The Blender AI Situation over the past month

General
, ,
1

Wanted to talk a bit about some AI related news to Blender recently, in which there are 4 main things that happened.

  1. Blender content creator Andrew “Blender Guru” Price is outed as an AI grifter
  2. Anthropic becomes a Corporate Patron of the Blender Foundation
  3. Anthropic launches Claude integration with Blender
  4. Blender revokes Anthropic’s Corporate Patron status

First, something lighthearted. Blender Guru’s been a bit of a controversial figure for a while in the Blender community, but his recent promotion of AI generated content has given people more reason to dislike him. The 3D artist twitter community spent about a week clowning on him, and retroactively criticizing some of his earlier work, like this one slide from a 2018 talk he did at Blender Con where he estimates the price of a traffic cone asset in production as costing $3,600, among other incredulous prices.

image
image1310×797 287 KB

Image Source: https://x.com/AndrewNameless/status/2045805676153205135?s=20

A lot of people responded by speedrunning traffic cone models, and eventually Blender Guru responded by setting his twitter profile picture to an image of him wearing a traffic cone.

image
image610×488 133 KB

The end result of all of this is that his reptuation has seemingly gotten much worse than it already was, and he has seemingly lost a lot of his good will and credibility as a 3D professional.

Then about three days ago, Blender announced Anthropic would be joining their development fund as a Corporate Patron, in which they’d be joining the likes of Wacom, Netflix Animation, NVIDIA, AMD, Intel, Meta, Adobe, and a bunch of other corporate partners (full list here). I honestly don’t know what the extent of a Corporate Patron is, whether they have power in what gets publicly updated to the software or pushed officially. Regardless, this received a lot of public pushback.

At the same time, Anthropic announced the ability to use Claude as a tool inside of Blender to perform a variety of tasks. In the video they showcase Claude applying textures to some rocks in a scene, and deleting and modifying objects. A lot of these batch operations have been possible through script automation in the past, so this isn’t anything new. It’s important to mention that Claude isn’t generating any meshes or textures with AI here. It’s just performing simple operations.

Whether or not Claude is actually useful is still up in the air. I’m not going to experiment with it personally, because the usecase honestly isnt that common in my work, and I cant see this saving much time for me that couldn’t be saved in easier and faster ways. A professional 3D artist George Crudo posted a thread on twitter where he tested the tools, and shows off its lack of competence. Crudo has a history of being critical against AI in 3D workflows, and has analyzed other AI tools similarly in the past, showing how ineffective they are as tools for artists.

Back to Blender. Today, they announced that they’d be revoking Anthropic’s Corporate Patron status and will instead accept Anthropic’s funds as a donation rather than as a Corporate Patron. It sounds like the Blender Foundation will now be more public on their decisions to interact with any AI company, but it does not seem to be ruled out entirely. Despite this incident, their Corporate Patrons page still lists Meshy.ai, an AI 3D Model Generation platform. So whether or not Blender will work with AI in the future is still up in the air.

(UPDATE: MAY 3RD 2026 - MeshyAI has been removed from Blender’s list of patrons)

My personal opinion is that AI as an Agent tool to perform tasks in Blender doesn’t sound so bad on the surface. Thereotically this would mean no generative content, no asset creation. Maybe it could be useful to perform batch operations using human language rather than script, but the cost of using such a tool kind of drowns out the usefulness in my opinion. Even if it were effective, which it is seemingly not, its only providing something you can already do for free, and at that point it feels like you’re throwing away money. By using Claude you would be restricted by their membership models as well. This isn’t even getting into the environment concerns of supporting LLMs or the political concerns of supporting anything related to Anthropic, given their history with the US Government even if that is now seemingly terminated.

I haven’t used any agentic AI tools, so I only hear the stories of AI destroying people’s entire repos and going rogue. It’s hard to take all these stories seriously though, with how many fake posts are made where an AI is prompted to “act like it’s going rogue”, or the fake moltbook situation earlier this year.

I try to take a pragmatic stance towards AI. If the tools can genuinely be used as tools and are effective and useful, then I’m willing to humor it. I’m strongly against image generation and video generation, as I believe the cost to be dramatically more expensive than text and am concerned of the environment impacts it would leave with rising power costs, and also the negative outcomes it could have for society at large, such as misinformation and harassment.

