Upvote Upvoted 27 Downvote Downvoted
1 2 3
TF2 Slopware
61
#61
-9 Frags – +

https://youtu.be/cPbAtRhFcpQ?si=xdqt6OrezvueytZ7
I need this in tf2

https://youtu.be/cPbAtRhFcpQ?si=xdqt6OrezvueytZ7
I need this in tf2
62
#62
47 Frags – +

guys, we should all be proud of ourselves, we have an active thread that isn't about someone being a pedophile!!!

guys, we should all be proud of ourselves, we have an active thread that isn't about someone being a pedophile!!!
63
#63
-22 Frags – +
WARHURYEAHguys, we should all be proud of ourselves, we have an active thread that isn't about someone being a pedophile!!!

I'm 100% sure someone will make thread like this after your post -,-

[quote=WARHURYEAH]guys, we should all be proud of ourselves, we have an active thread that isn't about someone being a pedophile!!![/quote]
I'm 100% sure someone will make thread like this after your post -,-
64
#64
3 Frags – +

slightly off topic but i enjoyed reading some of the other blogposts linked so i thought id share.

obviously this general issue has been an enormous point of discussion in maths recently and has been occupying much of my thoughts. if anyone is interested this site (https://proofsandprompts.com/) is a collation of essays/blogposts from mathematicians. most are relatively nontechnical and some have at least somewhat helped with my depression/anger/whatever due to recent LLM industry action in maths. this one (https://proofsandprompts.com/2026/10/07/the-grand-piano/) from today i especially liked

slightly off topic but i enjoyed reading some of the other blogposts linked so i thought id share.

obviously this general issue has been an enormous point of discussion in maths recently and has been occupying much of my thoughts. if anyone is interested this site (https://proofsandprompts.com/) is a collation of essays/blogposts from mathematicians. most are relatively nontechnical and some have at least somewhat helped with my depression/anger/whatever due to recent LLM industry action in maths. this one (https://proofsandprompts.com/2026/10/07/the-grand-piano/) from today i especially liked
65
#65
-10 Frags – +
wonderWARHURYEAHguys, we should all be proud of ourselves, we have an active thread that isn't about someone being a pedophile!!!I'm 100% sure someone will make thread like this after your post -,-

I was wrong

[quote=wonder][quote=WARHURYEAH]guys, we should all be proud of ourselves, we have an active thread that isn't about someone being a pedophile!!![/quote]
I'm 100% sure someone will make thread like this after your post -,-[/quote]
I was wrong
66
#66
dribble.tf
12 Frags – +

After having just pushed a lot of new AI built features to dribble.tf, stumbling onto this thread was really eye opening and I wanted to give my 2c

dribble has been around for 6+ years and the reasons for me continuing to work on it has changed, evolving from something I was intensely particular about in terms of craft/code - to something i was okay shitting out a ton of features in two days

It started as something I wanted to build for myself, and ultimately for the comp community. But seeing as I don’t play TF2 anymore or participate in the community - the main reason I (intermittently) continue adding to it is 1) for approval 2) hope it still benefits some people 3) to fully realize what i wanted the app to be

I work at a software agency, where to the majority of clients, the final output is what matters at the end of the day - regardless of the path you took. This has only been exacerbated by AI, and it’s immensely disheartening the dying emphasis on actual understanding

It was from banging my head over weeks that I better grasped 3D vector math, file/render optimization methods etc. which led to some great chats with other engineers who dealt with similar challenges, geeking out and learning from one another. It’s increasingly easy to lose that social aspect - and it all becomes about “what value added” are you bringing

Like a lot of others, I got a ton of joy from the problem solving and creativity that came with coding - but the industry almost induces a sense of guilt if you aren’t using AI. And in truth, it has managed to do things in dribble that I never would’ve been able to figure out - which further adds to the kool aid. Its scary that if you flip that switch of “fuck it llm go ball”, the productivity becomes addictive (or even straight mandatory to keep your job) - then you begin to not care how the sausage gets made because it tastes good enough

