Quoting: emphy

Quoting: LoudTechie...
Like all ML it's trained on data produced by people and thus at least subject to copyright.
Whether or not they've permission from the copyright holders to train their model with it I'm uncertain.
I would guess they've, because it's public which data they use and they've yet to be sued into oblivion.

It's the most auditable model I've encountered in quite some time, but you still can't usefully attach a debugger to it [External Link]
It's translations certainly aren't perfect, does mistranslated information count as misinformation?

Haven't got the time to search for the source, but I understood mozilla uses open datasets for their models, i.e. permission was granted or copyrights expired.

I find it amusing, by the way, that the single "ai" browser feature that is actually useful to me is rarely advertised as being such. Even the "anti-ai" vivaldi incudes it without it raising a single eyebrow.

This could be my fault.
I treat AI as equivalent to ML, but they're different words.

Wikipedia defines AI as

Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence [External Link]

and ML as

Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data, and thus perform tasks without explicit instructions. [External Link]

.

If we take these definitions.
Aimbots are AI by this definition, but is translation AI.
It's an application of Machine Learning, but that wasn't the question.
Do we typically associate translation with human intelligence and problem solving.
Is someone appear more intelligent when they can successfully translate a sentence or would we say that communication through a matching language is the intelligent part and thus using a matching a language to the one used toward you is AI.
I would argue the first, because people tend to think that people who use complicated text are more intelligent, meaning that understanding of language is more intelligent than the capacity to communicate. I disagree with my preceived majority, but that doesn't change that it's commonly associated with human intelligence.

View this comment - View article - View full comments

Quoting: Nezchan

Quoting: LoudTechieI don't think more chatbots and such will make Firefox better, but there is one recent AI feature of Firefox that although I don't use it I think is very good.

The build in translation capability. I had to download all kinds of sketchy extensions to get that to work right for my uncle and now I can just flip a setting.

As far as I'm aware, and I am no expert so I could easily be wrong, but translation is generally Machine Learning (ML), rather than Large Language Model (LLM), which the chatbots and generative AI are built on. It does fall under the "AI" umbrella, but doesn't have the same issues as LLMs such as rampant plagiarism, enormous energy use, black box processing that cannot be audited, generating misinformation, etc. So therefore not really a big issue despite both being under the same umbrella.

Machine learning is the base category. LLM is a subcategory of machine learning.

My understanding of the training methods of Mozilla are insufficient to classify it, it as a LLM, but it certainly is an LM(language model).
I would guess based on their github and budget that it's not(LLM requires self-reinforcement learning, which costs a lot).

Copyright wise all translation is the creation of derivative works although many nations tend to have personal use exceptions(in this case this matters, because I'm talking about the act of translating the webpage not the act of training the model).

Like all ML it's trained on data produced by people and thus at least subject to copyright.
Whether or not they've permission from the copyright holders to train their model with it I'm uncertain.
I would guess they've, because it's public which data they use and they've yet to be sued into oblivion.

It's the most auditable model I've encountered in quite some time, but you still can't usefully attach a debugger to it [External Link]
It's translations certainly aren't perfect, does mistranslated information count as misinformation?

View this comment - View article - View full comments

I don't think more chatbots and such will make Firefox better, but there is one recent AI feature of Firefox that although I don't use it I think is very good.

The build in translation capability. I had to download all kinds of sketchy extensions to get that to work right for my uncle and now I can just flip a setting.

View this comment - View article - View full comments

Quoting: BrandonGeneIt's shocking to me that none of these big companies have made the realization of "Hey, you know what would go over really well? Committing to NOT adopting AI in our product." Out of anyone, it should've been Mozilla.

Mozzilla is mostly(80%) funded by google the one with the money decides.

View this comment - View article - View full comments

Quoting: elmapulsome of those assets seems like they were draw on top of assets from pokémon or rpg maker...
or that they havent even changed anything on the original assets...
i wonder if this game is free from copyright violation or it might be taken down at any moment...

As Palword displayed.
Copyright isn't their largest threat.
It's patents.
They last shorter, but are much more thorough.

View this comment - View article - View full comments

Quoting: JarmerAs a follow up to my own comment LOL:

https://bsky.app/profile/zen-browser.app/post/3ma4y3npgjk2e [External Link]

So Zen is safe (for now). So seems like we have:

librewolf
zen
waterfox

as the privacy focused - fuck you ai - alternatives.

There are also tor-browser and thus mulvlad browser.

View this comment - View article - View full comments

Quoting: ChrisznixAnother one percenter (new releases) here. Guess i am an old man of habit. :)
I really lile this feature. While i get that this is a bummer for people that only play offline, it shows that they don´t collect any data then. Which is something i really like. They easily could have snuck in a service that tracks things offline.

or at least they don't share it publicly, which still is a great improvement.

View this comment - View article - View full comments

Quoting: vic-baythankfully steam controller 2 is around the corner, so i will be looking to that as my next controller.

wtf is firmware update in xbox controller? how and why it breaks any compatibility?

xbox controller should be a braindead simple device with design that has practically no changes for three decades, it is a dualshock 1 with dpad and left stick swapped. what firmware updates can it possible have?

i know what these updates do. they are made for the sole purpose of breaking compatibility with non-windows machines. although microsoft only shoot themselves in the foot with this decision.

at this point whatever they do just makes everyone hate them more.

I assure you that compatibility breaking frimware updates can have reasonable cause.
Things like EU cybersecurity laws, anti-cheat compatibility and cummulative memory corruption.
I do suspect you're right in this case though.

View this comment - View article - View full comments

Quoting: lilovent

Quoting: tpauThe question is how can we keep the almost monopoly of Steam in check and balance it out?
I buy my games only on Steam if there is no other option.

The answer is simple: provide competitive good enough services and shopping experience that are on par at least with Steam.

If any other game shop does not want to provide that or cannot do that, they won't get better market shares.

For example, Epic is still burning Fortnite money and any of their underhanded tactics did not work out, instead of treating customers better.

Steam does exactly that, that is providing in comparison to other stores the best services.

Steam's behavior is so interesting.
They behave themselves as a monopolist that fears anti-trust law, which is such a weird situation.

You know Valve's reputation of "winning by doing nothing".
That's how you win as a legal monopolist.
Let the economy of scale and ages of product refinement carry you, because a reaction would by definition be anti-competitive(quite literally you take actions to minimize the amount of competition you face).

View this comment - View article - View full comments

Quoting: scaine

Quoting: LoudTechie

Quoting: LestibournesI wonder, since Valve is now the OS vendor, if it can't implement its own kernel-level anticheat and let the game devs or anticheat devs make use of it.

It could, but the devs wouldn't trust it, because under the gpl they would be forced to release the source code defeating the security through obscurity kernel anti-cheat relies on.

That might be a strength in the long-run though, because if you're relying on security through obscurity, we all know how that ends. But if you implemented an open-source anti-cheat tool that was actually robust? That would be a huge win.

I agree.
I've done quite some research on how to do this and posted in great detail about it on this forum, so I will avoid geeking out again and making my post too long.
I'm just not as optimistic as you're and I don't think Valve is going to be that much of an innovator.
I'm blistering with ideas how to achieve it and even published a little bit of code for it, so I really really hope you're right and I'm wrong.

View this comment - View article - View full comments