Despite this I personally see no real use in a tool like what Claude advertises. Even if it was effective, this provides very little use in my day to day workflow in Blender. Maybe that’s due to my specialization as a Character Artist, maybe environment artists or someone working on a movie would find power here. Even if it worked, even if cost wasnt a concern, and even if the political reasons to never engage with AI were ignored, it seems pretty useless.

Somewhat interestingly, Unity I believe was investing in a similar AI tool that would allow you to prompt an AI to spawn gameobjects and place them around the environment. I dont know how far that really went, I think I saw this roughly 2-3 years ago, but it didn’t seem that useful then either.

12 Likes
2

To be honest, I kinda stopped paying attention to him since he was on the last big bandwagon, crypto. I just find him a way too gullible person, to be honest.

I do really think it’s important to push back against AI in creativity (code included). It seriously is going to create a huge issue in the future legally in addition to increased cost once these models are not subsidized anymore. That’s leaving out all the ethical issues…

8 Likes
3

update! meshy.ai is no longer listed as a corporate patron either. so theyre being consistent which is nice. curious to see how strong this stance on AI will stand in the longterm.

(Before)
https://x.com/slavoartist/status/2050576251618738582?s=20

(After)
https://x.com/IsThisA3DModel/status/2050715986597163096?s=20

6 Likes
4

It seems like Blender Foundation is taking the issue seriously, which is good to see.

4 Likes
5

I’m very much opposed to corporate control of AI (and the implicit goal of replacing workers that comes with corporate ownership).

Ultimately I think that the desire to make assets 100% with AI comes from the commodification of art. If art is a product where the goal is to ship and make money, then using AI makes sense. If the goal of art is to encode the user’s intent, and express ourselves in a way that words can’t, then AI is inherently self-defeating because there’s no intent behind it.

That said, I also try to take a pragmatic view and use these tools to improve my quality of life, but without robbing my agency.

I wanted to create a non-slip industrial grating material and then bake it as a texture. I would have been totally lost, because I don’t know the first thing about writing shaders. I basically explained what I wanted to the best of my ability to Claude, what shader to parameters to expose. It wrote a Python script to generate the material and then another to bake it to a texture. To a 3D artist this is probably trivial, but for me it was a blocker that would have taken hours to figure out myself.

image
image760×670 73.9 KB

Another thing I could see potentially being useful is cleaning up 3D models generated from photogrammetry. I have some retro hardware I scanned using RealityScan on my phone, as well as a $2,000 3D scanner. The phone scan is actually better overall (no giant gaps), but it is much noisier. I have access to an RTX 4090 and might spend a day playing around with some cleanup tools.

6

So, in other words, you:

  • robbed yourself of the opportunity to learn;
  • used a corporate-owned service to non-consensually copy (i.e. steal) other people’s work, while pretending it was doing the work itself, and obfuscating the origin of said work;
  • didn’t actually create anything, missing your own stated goal;
  • could’ve searched for an existing script (with explicit consent for copying, and authorship information), searched for a tutorial, or asked a person for guidance, but decided to replace the humans instead.

You’re not really being nuanced about it; you’re constructing a narrative to make it sound less frivolous, and contradicting your own stated views in the process. It’s not dissimilar to people with harmful drug addictions justifying their continued use of the drug as something beneficial and more important than an addiction problem.

Don’t take it personally - I also don’t judge people negatively just because they have an addiction problem. Rather, I consider the justifications to be a broadly harmful point of view to hold publicly.

7

I think if you want to convince others of your argument you have to tone down the contempt.

8

I think I have an appropriate level of contempt for the practice you’re endorsing, given its consequences. I did not intend contempt towards you as a person, tho I did intent a strong wording to underline a rejection of your endorsement, and flaws in your argument, for other readers. That being said, you seem generally uninterested in any opposing points of view, so I’m not sure what else you’d like me to respond with.