On another hand, it’s really awesome to be able to see dribble be closer to the app I always wanted, but maybe it also feels exploitative (?) because I don’t use it nor participate in the community any more. So it’s only potential benefit to me now is as a portfolio piece. And truthfully that is part of my motivation - I always was a bit embarrassed it didn’t work well on mobile or looked kinda jank. Although I don’t think it’s mutually exclusive from also wanting to create something useful

I keep finding myself having to justify: if someone were to look at my app (even prior to recent AI additions) in modern contexts would probably think it was AI generated. So the effort I put into the work years ago gets washed out. This sort of thinking certainly edges me towards focusing on the result instead of the journey. I end up thinking anything I produce will be compared to something that can/has been done with AI - that craft has little value anymore

I do wonder how dribble would’ve been taken if released one week ago instead of 6 years. I think it’s important to be a part of communities like this that care deeply and will argue; because a massive part of software is social. The feeling I love when using an app is when you can tell a person added this exact interface or interaction “for you” - like a “dw I gotchu”. If you’ve used Chinese mobile apps before, they are highly functional but feel sorta careless

Im def not an advocate of LLMs, the world before it was certainly better. This thread was just really interesting and all the spaces I’m in deal with software as a job, whereas in communities like this where there’s no financial incentive - motivations and respect matter a whole lot

With all that said, I know I won’t have the time to spend on dribble by hand anymore, and if I were to add more features, it would be with Claude. But ultimately the app I will try to build is the one I always wanted back then

Theres the morality and raping-of-the-earth part of AI but is a different discussion - but I wanted to just share as someone who’s had a project that went from one end to the other, and different involvement with the community

After having just pushed a lot of new AI built features to dribble.tf, stumbling onto this thread was really eye opening and I wanted to give my 2c

dribble has been around for 6+ years and the reasons for me continuing to work on it has changed, evolving from something I was intensely particular about in terms of craft/code - to something i was okay shitting out a ton of features in two days

It started as something I wanted to build for myself, and ultimately for the comp community. But seeing as I don’t play TF2 anymore or participate in the community - the main reason I (intermittently) continue adding to it is 1) for approval 2) hope it still benefits some people 3) to fully realize what i wanted the app to be

I work at a software agency, where to the majority of clients, the final output is what matters at the end of the day - regardless of the path you took. This has only been exacerbated by AI, and it’s immensely disheartening the dying emphasis on actual understanding

It was from banging my head over weeks that I better grasped 3D vector math, file/render optimization methods etc. which led to some great chats with other engineers who dealt with similar challenges, geeking out and learning from one another. It’s increasingly easy to lose that social aspect - and it all becomes about “what value added” are you bringing

Like a lot of others, I got a ton of joy from the problem solving and creativity that came with coding - but the industry almost induces a sense of guilt if you aren’t using AI. And in truth, it has managed to do things in dribble that I never would’ve been able to figure out - which further adds to the kool aid. Its scary that if you flip that switch of “fuck it llm go ball”, the productivity becomes addictive (or even straight mandatory to keep your job) - then you begin to not care how the sausage gets made because it tastes good enough

On another hand, it’s really awesome to be able to see dribble be closer to the app I always wanted, but maybe it also feels exploitative (?) because I don’t use it nor participate in the community any more. So it’s only potential benefit to me now is as a portfolio piece. And truthfully that is part of my motivation - I always was a bit embarrassed it didn’t work well on mobile or looked kinda jank. Although I don’t think it’s mutually exclusive from also wanting to create something useful

I keep finding myself having to justify: if someone were to look at my app (even prior to recent AI additions) in modern contexts would probably think it was AI generated. So the effort I put into the work years ago gets washed out. This sort of thinking certainly edges me towards focusing on the result instead of the journey. I end up thinking anything I produce will be compared to something that can/has been done with AI - that craft has little value anymore

I do wonder how dribble would’ve been taken if released one week ago instead of 6 years. I think it’s important to be a part of communities like this that care deeply and will argue; because a massive part of software is social. The feeling I love when using an app is when you can tell a person added this exact interface or interaction “for you” - like a “dw I gotchu”. If you’ve used Chinese mobile apps before, they are highly functional but feel sorta careless

