question for me to answer as I try to improve how I build software. Let’s talk about these tools! It’s really important!
To put it a bit more provocatively, I worry that any site that simply bans discussion of how to use these tools is consigning itself to irrelevance. I struggle to see how a community that is reflexively against even the discussion of the use of AI in software engineering will have anything useful to say about the actual practice of software engineering in the near future.
Anyway - I see myself as a visitor here, so feel free to disregard. I completely understand and endorse wanting to ban the endless existential career/social/blah blah dread type posts that are so frequent on other sites. But I really do not think banning the discussion of, for example, “how can we best use AI agents to monitor the site” is the right path for Lobsters.
score 3 · canonical
That’s a fair point. Would something like
ai-tool-devwork? Part of the reason for the name is to make sure thataiandai-toolsshow up near each other in the multi-select on the new submission page.From my perspective, if you’re talking about the user experience and the use of the tool in a cohesive body of work, that’s a topical article under the
ai-toolingtag. I don’t think having a straight discussion of use is topical.To slightly expand on that point, I would like to expand on my litmus test. The reason why I’m evaluating these articles by replacing the “LLM” in them with “consultant”, is try and see if the article could also work as an article discussing the management of skills and abilities of a human versus the skills and abilities of a product. If this is the case, I’m leaning towards that the article is more of an article which covers the management side of the industry and out of scope for the site. We are also, explicitly, trying to avoid the “I made a thing an LLM” style postings. Handing off a unfamiliar repository with a thin announcement associated with it doesn’t provide the pedagogical experience we are going for.
That point being said, I’m fully aware that there’s going to be an amount of which does not fit in
ai-toolingand fails the “consulting” test. My suspicion is that you are after a subset of that content for your “continuing education” needs. To that end, I’m not entirely sure what a definition of that content looks like that could be used to form the basis of a potential future tag. As demonstrated withai-tooling, the less ambiguous those definitions can be, the better the tag will function. If you (or anyone reading this) wants to make an attempt at it though, I’m more than happy to be a sounding board.Targeted, constructive, feedback is always useful. Don’t feel like participation is a gate to providing it. We (the moderation team), do read it all, even if we’re not responding to it.
score 1 · canonical
IMO that doesn’t really solve the original problem. Both
ai-toolsandai-tool-devrather imply “built with / using AI tooling” than “building AI tooling”. I’ve been raking my brain for a while, but I don’t think you can really imply otherwise while having AI be the first word.building-aiordeveloping-aiwould both imply what the tag is actually meant to imply.The submit page narrows by search anyway so someone looking for AI will see both
aianddeveloping/building-aiand I’m pretty sure if someone wants to filter the tag, then they won’t be bothered by having to scroll a tiny bit more on the Filter page.score 1 · canonical
I think we’re all slowly realizing that anything related to LLMs/GenAI that’s not covered by the two tag descriptions:
ai: “Developing artificial intelligence, machine learning, LLMs.”ai-tooling: “The design, ux, development of AI/LLM-powered tools.”is explicitly off-topic.
score 3 · canonical
Noted - evangelising and complaining are equally off-topic.
score 2 · canonical
That as far as I can understand it is also how you are meant to understand it.
score 1 · canonical
You’d think, but in the comment I’m replying to he says (emphasis mine):
score 3 · canonical
I see. Well, let’s hope most people have better reading comprehension than me then.
score 1 · canonical