9

I’m on these forums because I like having discussions. It kinda feels like you weren’t interested in having a discussion, you just see me as an “AI chud” (my words not yours) and an easy dunk. If the goal is to convince me to change my behavior or opinions I think there’s better ways to go about it.

you seem generally uninterested in any opposing points of view

I don’t know why you think this. I’ve never interacted with you before outside of like a paragraph-long forum post.

Idk, this is just kind of an insane way to treat a stranger on what is supposed to be a friendly forum. If you actually want to have a discussion with me about why I should change my POV, I’m down, but I usually try not to engage with people who are rude.

10

The community circle is still writing a code of conduct, under which the topic of AI is specifically addressed. But even so, our current guidelines ask of our users to:

@oufrost: In this conversation, you have claimed Seabass has something akin to a harmful drug addiction. This is an ad-hominem.

You also openly agreed that you wanted to display contempt for what they are doing with the intention of not actually engaging Seabass in conversation. This does not improve the discussion.

I want the forum to be a place where others feel comfortable having discussions, even for contentious topics like AI. This behavior does not help foster the positive, welcoming community we are trying to create. Please follow the guidelines the forum is currently running under.

4 Likes
11

I’m personally not confident in the ability of AI to provide useful meshes. I haven’t done any experiments with AI generated models personally though. The results of what George Crudo did (as mentioned in the OP) and some others I’ve seen in passing, it has never provided a practical product. These generators do not stand up to real world workflows.

You do not want to be condemned to the fate of the Yandere Dev Tooth Brush. That is effectively the direction you end up when you use 3d model generation tools.

That being said, AI-Powered tools, such as retopology ones, can definitely assist in cleaning up models from sculpts or 3D scans. Retopology here is still a manual process and not a one-click solution, but from my understanding there are AI tools that make the tedious aspects such as filling, much faster without the same overhead a full on AI model or generation tool incurs. And also without prompting. Heres an example of one that looks fairly impressive: New AI Tool To Automate Creation Of Animation-Ready Retopology

I’m not a big fan of this workflow though. If you’re a solo developer looking to make 3D models yourself, I would personally avoid sculpting/retopology/3D scans for 90% of your work. You will often times be able to make these things much faster if you make the model directly, using vertex modeling techniques. This mostly goes for props. Most solo developers cannot warrant spending over an hour for common props such as a computer monitor that’s just on a desk in the background.

For more complex objects that demand a lot of focus though, it could be worth the extra hassle. But I try to minimize the number of different workflows I use so I can instead be faster and more efficient. Its kind of like cooking. You could totally buy a bunch of kitchen gadgets. But almost always it is much faster to just use a knife, and you’ll have a lot less to clean up afterwards.

1 Like
12

I did a bit of research. Surprisingly as of today there’s no turnkey AI retopology tools. I saw something from Nvidia but it was mostly experimental and required a ton of VRAM. I ended up just doing some manual cleanup and then passing it through a traditional CPU-based remesher, but the results weren’t great, it just distributed triangles uniformly throughout the model.

I agree, I just wanted to give it a try. I thought it would be neat to have a scene with more high-fidelity assets for the main menu, but I couldn’t get 3D scans to look good, so I just bought an asset pack, because this kind of modeling is out of my depth. I also realized that whatever your highest-fidelity asset is, everything else has to match that level of quality, otherwise it looks out of place.

I also have found that I don’t really enjoy 3D modeling in Blender. There’s a lot of complexity that I’m just not interested in learning about. I think in the future I will stick to an art style that lets me use tools that a baby could understand like: https://www.blockbench.net/

This is another thing that I learned. For my pre-rendered UI I used FreeCAD (which means terrible geometry), Blender, and then Krita for cleanup. Having so many stages made it really inconvenient. I know now that I probably want a single streamlined workflow for every asset type.

1 Like
13

Thanks. The internet is full of volatile responses that escalate from 0-100 and it’s something that I am also trying to unlearn. I have limited the places where I post and interact specifically so I can get away from unproductive discourse.

We are also all in the process of learning and so I try to give people grace (if they are acting in good faith) because people usually don’t change their opinions from moral arguments, I have found that people usually only change their opinions through learned experience.