Im def not an advocate of LLMs, the world before it was certainly better. This thread was just really interesting and all the spaces I’m in deal with software as a job, whereas in communities like this where there’s no financial incentive - motivations and respect matter a whole lot

With all that said, I know I won’t have the time to spend on dribble by hand anymore, and if I were to add more features, it would be with Claude. But ultimately the app I will try to build is the one I always wanted back then

Theres the morality and raping-of-the-earth part of AI but is a different discussion - but I wanted to just share as someone who’s had a project that went from one end to the other, and different involvement with the community
67
#67
1 Frags – +
smesoI keep finding myself having to justify: if someone were to look at my app (even prior to recent AI additions) in modern contexts would probably think it was AI generated. So the effort I put into the work years ago gets washed out. This sort of thinking certainly edges me towards focusing on the result instead of the journey. I end up thinking anything I produce will be compared to something that can/has been done with AI - that craft has little value anymore

Dude, please don't peg yourself and your work against a bunch of vectors. You have more impact to this community than you realize. From a technical standpoint, you were able to move demos out of the game, fix replays and paved the path that eventually led to the current tools revolving demos. Heck, they for sure trained on your repo as well.

Please don't even think that those years that you made and hosted dribble pre-AI were wasted, I can't speak for everyone, but I very much respect those who shell our their time, energy in pursuit of something that they are passionate about. Sure AI kinda mogs some technical work when it comes to pure output, but you cannot deny that in all these years of development, you grew with the project, that in deployment, you celebrated with the community, and that in hosting, you were in a way supporting our little chud niche-gamemode-in-declining-game's livelihood as you bore the server bills.

From my perspective, you sowed the seeds, tilled the dirt, set out to and eventually grew the plant. In a world where people only care about fruits, of course the fertilizer is viewed as a breadwinner, but there would be nothing to fertilize if you did not set out to grow that plant in the first place.

hope ur doing well. it's nice seeing you again and I am still a big fan and user of dribble :)

[quote=smeso]I keep finding myself having to justify: if someone were to look at my app (even prior to recent AI additions) in modern contexts would probably think it was AI generated. So the effort I put into the work years ago gets washed out. This sort of thinking certainly edges me towards focusing on the result instead of the journey. I end up thinking anything I produce will be compared to something that can/has been done with AI - that craft has little value anymore[/quote]

Dude, please don't peg yourself and your work against a bunch of vectors. You have more impact to this community than you realize. From a technical standpoint, you were able to move demos out of the game, fix replays and paved the path that eventually led to the current tools revolving demos. Heck, they for sure trained on your repo as well.

Please don't even think that those years that you made and hosted dribble pre-AI were wasted, I can't speak for everyone, but I very much respect those who shell our their time, energy in pursuit of something that they are passionate about. Sure AI kinda mogs some technical work when it comes to pure output, but you cannot deny that in all these years of development, you grew with the project, that in deployment, you celebrated with the community, and that in hosting, you were in a way supporting our little chud niche-gamemode-in-declining-game's livelihood as you bore the server bills.

From my perspective, you sowed the seeds, tilled the dirt, set out to and eventually grew the plant. In a world where people only care about fruits, of course the fertilizer is viewed as a breadwinner, but there would be nothing to fertilize if you did not set out to grow that plant in the first place.

hope ur doing well. it's nice seeing you again and I am still a big fan and user of dribble :)
68
#68
15 Frags – +

As someone who _was_ about to enter a degree in computer science, and has had programming as their primary hobby ever since they were 14, it has been hard to feel anything but dread when browsing new upcoming projects on github or teamfortress.tv. There is no more excitement, in how the creators of these projects overcame issues, new tech they invented, something new they learned. I used to reach out to creators and ask about their process, I used to read through the code, discovering new styles and personalities, quirks and stories told right there in the code or by the people behind it. But these days are sadly gone. I no longer feel excited, I feel dreadful.

