Mozilla Launches a New Startup Focused on 'Trustworthy' AI (techcrunch.com) 61
On the eve of its 25th anniversary, Mozilla, the not-for-profit behind the Firefox browser, is launching an AI-focused startup. From a report: Called Mozilla.ai, the newly forged company's mission isn't to build just any AI -- its mission is to build AI that's open source and "trustworthy," according to Mark Surman, the executive president of Mozilla and the head of Mozilla.ai. "Working on trustworthy AI for almost five years, I've constantly felt a mix of excitement and anxiety," he told TechCrunch in an email interview. "The last month or two of rapid-fire big tech AI announcements has been no different. Really exciting new tech is emerging -- new tools that have immediately sparked artists, founders ... all kinds of people to do new things. The anxiety comes when you realize almost no one is looking at the guardrails."
Surman was referring to the rash of AI models in recent months that, while impressive in their capabilities, have worrisome real-world implications. At release, OpenAI's text-generating ChatGPT could be prompted to write malware, identify exploits in open source code and create phishing websites that looked similar to well-trafficked sites. Text-to-image AI like Stable Diffusion, meanwhile, has been co-opted to create pornographic, nonconsensual deepfakes and ultra-graphic depictions of violence. The creators of these models say that they're taking steps to curb abuse. But Mozilla felt that not enough was being done. "We've been working on trustworthy AI on the public interest research side for about five years, hoping other industry players with more AI expertise would step up to build more trustworthy tech," Surman said. "They haven't. So we decided mid-last year we needed to do it ourselves -- and to find like-minded partners to do it alongside us. We then set out to find someone with the right mix of academic and industry AI experience to lead it." Funded by a $30 million seed investment from the Mozilla Foundation, Mozilla's parent organization, Mozilla.ai is a wholly owned subsidiary of the Mozilla Foundation -- much like the Mozilla Corporation (the org responsible for developing Firefox) and Mozilla Ventures (the Mozilla Foundation's VC fund). Its managing director is Moez Draief, who previously was the chief scientist at Huawei's Noah's Ark AI lab and the global chief scientist at consulting company Capgemini.
Surman was referring to the rash of AI models in recent months that, while impressive in their capabilities, have worrisome real-world implications. At release, OpenAI's text-generating ChatGPT could be prompted to write malware, identify exploits in open source code and create phishing websites that looked similar to well-trafficked sites. Text-to-image AI like Stable Diffusion, meanwhile, has been co-opted to create pornographic, nonconsensual deepfakes and ultra-graphic depictions of violence. The creators of these models say that they're taking steps to curb abuse. But Mozilla felt that not enough was being done. "We've been working on trustworthy AI on the public interest research side for about five years, hoping other industry players with more AI expertise would step up to build more trustworthy tech," Surman said. "They haven't. So we decided mid-last year we needed to do it ourselves -- and to find like-minded partners to do it alongside us. We then set out to find someone with the right mix of academic and industry AI experience to lead it." Funded by a $30 million seed investment from the Mozilla Foundation, Mozilla's parent organization, Mozilla.ai is a wholly owned subsidiary of the Mozilla Foundation -- much like the Mozilla Corporation (the org responsible for developing Firefox) and Mozilla Ventures (the Mozilla Foundation's VC fund). Its managing director is Moez Draief, who previously was the chief scientist at Huawei's Noah's Ark AI lab and the global chief scientist at consulting company Capgemini.
Oh good (Score:5, Funny)
Surely this is the idea that allows them to break that 8% market share barrier.
Re:Oh good (Score:5, Funny)
Here's hoping they train the AI on the firefox codebase to get rid of all the bugs in firefox.
Re: (Score:2)
That's an oxymoron (Score:2)
...such as "jumbo shrimp" and "military intelligence".
Re: (Score:2, Informative)
Re: (Score:2)
So are your complaints: "I hate geezers".
By the way, what are the in-fashion examples of oxymorons?