@outfrost I do think you raise some valid points. I think AI usage is problematic in the same way that eating meat, driving a car, buying things from Amazon, having a smartphone, supporting the US military, and owning stocks are problematic. But in order to exist in this world without constantly self-flagellating or feeling guilty, you have to figure out how to coexist with all this stuff. Some things are hard red lines for me, others are not (I will never directly or indirectly support the military, however sometimes I do eat meat and order stuff from Amazon).

I would agree that this is a form of rationalization/compartmentalization, but unless I join an Amish community I’m going to have to rationalize my actions or go insane.

Also I would argue that most of us have some kind of addiction that we rationalize. Maybe not the clinical definition of addiction, but in the sense that there is some negative ritual we participate in because it provides us temporary relief. That’s also the nature of living in the 21st century.

1 Like
14

I suppose the counterargument to that is that anything you can do with generative AI you can do better by not using AI. Furthermore, it’s important to recognize that AI is heavily subsidized. Even non-excessive use of coding assistants can land you in the hole for thousands of dollars a month.

In short, I think it’s much easier to avoid than buying things from Amazon, having a smartphone, etc. You have to go out of your way (and dig deep in your pocket) to use it. I’ve managed to get this far without using generative AI in anything I do. That is not to brag, but more to emphasize that it was quite straightforward to avoid.

1 Like
15

Claimed, no. Implied, more willing to agree on that. But if that’s an ad hominem, doesn’t that mean we treat harmful drug addictions as failings of character? I don’t, and I wish we didn’t, but I take from this that I didn’t communicate that part well.

I see now that I should be more mindful of what kind of conversation one can expect to come out of how I say things. Building the correct tone to convey my thoughts has been a lifelong challenge, and it’s been kind of amplified with how difficult it’s getting to be critical of genai on a day-to-day basis. Not to say this justifies my words or anything, it’s more of a… introspection, I suppose.

16

edit: This is a reply to @Seabass. Discourse is being weird about displaying that.

Sorry about my initial tone. Re-reading back through what I said, it wasn’t going where I would’ve wanted it to, nor was it necessary.

I feel that, out there, in the world at large, it’s been getting difficult to have a principled discussion about and be critical of genai, and some part of it has been the ages-old excuse “oh but if I use it just a little bit it’s fine”. Altogether it made me approach the subject with exasperation, which was very much misplaced here.

I remember an interaction in discord, where you mentioned your use of genai, I pointed out problems with it and/or flaws in reasoning, and you replied with something along the lines of “I’ve had this discussion, I don’t want to have it again”. That’s where I got that sentiment. It now seems to me that that is no longer the case, or was something I misunderstood, and I’ve leaned too strongly into assumptions based on that.

I don’t agree with the comparison. I don’t think the things you list are inherently problematic (well, aside from direct support for US military…) from the perspective of an individual, even if they have (many) problematic use cases. They are very often means to a benign, well-justified end, and it is the system’s and powerful people’s fault that their use creates or perpetuates problems, and that better alternatives aren’t viable.

Generative ai, on the other hand, doesn’t solve any real problem, is completely unnecessary in achieving our well-justified goals, we aren’t forced to use it (outside toxic employer environments anyway), and we’ve been living with viable better alternatives all along. It is also the system’s fault that genai is allowed and incentivised to exist in such a harmful way, but it is unique in both how harmful it is to our cultural and intellectual heritage (and by extension to all other aspects of our world), and how much say we currently have in opposing it.

That’s why I’ve likened it to harmful drug addictions. The drug seems enticing as a solution to (or escape from) our problems, but it won’t in fact solve anything, and it will take a hefty toll on us and those around us if we keep using it. We should probably say “no” before we become dependent on it, and we should probably help those already dependent get out of it.

17

It’s no problem. I get where you’re coming from, I genuinely do. I have wrestled with the ethical conundrum for a while and I came to my current conclusions after thinking about the nature of work, what are we actually gaining vs losing, etc.

Oh ok I remember this. At that time multiple communities I am in had been having AI discussions ad-nauseam and I just didn’t care to continue to engage in them.