I tried to set an example and make something that others might find useful, and also for those who might feel the same as me to hold on to, that there are still projects crafted with love and not only with a finished goal in mind. Programming was as you said smeso, a creative and learning process, where you learn from and with others. But when I foolishly showed my WIP, my idea got taken, ran through AI and published within a day. So me and a friend, who I thankfully learned a lot from during this project, just gave up.
2 years ago I was excited to work a tiny bit on ETF2L, wondering what else I could do in the future. But when I got asked recently whether I want to help as an actual coder for ETF2L, I had to decline. I couldn't make myself commit to a project, where everyone around me only cares for results, and frankly, my human results don't compare to that of AI.
In my jobs so far I had to deal with these kinds of developers already, which is fair enough. But I wont commit my free time to prompting agents too or try to have a conversation with someone who rather asks claude than learn something new with me.
I would love to discuss the new C29 changes with a friend, but why bother when no human writes C anymore to feel these changes. So I sit there in silent.

Perhaps I should just mute the projects section on here, perhaps I should close github and accept that there is no place for people who value interaction, who wish to learn and work together. Perhaps.

But it pains me to see projects I used to look up to get flooded with code that nobody ever read nor wrote. The user does not care "where the tasty sausage comes from", and probably never has. And I can't change that.
There could have been a new generation of excited people willing to move TF2 forward together and create bonds, but productivity metrics and chasing goals lead to said next generation being claude and GPT instead.

I would love to end this segment with something optimistic, with some hope that there could still be a place for a community who'd rather not finish a project and make new friends, than sit alone at the end of a finished prompt execution. But there isn't, except for small friend circles that pretend the outside world doesn't exist to keep motivation.

As someone who _was_ about to enter a degree in computer science, and has had programming as their primary hobby ever since they were 14, it has been hard to feel anything but dread when browsing new upcoming projects on github or teamfortress.tv. There is no more excitement, in how the creators of these projects overcame issues, new tech they invented, something new they learned. I used to reach out to creators and ask about their process, I used to read through the code, discovering new styles and personalities, quirks and stories told right there in the code or by the people behind it. But these days are sadly gone. I no longer feel excited, I feel dreadful.

I tried to set an example and make something that others might find useful, and also for those who might feel the same as me to hold on to, that there are still projects crafted with love and not only with a finished goal in mind. Programming was as you said smeso, a creative and learning process, where you learn from and with others. But when I foolishly showed my WIP, my idea got taken, ran through AI and published within a day. So me and a friend, who I thankfully learned a lot from during this project, just gave up.
2 years ago I was excited to work a tiny bit on ETF2L, wondering what else I could do in the future. But when I got asked recently whether I want to help as an actual coder for ETF2L, I had to decline. I couldn't make myself commit to a project, where everyone around me only cares for results, and frankly, my human results don't compare to that of AI.
In my jobs so far I had to deal with these kinds of developers already, which is fair enough. But I wont commit my free time to prompting agents too or try to have a conversation with someone who rather asks claude than learn something new with me.
I would love to discuss the new C29 changes with a friend, but why bother when no human writes C anymore to feel these changes. So I sit there in silent.

Perhaps I should just mute the projects section on here, perhaps I should close github and accept that there is no place for people who value interaction, who wish to learn and work together. Perhaps.

But it pains me to see projects I used to look up to get flooded with code that nobody ever read nor wrote. The user does not care "where the tasty sausage comes from", and probably never has. And I can't change that.
There could have been a new generation of excited people willing to move TF2 forward together and create bonds, but productivity metrics and chasing goals lead to said next generation being claude and GPT instead.

I would love to end this segment with something optimistic, with some hope that there could still be a place for a community who'd rather not finish a project and make new friends, than sit alone at the end of a finished prompt execution. But there isn't, except for small friend circles that pretend the outside world doesn't exist to keep motivation.
69
#69
serveme.tf
8 Frags – +
labricecatBut when I foolishly showed my WIP, my idea got taken, ran through AI and published within a day.

Since this is about me, I'm sorry it happened that way. I didn't want you and your friend to stop working on the project, quite the opposite. I would have liked your project to be built in a way that wouldn't require a sourcemod plugin transmitting over websockets at 5 tick and sending all STV connects to a server not controlled by serveme.tf. I enjoy integrating interesting projects with serveme.tf in a way that benefits the community. It's why I helped mitch with a new API to get live loglines of any server (but without IPs). It's also why I showed you the vibecoded proof of concept within a day. I was already parsing STVs for other purposes, so I figured it was technically possible. Seeing what else could be done with STV got me carried away.

