I think the trouble is, what baby are we throwing out with the bathwater in this case? We can’t prevent LLMs from hallucinating (but we can mitigate it somewhat with carefully constructed prompts). So, use cases where we’re okay with that are fair game, but any use case (or in this case, law?) that would require the LLM never hallucinates aren’t attainable, and to get back earlier, this particular problem has nothing to do with capitalism.
This is a thing that is true of all LLMs, but it seems like you’re misunderstanding the core issue. It CAN give outputs like that sometimes. What we CAN’T do is force it to give outputs like that ALL the time.
It will answer “I don’t know” if its predictive text model guesses that the most common response to this would be “I don’t know”. To do that, to simplify a little, you could imagine that it reads your question, compares that to all the text in its training data, and tries to find the conversation that looks most like the question you asked, then answers whatever the person in the training data answered. But your exact question wasn’t in its training data, so if you took that mental model, and instead had it compare to 1000 similar looking things in its training model and average them, then it would hopefully do a better job of coming up with something at least close to what you actually asked. Now take it to a million, or a billion.
When we’re asking questions about the real world, we would prefer for it to answer based on knowledge about the real world. But what if it “matches” data from a work of fiction? Or just someone who doesn’t know what they’re talking about? Or true information, but about a different subject?
It doesn’t know anything. It doesn’t understand anything you say. It just looks at patterns that it learned from the training data and tries to guess what words are most likely to be said in that case. In other words, “here’s one case where it didn’t hallucinate” and “it will never hallucinate” are not the same thing at all.
Edit: To clarify, it doesn’t search its training data to answer your question, so asking “was this in the training data” is impossible. By the time you interact with it, the data is long gone. It was just used for training.
I agree in general, but 20 years ago, people were using email to actually talk to each other. There are problems with the protocol, but those aren’t related to the way it is federated imo. The reason people stopped using email to talk to each other was because the features of newer options were better – things like IMs and Skype, which have continued to evolve into stuff like WhatsApp or whatever people use now. But, unlike email that was devised in an era when things were still being driven largely by the education sector etc, all these other solutions were made by post-dotcom era profit-driven companies.
So I agree that email has lots of problems, and some of those are certainly related to its federation (e.g., the protocol has not really been able to advance in significant ways since making changes to it is nearly impossible). But I still think it’s the best example of a federated messaging protocol we have today.
Anyway that’s all a bit afield, as you said. I think the bottom line for me is that whichever protocol it is, if one of these current attempts at federation is going to meet my goals, then eventually there should be a large number of commercial entities participating. I know that’s not everyone’s goal though, but there’s a reason I don’t use IRC for example.
Something like 80% of email goes through Google and Apple. But, email is just about the most successful federated protocol we have. Also, I believe that these services would have become huge regardless, and I’m glad that they are dominant while using an open protocol instead of something they can exert much more control over.
In an ideal world, I believe the goal for federated social media is that you don’t care what platform other users you interact with are on, and they can freely move to other platforms without compromise. It’s scary if a big corpo controls too much marketshare and can break compatibility with other apps. But, if the protocol is truly open, there can’t be any barrier to corpos launching services on the protocol either.
I tend to agree when everyone is worried about an already existing major player joining federation (e.g. FB with threads). But bluesky is a new entrant to the space; they will have to fight the existing giants for market anyway. And if they’re starting small, then them being federated means that as soon as they start to get credible traction, any other company would be able to launch their own app in the same space. If the scare of big players is that they’ll choose to one day stop playing nicely with federation, then it will definitely be easier for them to say “you can no longer chat with a few random FOSS weirdos” than to say “you can no longer chat with this other major app”.
tl;dr, for me the goal isn’t to have a protocol that can only talk to other people who care about FOSS; it’s to have a way to talk to everyone. Eventually, that means that I hope we do hit a critical mass of “big players” buying in, even if they’re motivated by profit.
To some extent I feel like the inverse of this would be “do you wish Gmail wasn’t federated?”
There’s nothing implicit about “opening the project in unity” that needs to be a trigger for terms to change.
If you make and distribute a game made in unity, then you are distributing some unity IP. You would need the license holder to grant you permission to do that. The terms you agree to with unity are what grant you the right to distribute this.
So this has very little to do with “have you opened the editor lately”, and is more similar to when e.g. Dead By Daylight has to stop selling a dlc character because they don’t renew an agreement with the rights holders.
To my knowledge, almost zero games incorporate licenses that actually give any legal space or protection for streaming, it’s almost always a “we 100% have the right to sue you but we probably won’t, we totes promise fr fr” kind of situation.
But for this case in particular, what’s actually happening is that Japan is one of the strictest countries in the world w.r.t. copyright law; I can’t know the laws of every country in the world, but in 90% of jurisdictions the worst you’d expect to happen is the videos get taken down, maybe your channel gets deleted.
Don’t screw around with copyright law in Japan though.
Honestly I haven’t heard much rhetoric around anyone banning these terms. But if moving away from them IS good, and the entire catalyst for this conversation is “YouTube chose to use newer, more preferable terms”, then isn’t that a good thing?
Unfortunately from their perspective that isn’t apples to apples. They can charge higher rates for targeted ads.
There are cultural traditions of using colors as symbols, many of which are harmless – red for anger, blue for sadness, green for envy. Whitelist and blacklist come from the very long-standing theme of using white to represent good and black to represent evil.
Regardless of how you feel about the origin of those themes, it makes sense to start moving away from them now. Whether intentional or not, they can be harmful and aren’t really necessary.
I had never heard about this, this is absolutely awesome
I wish I could give this a million upvotes.
While this obviously would never help in any meaningful way, I do absolutely love it
So you’re saying the problem is that it’s infeasible to distribute the source code, which they already distribute to all of their developers with no problem, while there are numerous platforms that will host it for you for free if it’s public FOSS?..