The attitude I had at the time was that nothing I can say really matters in the grand scheme of things. I don’t mean that as an excuse to abdicate myself from personal responsibility, but going back to the veganism example, market forces are probably going to do more work to get people off of meat than decades of animal rights activism. Climate change and environmental degradation, animal-borne disease jumping to humans, the resurgence of parasites like screw worm and ticks, these are going to reduce meat consumption more than animal rights activism. That’s not to say that kind of work isn’t important (I myself am pretty politically active) but the way I have to frame things to prevent myself from going insane about my lack of control over large-scale systems, is to focus on what I can control, my own actions and my own immediate environment.

(Similarly I would argue that the corruption of the US military industrial complex is more responsible for defanging the US military than any anti-war movement since the country’s inception. The system is its’ own gravedigger to paraphrase a famous economist)

To fully explain my POV at the time, I would have had to type out a giant response and I just didn’t want to do that at the time. So here is that response, but very delayed.

We could be a really long discussion about what AI is (is it a giant IP theft machine that often copies stuff verbatim? (yes) Does it make you dumber? (yes)) but I think that we actually agree on a lot about the nature of AI. The source of conflict is a divergence what we think is the best way to mitigate the negative effects of AI. I personally think that the choices aren’t a world with zero “AI” usage and maximalist “AI” usage, but I think the choices are a world where AI is run in giant datacenters that require enormous amounts of capital, and a world where local models are run by individuals or small organizations on hardware they control. I don’t think anything I personally do will change the outcome of the conflict between the two worlds. It seems like it is just going to be a function of competition between China and the US, and whatever physical rules determine how “smart” of a model you can fit on a consumer GPU.

Because I can’t control the AI war any more than I can control a real war, my relationship with this technology ultimately boils down to how it affects me, personally.

(Also, I think an argument could be made about how paying a $20 monthly subscription enables the company to exist and creates all of the negative externalities like building datacenters in poor neighborhoods, etc. And I think that’s also true, but generally that argument can be made for participating in the economy at large.)

Generative ai, on the other hand, doesn’t solve any real problem

I see what you’re saying, there are always alternatives, but I think saving time is solving a problem in itself.

My strongest use for AI is writing bash scripts and powershell scripts. It’s not something I care about, and so before LLMs were a thing, I would basically just cobble together random bits and pieces from Stackoverflow until something worked. I cannot absorb information if it’s something that I’m not fundamentally interested in. I still have to look up git commands because I just don’t care about how it works. The information leaves my brain as soon as the task is accomplished. That probably puts me at odds with many people here, who I think enjoy tinkering with systems for the sake of it, but we’re all different.

I understand if someone has the response “if you’re just cobbling together code like a script kiddie then you have no place writing bash scripts that your company relies on” and that’s probably true (lol), but “ought” and “is” are usually not the same thing.

So for me personally, with this project I realized that Blender is too large and too complex of a program for me. I need to work in something simpler that lets me focus on making the thing I want to make, and spending less time learning these very deep tools. Knowing that, I know I am not interested in learning how the visual shader node system works. I just want to finish the grate material so I can finish the project so I can move on to something else. Part of the goal of this project was to make something with the goal of finishing it, because I knew I would learn a lot from that (not just technically, but like, how does my relationship with a hobby change when my goals change) and I realized that actually it makes the experience completely miserable at times.

To that end, any future projects are just going to be for fun. Doing whatever I want at my own pace, with no pressure to finish, because that feels the most enjoyable and freeing to me. I don’t feel the pull of an optimization machine saying “hey, use me as a shortcut!”. I dislike that feeling and want to distance myself from it as much as possible. I probably won’t even use Godot because Godot also abstracts away the fun parts for me. I could have used something else for this project, but once you tell yourself you have to finish, you become an optimizer, just reaching for whatever finishes the job the quickest. At least that’s how my brain works, personally.

I brought up the smartphone as an example because I think that my relationship with my phone is much more negative than my relationship with AI. I moved to GrapheneOS and don’t have any social media really, but still the phone manages to be a huge attention sink. I can’t really ditch it though because so many things in the physical world require an app to use.

As far as the addiction comparison goes, you’re not wrong (although, as an ADHD-haver, the addiction thing kind of falls flat for me because I literally can’t get anything done without stimulants, granted I stick to caffeine and exercise because Adderall and nicotine are way worse for your heart).