I don't think I took your idea, though. Live match info for casters, a cheat feed, a better autodirector and a minimap have been around for years. I built three of those myself over a decade ago, and better people have since made better versions. It's interesting to me to see different people's takes and improvements on these ideas.

I suggested building on STV and logaddress instead of a custom plugin. That didn't fit how you wanted to work, which is fair, so I kept building my own thing. It does overlap with yours: it has a minimap and a larger map too. After a few more weeks, most of it is now something else, though: shared STV relays, instant scrubbing through a match, in-game demo review, telestrator functionality, an event timeline, team comms injection.

It all works on any TF2 server, and the same techniques would work for your project too. I'd like to see yours finished.

Hopefully you'll still start your CS degree. I started mine in 2004, facing the massive disruption of click and drag programming, low code and UML diagrams making programmers obsolete. Newspaper headlines questioning what all these useless programmers currently in school would be doing later. In the years before that, there were those who thought OOP would make writing new code useless: just reuse all the things. And before that, COBOL would let business people write the code themselves. We're all still here, even the COBOL programmers.

[quote=labricecat]But when I foolishly showed my WIP, my idea got taken, ran through AI and published within a day.[/quote]
Since this is about me, I'm sorry it happened that way. I didn't want you and your friend to stop working on the project, quite the opposite. I would have liked your project to be built in a way that wouldn't require a sourcemod plugin transmitting over websockets at 5 tick and sending all STV connects to a server not controlled by serveme.tf. I enjoy integrating interesting projects with serveme.tf in a way that benefits the community. It's why I helped mitch with a new API to get live loglines of any server (but without IPs). It's also why I showed you the vibecoded proof of concept within a day. I was already parsing STVs for other purposes, so I figured it was technically possible. Seeing what else could be done with STV got me carried away.

I don't think I took your idea, though. Live match info for casters, a cheat feed, a better autodirector and a minimap have been around for years. I built three of those myself over a decade ago, and better people have since made better versions. It's interesting to me to see different people's takes and improvements on these ideas.

I suggested building on STV and logaddress instead of a custom plugin. That didn't fit how you wanted to work, which is fair, so I kept building my own thing. It does overlap with yours: it has a minimap and a larger map too. After a few more weeks, most of it is now something else, though: shared STV relays, instant scrubbing through a match, in-game demo review, telestrator functionality, an event timeline, team comms injection.

It all works on any TF2 server, and the same techniques would work for your project too. I'd like to see yours finished.

Hopefully you'll still start your CS degree. I started mine in 2004, facing the massive disruption of click and drag programming, low code and UML diagrams making programmers obsolete. Newspaper headlines questioning what all these useless programmers currently in school would be doing later. In the years before that, there were those who thought OOP would make writing new code useless: just reuse all the things. And before that, COBOL would let business people write the code themselves. We're all still here, even the COBOL programmers.
70
#70
19 Frags – +

While I'm personally increasingly disappointed in the amount of LLM code being used "around here", given the very large list of reasons why LLMs are problematic. I'll continue creating my own human-authored bugs code for the foreseeable futures, even if it means re-implementing a bunch of work from scratch to avoid having slop in my dependencies.

I would encourage others to keep doing their good work despite everything, but I can't blame people from getting discouraged and giving up.

While I'm personally increasingly disappointed in the amount of LLM code being used "around here", given the very large list of reasons why LLMs are problematic. I'll continue creating my own human-authored [s]bugs[/s] code for the foreseeable futures, even if it means re-implementing a bunch of work from scratch to avoid having slop in my dependencies.

I would encourage others to keep doing their good work despite everything, but I can't blame people from getting discouraged and giving up.
71
#71
8 Frags – +

Heya, a note here as ETF2L was mentioned. Maybe it was a bit of a misunderstanding but as of right now, at least to my knowledge, none of the live code for ETF2L was written by an AI. All the old website changes like the result submissions through logs.tf or the map draws addition for this 6s season were all done by banging my head against the codebase and hoping it works.

