Crux is that Facebook/Meta has now been almost a decade in the VR space and they still have no idea what to do with it. They are just stumbling in the dark wasting tens of billions of dollar with little to show for. They sure have the money and will to build the next big thing, but only a very vague idea of what that thing might even be to begin with. It doesn’t help that they basically fired everybody of the original Oculus crew that got the VR space up and running again in the first place. Even their Metaverse that they spend so much effort hyping up is a complete nothingburger, it’s not just that nobody cares, it’s that they haven’t even managed to build anything worth calling that, they are still playing catch up with features from PlayStationHome 15 years ago (or Habitat from over 37 years ago).

depth is reliable

What if one of them is dirty? What if you are driving with the sun right in front of you? What on a foggy winter day? The big problem here isn’t even what the cameras are or aren’t capable of, but that there is little to no information on all the situations Tesla’s autopilot will fail in. We are only really learning that one deadly accident at a time. The documentation of the autopilots capabilities is extremely lacking, it’s little more than “trust me bro” and “keep your hands on the wheel”.

The fact that it can’t even handle railroad crossing and thinks trains are a series of trucks and buses that blink in and out of existence and randomly change directions does not make me wanna blindly trust it.

Humans don’t have the best track record when it comes to safe driving, might not be the best idea to imitate them when there is better tech around.

It’s a general problem with ChatGPT(free), the more obscure the topic, the more useless the answers will be. It works pretty good for Wikipedia-style general knowledge, but everything that goes even a little deeper is a mess. This is true even when it comes to things that shouldn’t be that obscure, e.g. pop-culture things like movies. It can give you a summary of StarWars, but anything even a little more outside the mainstream it makes up on the spot.

How much better is ChatGPT-Pro when it comes to this? Can it answer /r/tipofmytongue/ style question?

That depends heavily on the framing. AI right now is overrated, it can do a ton of stuff, but rarely anything useful. AI in the (near) future however will eat all our jobs and everything else.

It’s always worth keeping in mind under what restrictions current AI still operates. ChatGPT never saw a compiler or a shell, never could test its code. It’s pure book-knowledge without any practical experience. Yet it still can produce quite good code, at least for small problems. What it might be capable of once it has access to a compiler and runtime environment could far surpass what we have today. But it’s speculation, since we don’t have that yet.

ls reaction to this is unexpected:

$ mkdir foo
$ echo Foo > foo/file
$ chmod a-x  foo
$ ls -l foo
ls: cannot access 'foo/file': Permission denied
total 0
-????????? ? ? ? ?            ? file

I expected to just get a “Permission denied”, but listing the content it can still do. So x is for following the name to the inode and r for listing directory content (i.e. just names)?

We could absolutely do it

Every exploration into hostile environments heavily relies on goods and services imported from the rest of Earth. Biosphere 2 is as far as I know still the only time we ever tried to actually build a completely independent ecological system, but that was 30 years ago, in a non-hostile environment, only run for a short amount of time, still had tons of problems and would still be missing a lot of stuff to be truly self sustaining for long time periods (e.g. no industrial facilities).

They store it encrypted on your device such that you can’t access it yourself, only their app can access it.

they don’t keep your data.

They don’t allow you to export your chat history, it’s locked in the app:

https://www.reddit.com/r/signal/comments/t5m666/exporting_a_single_signal_chat_history/

That’s not only crappy app design, that outright violates European law:

https://gdpr-info.eu/art-20-gdpr/

Ain’t perfect, but close enough for now: https://imgur.com/a/mHHeJyn

The human expertise is in the data.

There are no art lessons in the training data, just labeled images. How to actually draw the AI has to figure out by itself. Even the labels on the data aren’t strictly needed, they are just there so the humans can interact with the AI by text.

Same with AlphaZero, it didn’t learn playing Go from humans, it learned that by playing against itself and not only beat humans, but previous versions that were still based on recording human Go games.

That’s the bitter lesson of AI research: Throwing data and computation at the problem gives you much better results than human experts.

And we have barely even begun to explore this. What ChatGPT does is still just reciting information from books and websites, it can’t interact with the real world to learn by itself. It being based on books written by human experts is not a benefit, but what’s holding it back.

Examples are welcome.

Not that it really matters in this case, but AI art just seems inconsistent in silly ways.

It’s to be expected given that the AI has no access to the previous image.

The whole point of AI is that those systems aren’t fundamentally different. There is little to no human expertise that goes into those systems, it’s all self learned from the data. That’s why we are getting AIs that can do images, music, chess, Go, chatbots, etc. all in short order. None of them are build on decades worth of human expertise in music or art, but simply created by throwing data at the problem and letting the AI algorithms figure out the rest.

Why, when speaking intelligently and thoughtfully in the subject, is he so wrong in his assessment, when you, in one lazy sentence, are so right?

Obama is employing good old human exceptionalism and moving the goal post. A tried and true method of argumentation that has continued to fail for the last 50+ years when it comes to AI. “AI is good at X, but not Y” becomes “AI is good at X and Y, but not Z” the next year. Focus on a tiny niche that AI hasn’t covered yet, while ignoring the pace at which AI is advancing. Wasn’t too long ago that people where proclaiming that computers could never be creative. Nowadays that switched to “but it can’t beat the human masters”. Well, guess what? That did neither hold true for Chess nor Go and it won’t work out for Bob Dylan music either. Be prepared for a future where AI is better at everything. It will come and much sooner than people expect.

It’s also worth keeping in mind the quantity of AI generated content. I still hear tons of artists talk as if AI were competing with them on a level playing field. But in the time they finish one image, AI finished thousands or even millions. This is not just about AI replacing the human, but completely shifting how we deal with information in general. Something like ChatGPT isn’t interesting because it can write better websites than a human, but because it completely bypasses the need to visit websites in the first place. You ask the AI and the AI delivers the answers. There is no intermediate step where knowledge needs to get dumped into a static website or a book.

so of course this looks like impressive art to younger people and no one else;

So what exactly do you consider “impressive art”?

you’ll love the AI music we sell you too!

The money argument is frankly a bit pointless at the moment, since nobody has figured out how to make money with AI art. Any new AI tool is replicated by another company within months. The AI art itself is essentially worthless, since everybody can just generate it by themselves for free. A lot of the tools are also Open Source and can run on your own PC locally.

Maybe one day we’ll have some AI-TikTok with an endless stream of user-targeted AI content and ads in between, but that doesn’t exist yet. For the time being, AI is just costing companies money, while users can enjoy it for free.

Also, no need to be dramatic.

Signal is an over hyped piece of shit that grossly violates numerous core tenets of pricey privacy and data freedom.

I’m gutted that you can’t tell the difference.

Then fucking enlightenment. Everybody keeps talking “AI art bad” yet nobody keeps showing what good art is supposed to be.

Just take a look at the variety of garbage available on all streaming platforms right now

Yeah, and that was created by humans. Guess why I like AI art? It can produce way more interesting stuff than that. The bar for “human art” is pretty f’n low. What do you consider “good art”?

Could you please answer the question: What is good art?