Much like real drug addiction, I think it is different for everyone. I have seem some people (including very brilliant senior engineers) get sucked into AI coding and burn literally tens of thousands of dollars worth of tokens and then get chewed up and spit out. I have seen people try it and very quickly say “this isn’t for me”. I have also seen people get sucked in and haven’t had to pay the piper yet, so they keep hitting the vibe coding slot machine. Like with real drug addicts, I don’t think any of these people could be convinced by argumentation. I think they will have to essentially hit rock bottom, and then go to “AI rehab” (writing code by hand again I guess?).

1 Like
18

The problems with LLMs don’t go away just because the model is smaller and run on a different computer. Sure, it may no longer be directly controlled by the oligarchs, but most everything else persists - how training data is sourced, who has control over the training data, how much energy is used for training and runtime, how inefficient, incorrect and unreliable it is compared to normal tools and resources, how much it deskills, how negatively it affects human knowledge, work, culture, etc.

And I don’t think we’ll get a “zero ai” world in any reasonable timeframe either, but we have to show up to the negotiating table with demands that favour us - otherwise we wouldn’t be negotiating a compromise, we’d be capitulating.

I think we all have some level of responsibility for shaping its cultural perception. We also can’t control whether homophobia exists, but we sure shouldn’t be laissez faire about homophobic behaviour.

What’s the problem with continuing to do that? I know we (as a profession) used to meme on copypasting from stackoverflow, but that’s infinitely better than LLMs.

Of course, stackoverflow is now actively becoming an ai cesspit, and trying to search the internet for anything is a nightmare, completely on purpose and orchestrated by ai companies, but you can still do it. For now at least.

Tbh I would infinitely rather you look like a “script kiddie” and cobble it together yourself, than take whatever the impostor machine spits out. It breaks in more predictable ways, you know what you did and didn’t account for, less wasted heat, fewer data protection risks, etc etc.

Was the grate material necessary in order to finish the project? But more importantly: if you make projects with the goal of finishing them, doesn’t the use of ai kind of ruin the notion of finishing something? (Forgive me if you’re basically saying the same thing, I’m not trying to be contrarian.)

Yeah, I do distinguish harmful ones for that reason. I don’t think it’s inherently bad to be addicted to caffeine and such, and having to take addictive drugs in order to get around another condition that impairs your functioning is of course easily justifiable, at least in the absence of efficacious alternatives. What I mean by harmful drug addictions I’m comparing LLMs to, is obligatory use of substances which are not necessary, and whose long-term effects are (almost) entirely destructive, e.g. cocaine or nicotine. I don’t fault any individual for having a harmful addiction, but it is important to recognise it as such, try to prevent it from happening, and avoid deluding oneself about its benefits.

19

My coffee is wearing off so my bad if this is incoherent. Also when I say gen AI I’m talking exclusively about coding agents. I think using it for images or video is pointless due to the nature of art.

Not to be like, overly-doomer but I don’t think there’s any meaningful negotiating that can be done. Like, the fate of “AI” is just going to be dictated by market forces. If it truly isn’t useful, eventually the accumulated tech debt will cause everyone to abandon it. We’ve already kind of seen that earlier this year-- I believe it was IBM that went back to hiring juniors because you still need to train “human capital” and can’t just rely on AI because you have an ouroboros situation on your hands.

What’s the problem with continuing to do that? I know we (as a profession) used to meme on copypasting from stackoverflow, but that’s infinitely better than LLMs.

It writes better quality bash scripts than me like 10x faster than I could. It feels wrong to say but that’s the unfortunate truth.

I guess this is the crux of my argument. If you could go back to 1910 and tell everyone how cars will pretty much destroy the environment (both the climate and also our lived environment) they would just keep using cars because it was convenient. Anyone who refused to use a car would eventually get overtaken because of the efficiency feedback loop. The excess mortality and environmental degradation that cars introduce is very real, but that’s never going to stop people from using them-- even before we lived in an society that required a car.

You couldn’t trick people into buying cars if there wasn’t some immediately tangible benefit to them, for the same reason I think you can’t “trick” people into using AI no matter how much marketing you pour into it. People (or corporations) either have some sort of immediate benefit or they don’t. So I don’t think there’s much use in fighting a war over the narrative, because people will just make up their own minds about whether it has utility to them.

