World population and our standard of living have improved drastically over those years too, we aren’t burning that additional energy for nothing.

I don’t think energy use is a serious problem, that just seems to get thrown around just because it’s trendy. Does it even matter compared to gaming or crypto? It’s also an easily solved problem, just install more solar. Training the initial model isn’t time critical or depended on location, so there is a lot of flexibility here that you wouldn’t have in other applications. Meanwhile running the already trained model is very cheap, it’s literally the most efficient way to solve the problem. Trying to replicate what StableDiffusion is doing with a 3D renderer and you’d need to burn a heck of a lot more cycles, as well as hire a truckload of artists, which would all use substantially more energy.

Basically, people are going to use AI when it makes better use of time/money/energy than the competition. Nobody is going to use AI to burn energy just for the fun of it, it has to improve on what we already have.

As for the concentration of power and wealth, that can certainly happen to some degree, but I could also easily see that get balanced out by the amount of freedom that local models give. Right now I can generate subtitles for video with Whisper, generate voices with tortoise-tts, generate images with StableDiffusion as well as play around with LLMs on my local machine with OpenSource’ish models. Nobody controls what I do and I am not paying for anything. There are obviously still aspects that those models can’t do, local LLMs aren’t up to GPT-4, but already quite close to ChatGPT for some tasks, StableDiffusion isn’t quite as good as Midjourney for plain txt2img, but state-of-the-art in a lot of other aspects (custom training, ControlNet, LORA, etc.). But for a lot of tasks those models are already “good enough” and they are constantly getting better. Meanwhile ChatGPT or BingChat are so heavily censored that they flat out just don’t work for a lot of task, even seemingly simple things like summarizing movies (too much violence). Nobody even talks about DALL-E2 anymore, due to being surpassed by everything else out there.

Now centralization can still happen, Google is sitting on more data than everybody and if they make some multi-modal model that is trained on it all, that could be a very potent offering. But for the time being at least, everything that was released was outclassed by another thing within a few months. Nothing in the AI space so far lasts very long and the fact that AI models can use other AI models to improve themselves, hopefully makes that continue for a while. With the censorship going on I also have a hard time seeing local models disappearing anytime soon, as so far none of the commercial offerings had the balls to just build a model that knows everything.

The issue isn’t that the competition doesn’t offer similar functionality, but that there is no way to move your data to another hoster. If you move CI, you have to rewrite it as everybody uses a different language. If you move pull requests, you lose contact with all the users that made those pull requests, as Github doesn’t allow PMs and doesn’t publish emails by default.

I can move a Git repository in a single line, I can even mirror it to multiple hosts at the same time with ease. With all the surrounding aspects of a project that isn’t possible.

Though worth pointing out that this isn’t a GitHub specific problem, all software hosting suffers from this. Moving data between different Open Source bug tracker ain’t exactly easy either. There aren’t very many tools that are properly distributed in the way Git is and the few that there are, don’t seem to have very wide adoption (e.g. git-bug).

For example, if I burn a copy of your Disney DVD, watch it, and then write a review, then I’ve violated copyright. The review doesn’t violate copyright, but the DVD I burned does. Even if I throw away my DVD after publishing my review.

1. No, you haven’t. Making private copies is completely fine under copyright, that was decided back in the VHS days. You might violate DMCA, but that’s not an issue for AI, as everything they were trained on was publicly available and unencrypted.

2. When you download from the Internet, the server makes the copy, not you.

3. Your review still didn’t violate copyright. You are even free to include some images of the movie in your review under Fair Use, as long as they are small and insubstantial enough to not stop people from seeing the movie.

4. Watch any art tutorial, step one is gathering reference images from the Internet. If that would violate copyright, than a lot of artists would be in big trouble.

Copyright is about specific works, not vague ideas or styles. You can’t just claim that work X violated the copyright of work Y because it has some similarities and competes with it in the market place. You have to show that work X copied substantial parts of Y. And with AI models that’s going to be difficult, as the average image contributes about a single byte of information to the model. When the model was properly trained, without excessive duplicates, there is no way to get back to the original image from the model (some exceptions do exist here, e.g. Mona Lisa).

It also worth pointing out that in all these months of discussion, nobody ever managed to show a single image that the AI model would violate the copyright of. If AI stole your stuff, it shouldn’t be that hard to find some evidence for that.

I wonder if this can really be stopped.

We already have Generative Fill in Photoshop. This “fight” was over before it even started, AI will be everywhere going forward. And that’s fine.

The issue is more how it will be used. It’s very easy to create lackluster art with AI and it can be insanely frustrating trying to get specific and consistent art with AI. But at the same time AI is way better at a lot of things than human artists, textures, lighting, etc. AI nails that every time and humans artists don’t. People often make the mistake of comparing the best of the best of human artists with some average AI image, when the reality of most human art production isn’t exactly the best of the best. Most human art is extremely lackluster, even mainstream 100 million Hollywood movies have posters and box art full of anatomy mistakes and bad copy&paste.

If AI art is properly directed by a human, you can do great stuff with it and create much better work than just the human by itself. But that requires quite a bit more effort than just typing some text into a prompt and picking the first image.

As always with art: Process doesn’t really matter, only the result does.

Nothing too dramatic yet, but a lot of features GitHub provides are GitHub specific, not Git, which creates a lock-in and dependency that will cause problems sooner or later and make moving difficult.

The big and growing issue is that too much functionality is in GitHub and not in Git itself. So while you can move or mirror your repository very easily, moving your issue tracker or pending pull requests is a lot harder and comes with huge loss of information (e.g. there is no way to contact the submitter of a bug report, as all you get is a GitHub username, not email and GitHub doesn’t even offer PMs).

That said, I’d happily ditch GitHub for anything more distributed, e.g. hosting Git repositories on IPFS, integration with git-bug, etc. You can mostly DIY that today, but a hoster that provides some free storage would be very much welcome to help with availability.

Another more basic thing I am missing today is a redirect service for repository names , having https://github.com/User/Project.git spread all over the build files makes it hard to move hosters or provide backup repositories. GNU Guix has mirror:// to solve that, but that’s about the only place I can think of with mirroring build in.