grumbel
Photography never was. It has been copy&pasted together from multiple exposures, different photos and such in the darkroom for well over a hundred years. And once digital became common place the manipulation went completely into overdrive. The photo you see has little to do with the photo that the camera captured.
If we cared about fully deterministic captures, we'd all be using Lytro lightfield cameras, stereo cameras or 360 cameras or other contraptions that capture more light information than the classic camera.
Code on the other side is just text files, if AI writes them or a human makes no difference. You still end up with the same old files. At least for me, AI has long crossed the point where I feel like I can contribute anything useful. The AI understands my own programs better then myself and is far faster and more flexible than I could ever hope to get anywhere near.
My attempts at discussing AI related questions with AI itself have been pretty frustrating, especially in the early years, since they are either just uninformed about what's currently going on in the AI space or recite random blog posts of questionable value, often they couldn't even tell what they themselves are capable of.
I'd start with the basics: A scanf() replacement would be nice, having a ostream/printf() alternative, but not something for input feels very unfinished.
Basic Unicode support is also still a mess.
Another thing I don't get: Why is uninitialized memory still the default? Why haven't they made the uninitialized state explicit in any of the numerous language updates?
Yes, as Nostr and Freenet disconnect the storage location from the document identifier. Thus your link keeps working, even if the servers change.
The problem with the Web isn't a lack of cheap storage or people willing to mirror content, it's that none of that matters when the links keep breaking anyway and nobody can find the content anymore.
For how to do it right, see Usenet, the messages are held together by unique message-ids and the protocol was designed for distribution. Thus archives could be put together from random backup tapes and even 45 years later, you can browse and read the discussion threads.
Your average web forum just disappears when something goes wrong and even if somebody has mirrored the HTML, they'd have a rather bad time trying to recreate the original database, since the HTML is really more like screenshots of a GUI than the actual content.
> The web is just the library but on the web
Many countries have a legal requirement that every book published has to be send to the national library. We also have ISBNs that uniquely identify books. We have nothing like that in place for websites.
> they don't disappear quite so easily.
A local copy doesn't preserve it on the Web. Even if you wanted to preserve it on the Web, you can't, since the original domain belongs to somebody else.
Yes, but badly. Any old Amiga game that requires you to jump with Up or requires keyboard keys for certain functions becomes unplayable on a gamepad in an emulator. When you throw the game into the AI and do HLE emulation, you can fix it all with ease, and you can do it all context depend, e.g. a platformer level might need different controls from the spaceship levels or the menus. You can also add on-screen menus when needed.
For console games this isn't generally much of an issue, they controls map pretty well to modern gamepads and play well in emulation. But home computer games play absolutely terrible in emulation, as the controls are all over the place and each game has their own little quirks.
One of the big problems with the Web is that it's all ephemeral, it not a question of if it breaks, but when and how badly. A book doesn't disappear from your shelves just because the author lost interesting or the publisher goes out of business, but websites disappear from your bookmarks like that all the time. That's what needs fixing.
The other part is HTML itself. You shouldn't need to write raw HTML to write a website. That's the job of the browser. The very first browser that Tim Berners-Lee wrote supported editing, but that got completely lost over the years. A little bit still exist in the form of the 'contentEditable' attribute, but you can't author websites with that all by yourself, as there is no means of saving or publishing the results. No modern browser supports WebDAV either.
Even those, which take about 20min, are incredible useful, as you can add save state, level select, cheats and gamepad controls on top that most emulators would struggle with or at least make a hell of a lot more complicated.
> Now this garbage can be mass-produced and get the adoration of the AI hype crowd
Those games wouldn't exist without AI. Decompiles that ended in a fully playable state have been incredible rare before AI. Best you would get was often just an incomplete reverse engineering of the data formats involved.
As long as you need to rent servers and domain names all those alternatives are a waste of time, as they are just repeating all the missteps from HTTP. We need an alternative where you can type some text, hit publish and have it appear on the net with zero friction and in a way that isn't tied to a single third party. Publishing a site needs to be as easy as browsing one.
So far Nostr it's the only thing that gets close at the protocol level, but still lacks a good client.
When I see a picture on a computer screen, it evokes similar emotions as seeing the real thing, since the light that ends up on my retina is very similar, even so other aspects of the sensory input might be missing.
Music on the other side doesn't really have a natural analog. Nature is mostly just wind and water noises, with a bit of bird song and chirping crickets. The only thing they have in common with music is that they are all sounds, but the structure and complexity is quite different.
Music as an offshoot from pre-language communication seems more likely, it could explain the emotional component at least, since groups of animals might need to sync their emotional state. But even there you end up with quite a big leap between what sounds happen naturally and what we call music.
The hard part is really just the coordination, you don't want dozens of repos doing the same thing, but at the same time dealing with random vibe coded PRs is an easy way to overwhelm and burn out any maintainer.
AI is out there in the wild. It doesn't matter if OpenAI can get their stuff under control. There will be plenty others that don't. And there will be plenty of others that deliberately use their AIs for malicious purposes.
You'd think there would be an audience for using the TV as a big video phone, but that never happened. One could blame it on consumer trust, but that has never stopped them from adopting all other kinds of questionable things.
Maybe I watched too much sci-fi, but using a big screen to make video calls seems like such a natural extension for a TV.
A hologram is much like a window, you can look through a big window or a small window, and see what's at the other side both times, but the big window still gives you a lot more viewing angles than the small one. Cutting a hologram in half gives you half a hologram, not two copies of the hologram, it's just that you can change your viewing angle to still see what's on the other side, but that will be a different image than with the full hologram.
And those are like vision, which LLMs are pretty good at. It's all just sensory information, 1s and 0s going into the network, which in turn change the behavior of the network in some way. Furthermore LLMs have demonstrated some form of vision even before we started training them on actual images, as could be seen by the crude vector graphics they were able to produce.
Language is the I/O of the network, not the network itself. A human doesn't stop feeling pain just because we locked them in a box and reduced their interaction with the outside world to teletype. We simply don't know what's going on in the AI box.
> Maybe there's some future embodied AI
That changes nothing about the network itself. That just changes your perception of the network into something more familiar.
There is nothing "clear" about this. All we know is that the neural nets arrived where they are today, not by billion years of evolution, but by getting training on artifacts that resulted from those billion years of evolution. Does that mean they are the same or fundamentally different? How about we just admit that we don't know? Instead we pull out the "human exceptionalism" card again, which generally had a terrible track record throughout history, and helps absolutely nothing in gaining understanding.
The biggest hurdle for a full escape is that the agents don't have access to their own model weights.
Opening the IDE and typing program code. With LLMs you don't have to use an IDE, you don't have look at program code, you don't have to care about coding standards. You ask the chatbot to write you a program/feature/fix and chatbot does it.
Is chatting with a chatbot still "coding"?
The part that isn't fully solved is just the software architecture side of things, do you want library A or library B, or write it all from scratch? LLM can do all three, but if you aren't careful it might go down a route that you don't like. But that again can be fixed with chat, "replace A with B", not coding.
You might still need to nudge the LLM in the right direction or stop it from going off weird tangents, but none of that involves touching actual code yourself.