However, I can not give any certainty that this is a permanent state for ETF2L or tools made for it. The project is volunteer driven and if a volunteer makes a project with the aid of AI that fulfills a gap for us it would be a shame to blindly decline due to AI use. That being said even if I can distinguish every time if a PR was made with AI assistance or not.

However, that does not mean that any generated code/project will get accepted. I want projects to be human driven with AI being a potential assistant. Meaning the quality, understanding, and love and care for what is being built are the driving force. I want people that are working on the projects to be able to learn from what they're building, as well as the other volunteers along the way.

Future generations might look back on these projects and still be maintaining them in 20 years, similarly to how we are currently doing, so willy-nilly generating things is not what I want.

Heya, a note here as ETF2L was mentioned. Maybe it was a bit of a misunderstanding but as of right now, at least to my knowledge, none of the live code for ETF2L was written by an AI. All the old website changes like the result submissions through logs.tf or the map draws addition for this 6s season were all done by banging my head against the codebase and hoping it works.

However, I can not give any certainty that this is a permanent state for ETF2L or tools made for it. The project is volunteer driven and if a volunteer makes a project with the aid of AI that fulfills a gap for us it would be a shame to blindly decline due to AI use. That being said even if I can distinguish every time if a PR was made with AI assistance or not.

However, that does not mean that any generated code/project will get accepted. I want projects to be human driven with AI being a potential assistant. Meaning the quality, understanding, and love and care for what is being built are the driving force. I want people that are working on the projects to be able to learn from what they're building, as well as the other volunteers along the way.

Future generations might look back on these projects and still be maintaining them in 20 years, similarly to how we are currently doing, so willy-nilly generating things is not what I want.
72
#72
1 Frags – +

As someone who spent roughly 7 years without the clankers at a web dev job and started using these tools more over the past few months, these are my two cents. (Sorry for the nerd essay up front)

The slop that is not good will simply not get attention and people will move on. That does not mean everything that is made with AI is inherently bad.

I do think these agentic AI tools are probably the way forward given how good these have gotten over such a short amount of time even with this transition period being rough, but I do think in a way a lot of things were lost in the process.

If you had told me a year ago that AI would get to the point that you could write three sentences, put a screenshot up and something pretty good would come out that could be refined with like maybe two extra prompts, I would have laughed at you. But I think it is kind of there already now if what you are making is essentially just a CRUD app, and it is not looking like this will slow down whatsoever. A lot of people you are building the product for in a business setting also just do not really care as long as the result is good.

I don't think things that are made with AI are bad but a lot of these projects regress to the mean: meaningless explanatory all caps subtitles everywhere, overuse of purple gradients (I love purple but now I instantly think "AI" when I see it), three nested cards, a severe lack of attention to detail...
That does not mean people cannot make good things with it. In the case of protfx; who cares that its development is AI-assisted? People go there to play pugs, it does that job well. If someone cares enough and has enough knowledge they can draw from on how stuff should work, I don't see why people would be upset about that.

For technical people, I feel like a lot of the things that mattered a lot just matter less now, or rather they still matter but these models will work their way around a lot of the things for you instead so they are not really your concern anymore.
When I did work on making a V2 for the ETF2L API, I had to manually review several queries from the Wordpress codebase, make sense of serialized PHP objects to show those in proper ways in the output format.
Making something like charts.tf years ago (artisanal purple slop before AI was mainstream), I had to learn about how Steam OpenID Connect logins worked, study .zip files from logs.tf to find specific log entries that I wanted to show statistics for and parse them into a database, find ways to integrate with data from leagues (which ended up leading to building the V2 API). I did things like find some random archive on Reddit of Sizzlingstats and try to visualize what was in there, painstakingly try and find Steam profiles of TF2 players based on their ESEA alias with this site then organize them in a spreadsheet with the help of a friend. Or thinking of clever(er) ways to find specific logs of matches when dealing with edge cases (mercs, different map versions, people uploading logs days later if the server didn't upload...).
A lot of this stuff I found fun/challenging at the time and it was part of building it, some of these required programming knowledge, others were more about understanding data. In the end though, the main thing people look at is the numbers on those stats when it comes down to it. Most will not care about how things got to that point. In a way, that is a little sad to think about. I could now type "Claude, look at this Reddit post from 13 years ago, download that file from MediaFire and add this" and it would probably get the job done 70% and you could follow-up prompt the remaining 30% to get it to where you want. The way there has changed, the end result would probably not be hugely different depending on the complexity of the task.

