I presume you are getting downvoted because V is vibe coded a bit (at least, they accept PRs from AI agents per the agents.md file).

You can hash the entire video. It would be very, very difficult (impossible?) to create another video with a hash collision that changed even a couple pixels from the original hash. You just need the hash computed from that unaltered video uploaded at time of encoding to the blockchain. It doesn’t change that someone could have created an entire AI video and uploaded it to the same proof of existence, but it does mean you could verify that a given video was observed on the blockchain at time x, and that it has not been altered in some way. Time of video creation would also be included in the hash (and metadata) so that you could verify that the video was created, by what program, at what time; and you’d have (ideally a less than 5s later) time stamp of it being verified on the blockchain.

There’s likely a better way to do this as well using zero knowledge proofs and/or homomorphic encryption but there probably exists a way to do this which would be tractable.

It’s a hard problem, but there is likely a way with cryptography and upload to server that would be verifiable. Honestly this may be the only legitimate use of blockchain.

It should be illegal to do company stock buybacks within 1 year of any workforce reduction which is the lesser of 3% of work force or 100 employees.

I think it’s more likely that we’ll need some cryptographic verification at time of recording to be valid, or a physical medium like film to be present. So perhaps film cameras or discs (non rewritable) have a resurgence.

Wondering if this is just nano hydroxyapatite with aerogel just for kicks, or if there is meaningful difference / improvement.

The market can remain irrational longer than you can remain solvent.

I think they meant on these neo competitors, you could install Linux, which they thinks is better than macOS.

I have a MacBook Pro and agree. I just wish I could get iMessage on Linux for my family that can’t / won’t change to signal or similar.

Currently, due to recent litigation, importers and companies are able to request tariff refunds. So if you paid a tariff directly, then you can request a refund and the government is required to pay you back. This is already decided and there is a refund request website.

Current lawsuits like this one are saying that Amazon requested the refund because they have the tariff receipt, and they’ll get the refund. Folks are suing Amazon because while they have the receipt, they passed on the charge, meaning they didn’t really pay the tariff in actuality. So they’re arguing that if the tariffs are illegal (already decided), and that tariff refunds are being sent out (already decided), then companies should also be required to refund their customers for the increased costs they passed along (lawsuits like this one).

It’s common sense. If a company charged 10 dollars for a product before the tariffs, charged 15 after the tariffs because it cost them 5 dollars in tariffs, then they still made the same profit after the consumer bought the product, and the consumer paid the tariff. So when a refund goes out, companies should have to return that tariff charge to the consumer. They’ll literally make the same profit and the consumer will be reimbursed then for the tariff charge they paid. This is the precedent we want to set, because otherwise consumers get screwed both ways while large companies get to pocket tariff costs. This is class warfare; working class and small business owners are losing.

A data center that will employ like 100 people and destroy their local air quality (probably), blast tons of light at night (definitely) and otherwise destroy the local ecology.

If a company adds a tariff charge to its bill when a consumer pays, or it can be shown that they passed along a charge by raising the price by the tariff amount or similar, then it is clear the consumer paid. The party paying the tariff is the consumer. Suing for a refund sets precedent, and how the current round of tariffs were deemed illegal in the first place.

Thanks for sharing this, though I still can’t get this to properly work (errors with bad vulkan layer / some swap chain error). I’ve installed gamescope with 25.08 and mango is also 25.08. Used proton up qt to install proton ge 10.34. Alas, until 64 bit comes out I prefer steam flat pak but it’s a little sad not to have hdr.

Traditional software was developed by humans as an artifact that, and to the degree that humans improved the software for some task, got better, but it was not guaranteed. Windows 11 is proof of that, and there are a laundry list of regressions and bugs introduced into software developed by humans. I acknowledge you say usually and especially for open source — I lukewarm agree with that statement but disagree that large LLMs or other generative models will follow this trend, and merely want to point out that software usually introduces bugs as it’s developed, which are hopefully fixed by people who can reason over the code.