It’s not the same as homophobia or veganism where there’s no real utility to being homophobic or eating meat (one could argue there’s like an in-group signaling thing going on but you get what I mean). You can convince someone through argumentation that a moral position is wrong, as long as the moral position is abstract and not tied to some economic thing. You can convince a worker that the police state is a bad idea and against their best interests, you’re not going to be able to convince business owners of that perspective because that system is literally built for them.

Because of that, if people see some kind of utility and are never going to quit of their own accord, I think that the solution is local models trained on an open dataset, running on efficient hardware. There’s a lot of prerequisites there obviously, but this has generally been the trend with computing so idk.

I think we all have some level of responsibility for shaping its cultural perception. We also can’t control whether homophobia exists, but we sure shouldn’t be laissez faire about homophobic behaviour.

This is a fair criticism and I can just shut up about it rather than give free advertising. My point was not “OMG LLMs are epic!” more like “hey this meteor came and killed 99% of life on earth-- I have found out that I can make pottery with the ejecta though so that’s nice”. Where I’m coming from I guess requires the reader to accept the prerequisite that the meteor (AI) hit and killed all the dinosaurs (enjoyable tech jobs) and they are never coming back. I think that premise is itself controversial, but every software developer I know is saying the same story-- work is now just managing agents, and it’s fucking miserable/amazing, depending on the invidivual. I was reading the creator of Rust’s blog yesterday and even he was accepting it with defeat: https://graydon2.dreamwidth.org/322732.html.

Was the grate material necessary in order to finish the project? But more importantly: if you make projects with the goal of finishing them, doesn’t the use of ai kind of ruin the notion of finishing something? (Forgive me if you’re basically saying the same thing, I’m not trying to be contrarian.)

Yeah that’s what I’m saying. I enjoy the process of making things. Starting something with the goal of finishing rather than the goal of enjoying making the thing, basically ruins it for me. AI or not, I will take the path of least resistance when my goal is a destination rather than a journey/ritual.

Anyways I’m not trying to convince anyone of anything. I find debate exhausting because I end up having to defend/attack some position that ultimately doesn’t matter. I could write a million words denouncing AI as a great evil and a million words making Linkedin posts about how it is revolutionizing b2b saas and it wouldn’t change anything. It just feels like a waste of my time. I will capitulate to whatever your points are because I just find it so tiresome. Every community I am in is just a nonstop debate about this stuff and I just wish it had never come into existence, or I could hit fast-forward on its timeline to the point where everyone ditches it, or it just consumes software development in totality and I can switch to being a high school teacher or something.

Really I think the solution to all of this would have been for software developers to unionize in the 2010s when they had power, but you don’t unionize when things are going good. By the time people start unionizing, they’re usually in a dire position. So now they’re at the mercy of their bosses and the bosses see number go up.

I guess the point of my original comment was that I just wanted to unpack how this whole thing has made me feel and how to process my relationship to work and craft-- and highlight a silver lining, which is really more personal than anything else and I can always just keep that to myself.

EDIT: also, can I just say how fucked up it is that now when people see em-dashes they automatically assume the output is AI generated. I have always just put commans and em-dashes everywhere (to the chagrin of my english teachers) and now they have been co-opted. I think my stream of consciousness writing style is distinctly not AI thankfully.

20

I think programming is just as much an art as the visual arts, it is just a programmatic art, and I enjoy the process of figuring things out and learning new patterns and gotchas and such. Though for some it might just be a means to an end and they might be looking at it for the utility versus the creative endeavor. I think this varies from person to person and moment to moment in the same way that some might treat a car like a simple object of transport and others might treat it like a work of art and get special parts or restore and old car and such; or how someone might cook food just to make a meal to eat and someone else might get into culinary arts to make something special or to enjoy the process.

I suspect a lot of disagreement might come from those kinds of differences, where sometimes someone just wants the end result and doesn’t need or want to explore the process itself. And this isn’t set in stone per person either, I think someone’s view might vary per project or per moment depending on need or interest! Someone who loves cars might have an everyday car they treat like a utility and then a project car that they dote on and dream about.

Just my 2c.

2 Likes