There are many reasons to hate AI from the ecological side, prices for hardware being ludicrous, it being trained on complex hand-written work of others with no attribution... but unfortunately the models have gotten to the point where they are outperforming the average junior engineer. People will build things with it, some of it will be hideous and not get any attention, others will build cool projects or add features to existing ones that they would have never built because it would have taken too much time otherwise.

As someone who spent roughly 7 years without the clankers at a web dev job and started using these tools more over the past few months, these are my two cents. (Sorry for the nerd essay up front)

The slop that is not good will simply not get attention and people will move on. That does not mean everything that is made with AI is inherently bad.

I do think these agentic AI tools are probably the way forward given how good these have gotten over such a short amount of time even with this transition period being rough, but I do think in a way a lot of things were lost in the process.

If you had told me a year ago that AI would get to the point that you could write three sentences, put a screenshot up and something pretty good would come out that could be refined with like maybe two extra prompts, I would have laughed at you. But I think it is kind of there already now if what you are making is essentially just a CRUD app, and it is not looking like this will slow down whatsoever. A lot of people you are building the product for in a business setting also just do not really care as long as the result is good.

I don't think things that are made with AI are bad but a lot of these projects regress to the mean: meaningless explanatory all caps subtitles everywhere, overuse of purple gradients (I love purple but now I instantly think "AI" when I see it), three nested cards, a severe lack of attention to detail...
That does not mean people cannot make good things with it. In the case of protfx; who cares that its development is AI-assisted? People go there to play pugs, it does that job well. If someone cares enough and has enough knowledge they can draw from on how stuff should work, I don't see why people would be upset about that.

For technical people, I feel like a lot of the things that mattered a lot just matter less now, or rather they still matter but these models will work their way around a lot of the things for you instead so they are not really your concern anymore.
When I did work on making a V2 for the ETF2L API, I had to manually review several queries from the Wordpress codebase, make sense of serialized PHP objects to show those in proper ways in the output format.
Making something like charts.tf years ago (artisanal purple slop before AI was mainstream), I had to learn about how Steam OpenID Connect logins worked, study .zip files from logs.tf to find specific log entries that I wanted to show statistics for and parse them into a database, find ways to integrate with data from leagues (which ended up leading to building the V2 API). I did things like find some random archive on Reddit of [url=https://www.reddit.com/r/truetf2/comments/1ocfnp/data_mining_sizzling_stats_what_kinds_of_stats_do/]Sizzlingstats[/url] and try to visualize what was in there, painstakingly try and find Steam profiles of TF2 players based on their ESEA alias with [url=https://legit-proof.com]this site[/url] then organize them in a spreadsheet with the help of a friend. Or thinking of clever(er) ways to find specific logs of matches when dealing with edge cases (mercs, different map versions, people uploading logs days later if the server didn't upload...).
A lot of this stuff I found fun/challenging at the time and it was part of building it, some of these required programming knowledge, others were more about understanding data. In the end though, the main thing people look at is the numbers on those stats when it comes down to it. Most will not care about how things got to that point. In a way, that is a little sad to think about. I could now type "Claude, look at this Reddit post from 13 years ago, download that file from MediaFire and add this" and it would probably get the job done 70% and you could follow-up prompt the remaining 30% to get it to where you want. The way there has changed, the end result would probably not be hugely different depending on the complexity of the task.

There are many reasons to hate AI from the ecological side, prices for hardware being ludicrous, it being trained on complex hand-written work of others with no attribution... but unfortunately the models have gotten to the point where they are outperforming the average junior engineer. People will build things with it, some of it will be hideous and not get any attention, others will build cool projects or add features to existing ones that they would have never built because it would have taken too much time otherwise.
1 2 3
Please sign in through STEAM to post a comment.