Which brings us to AI models, and really they should just be called transformer models; they are statistical tensor product machines. They are not software in a traditional sense. They are trained to match their training input in a statistical sense. If the input data is corrupted, the model will actually get worse over time, not better. If the data is biased, it will get worse over time, not better. With the amount of slop generated on the web, it is extraordinarily hard to denoise and decide what’s good data and what’s bad data that shouldn’t be used for training. Which means the scaling we’ve seen with increased data will not necessarily hold. And there’s not a clear indication that scaling the model size, which is largely already impractical, is having some synergistic or emergent effect as hoped and hyped.

Also, we’re really not in the infancy of AI. Maybe the infancy of widespread hype for it, but the idea of using tensor products for statistical learning algorithms goes back at least as far as Smolensky, maybe before, and that was what, 1990?

We are in the infancy of I’d say quantum style compute, so we really don’t have much to draw on beyond theoretical models.

Generative LLM models have largely plateaued in my opinion.

In my experience it is obvious. Calling people on it also makes them feel embarrassed usually. I put something like “I can just ask an LLM myself if I wanted this output. Please provide your own commentary.” If I were a manager and I had an employee just copy pasting that kind of output, I’d probably wonder if that employee actually contributes anything.

I think this is the way. A certain number of times of “[coworker] wasn’t asked because they only respond with LLMs, so I just ask the LLMs directly. I am not sure what [coworker]’s expertise is anymore, I just don’t consult them” and I suspect coworker may in fact stop responding with LLMs.

There’s probably is a path to female only reproduction using artificial means. I suppose it would be possible with men too, if we could synthetically gestate a child, but it’s probably easier in women since they possess the evolved ability to gestate. Parthenogenesis has occurred in other organisms, possible perhaps for humans too, with help.

This already happens intrinsically in the models. The tokens are abstracted in the internal layers and only translated in the output layer back to next token prediction. Training visual models is slightly different because you’re not outputting tokens but pixel values (or possibly bounding boxes or edges, but not usually; conversely if not generative you may be predicting labels which could theoretically be in token space).

The field itself is actually fairly stagnant in architecture. It’s still just attention layers all the way down. It’s just adding more context length and more layers and wider layers while training on more data. I personally think this approach will never achieve AGI or anything like it. It will get better at perfectly reciting its training data, but I don’t expect truly emergent phenomena to occur with these architectures just because they’re very big. They’ll be decent chatbots, but we already have that, and they’ll just consumer ever more resources for vanishingly small improvements (and won’t functionally improve any true logical capability beyond regurgitating logical paths already trodden in their training data but in a very brittle way, because they do not actually understand the logic or why the logic is valid, they have no true state model of objects which are described in the token space they’re traversing probabilistically).

Sorry, I’m not saying that’s a good thing. It’s not just the context that’s expanding, but the parameter of the base model. I’m saying at some point you just have saved a compressed version of the majority of the content (we’re already kind of there) and you’d be able to decompress it even more losslessly. This doesn’t make it more useful for anything other than recreating copyrighted works.

Agreed. Continue the momentum and soon perhaps Mexico will go with a 4 day workweek.

Current models are speculated at 700 billion parameters plus. At 32 bit precision (half float), that’s 2.8TB of RAM per model, or about 10 of these units. There are ways to lower it, but if you’re trying to run full precision (say for training) you’d use over 2x this, something like maybe 4x depending on how you store gradients and updates, and then running full precision I’d reckon at 32bit probably. Possible I suppose they train at 32bit but I’d be kind of surprised.

Edit: Also, they don’t release it anymore but some folks think newer models are like 1.5 trillion parameters. So figure around 2-3x that number above for newer models. The only real strategy for these guys is bigger. I think it’s dumb, and the returns are diminishing rapidly, but you got to sell the investors. If reciting nearly whole works verbatim is easy now, it’s going to be exact if they keep going. They’ll approach parameter spaces that can just straight up save things into their parameter spaces.