What do you consider an "LLM provider"? Is it a website where you interact with a language model by uploading text or images? That definition might become too broad too quickly. Hard to ban.
the bulk of money comes from enterprise users. Just need to call 500 CEOs from the S&P500 list, and enforce via "cyber data safety" enforcement via SEC or something like that.
everyone will roll over if all large public companies roll over (and they will)
Logseq + Syncthing has been working quite well. Logseq is well maintained, has a fully functional Android app, and a thriving plug-in ecosystem. I have been using the combination for a year and it's an excellent long term solution.
> Mozilla is starting to seriously have a long list of highly questionable if not directly user hostile behaviors.
Would you care to provide examples? I am a longtime user of Mozilla products unfamiliar with the topic and I am genuinely curious.
> What should we think of their VPN they try to promote so much
Mozilla does not have its own service but rather resells Mullvad, one of the most privacy focused services in existence. Is there more to this story that I am unaware of?
Allow me to add this to the other sibling comments: Pocket was an... interesting series of choices.
"""
Mozilla replaced a feature that was end to end encrypted with one that sent private data to a third party for data mining.
They denied getting paid for the integration. That was technically true. They eventually admitted they got paid for referrals.
They bought the company in 2017 and promised to release the source code. They still haven't.
The Pocket website says "as a member of the Firefox family, privacy is paramount."[1] The first part is misleading and the second part is simply false.
Open Sourcing something is never a easy task especially if it calls for a complete rewrite which i assume is why it still has not been open sourced yet
Buying a technology company, they buy a proven idea. If the bought tech has a diffrent stack than everything else Mozilla already had then rewriting it is going to be a good long term idea.
> everyone copy-pastes commands other people wrote straight into the terminal
I know a lot of people that use Linux and not many of them operate this way. Most care about their software sources. "Everyone" is certainly not the case.
And yet when I complained about `curl | sh` on HN the other day, I got ridiculed. "Everyone" is too much, but even on a purportedly "hacker" website, people find the idea of perusing a shell script before executing it preposterous.
Something that's hard to remember, but helps a little: if you get 3 people saying stupid things, that's only 3 people -- not necessarily representative of the people out there.
But `curl | sh` is no less secure. Download this file and execute it. Functionally the same outcome. Tell me how doing that is materially different than `apt get`. Both employ signing and checksums (just with different PKI). One delegates trust to a package maintainer while the other trusts the author directly. I truly don’t understand the paranoia and consider it tinfoil hat security theater.
While I agree with the sentiment that Tesla has wonderful UI/UX, your 10+ step EV-to-ICE comparison is disingenuous. Tesla does not scrape the ice by itself.
What is a legitimate use case for this? I can think of a hundred applications for deceiving others but struggle to come up with a scenario where one would want their voice cloned or reproduced.
You're recording a podcast and want to tweak some of your own words, without the hassle of re-recording.
You're an indie game developer, and want to have vibrant NPCs with their unique voices and dialogues powered by a LLM.
You're producing a movie, and want to tweak certain lines of dialogue; with the consent of the talent.
You suffer from health conditions and are gradually losing your voice, but you still want to communicate.
There are certainly legitimate use cases of this technology. I personally believe illegitimate use cases overshadow the legitimate use cases, but I don't think it's fair to say there are no legitimate applications.
We should strictly regulate the use of this technology by criminalizing abuse; not by banning it altogether (which is pretty hard in the case of software and small models).
> You're producing a movie, and want to tweak certain lines of dialogue; with the consent of the talent.
The latest agreement to end the last round of strikes was to prevent this very thing.
Of your list, the medical condition to give someone their real voice instead of a Hawking voice would be the most legit reason. Everything else is a skewed sense of morally acceptable as I think they are shady
I wouldn't go as far as that. Plenty of indie to AAA games are produced using commodity assets / resources (e.g. why make your own tree model, when there's plenty of pre-mades in the marketplace). Yes, that takes away work from artists, but it is part of productivity and game development.
Centuries ago, elevators were manned. Today, they're all electronic. It is the inevitable march of progress and productivity.
There's a difference from starting the project by using OpenVoice vs hiring an actor for 90% of the work but then cloning their voice because you can' be bothered to reschedule the same actor for creative changes.
But if you start that project by making a voice sound like Morgan Freeman because you can't afford Morgan Freeman but you feel entitled then you can go pound sand. So your choice of making a generated voice should be of a voice that someone else isn't already using.
You may not be trying hard at all then. The first thing I thought was to clone your voice to use in real-time translations. I can probably think of several others mentioned in comments below, but this is a 100% always-useful never-nefarious(assuming perfect translation not being maliciously used) application.
Yes, I have, just now, been thinking about how few, and featureless, syllables I can utter to determine if an unknown caller has a legitimate reason for calling, such as a large-delivery driver, medical lab or other real call.
It seems the best tactic is to make the caller do as much of the talking as possible.
Or use a generic cloned voice to interact with unknown callers.
I have a friend with a paralysed larynx who is often using his phone or a small laptop to type in order to communicate. I know he would love it if it was possible to take old recordings of him speaking and use that to give him back "his" voice, at least in some small measure.
Unfortunately I have yet to see something that can do this and provide a voice model that you could plug into Android TTS and/or Windows which are what he uses.
Fixing small errors in narration, voiceovers, or other recorded content.
Translation of recorded content with the original voice into new languages.
Comedy as long as it’s obvious that it’s a fake.
Actually intentionally selling your voice to be the voice of some text to speech product. Maybe I want Alexa to have the voice of Danny Devito, as long as he’s ok with it and getting paid.
My wife has been sick all week and has to communicate over text because her voice has gone. We’ve been talking about making voice clones of ourselves for situations like this. Some people never regain their voices so preserving them before they lose it is super valuable.
I imagine training people, and having everything I say be available in any language, matching my tonality, and being able to reach a global audience. I'm very much looking forward to this.
Indie Gamedevs can do their own Voice Acting?
See also indie film, same use case.
Actor dies / hit by buss before a work is finished - Create a few more lines posthumously (it'll be in the fine print of the contract that you allow voice & image fakes in the event not able to do them).
Satire, Pranks, and alleged pranks (stuff that makes folk laugh).
What do you consider an "LLM provider"? Is it a website where you interact with a language model by uploading text or images? That definition might become too broad too quickly. Hard to ban.
reply