Re: (Score:2)
That depends on how you understand "trustworthy". I can believe that they are aiming for an AI that doesn't state fantasies as truth, and that doesn't intentionally lie. And that's one sense of "trustworthy".
Re: (Score:2)
You're an idiot.
Re: (Score:1)
"Woke" is culturally appropriated from the African American community and essentially useless now. Thank you white people!
The correct usage, in your sentence above, is "imaginary."
Re: (Score:2)
Isn’t there a term for people such as yourself? Ah yes, snowflake.
Re: (Score:2)
Re: (Score:3)
That sounds impossible, more likely it's about being fully open source and having transparency in training data and methods
Re: (Score:3)
That depends on how you understand "trustworthy"
TFA: "For example, let users interrogate the sources behind the answers that AI chatbots give them. [...] Create systems that give users more control over content recommendation AI [...] Systems that optimize a recommender for individual or community values"
Re: (Score:1)
George? I thought you were dead but it's like your cold hand is reaching forth from beyond the grave.
Can we get a Ron Paul AI? (Score:2)
What does trustworthy mean? (Score:4, Insightful)
What is 'trustworthy' in this context?
The launch announcement [mozilla.org] seems to define it as
"AI that has agency, accountability, transparency and openness at its core ... Mozilla.ai’s initial focus? Tools that make generative AI safer and more transparent. And, people-centric recommendation systems that don’t misinform or undermine our well-being."
To some extent I can understand what transparency and openness might mean (public availability of the training data, models and algorithms, and at a technical level, perhaps AI that can explain why it gives the answers it gives).
"Don't misinform" could mean that the training data has been fact-checked, or somehow assigned truthiness values, and that the AI is somehow able to indicate the truthiness of its outputs.
The other goals I don't really understand. If I ask it for the best way to kill myself, does it answering that truthfully undermine my well-being?
I'd worry that "safer" etc really mean "infused with the biases of its creators". Nonetheless if these are transparently and openly stated, that's a step forward.
Re: (Score:2)
Nonetheless if these are transparently and openly stated, that's a step forward.
So, nothing different from whats already out there.
Re: (Score:2, Insightful)
Well, a truthful answer is not necessarily one that operationally useful. So, yes, it could give you a truthful answer to how to kill yourself that would also not adversely impact your well-being. Or that did in socially acceptable ways. E.g. it could suggest that you volunteer as a medic-assistant in the Ukraine front lines, or some other socially acceptable but extremely dangerous occupation.
Re: (Score:2)
Any AI that isn't intentionally a PR engine is going to be "infused with the biases of its creators". There's no way around it, as to the creators this will seem what's necessary for useful and honest. (The ones that *are* intentionally PR engines will also have biases, but they'll be ones that the creators don't [necessarily] hold.)
Re: (Score:2)
To some extent I can understand what transparency and openness might mean...
I find it amusing how often some organizations need to remind us how transparent and open they are.
Prediction (Score:5, Funny)
Re: (Score:2)
It will find no valid training data for that, so it will just crash.
Re: (Score:1)
What about basic social rules makes you so uneasy? You don't like that someone might not want to interact with you if you act like an asshole? You think it's your right to be as shitty as you want to everyone and should just deal with it? Oh, but heaven forbid someone treat you with anything but the deference you demand!
Oops, sorry, I forgot that you're a right-winger. 'Deference' means 'respect'.
You couldn't handle a world without the social grace you refuse to extend to others. You'd get treated exact
Re: (Score:1)
Just imagine this scenario. You invite a friend over to your house. Before he enters you ask him to remove his shoes. He replies back with sounding a bit like Eric Cartman “fuck you I do what I want”. Then proceeds to complain about being persecuted over political beliefs and saying they feel like a Jew in 1930s Germany. Then doesn’t understand why he’s not invited over again.
Re: (Score:1, Insightful)
Let's define "basic social rules" first. I suspect that yours involve a requirement to say certain things and being forbidden from saying other things. So you're really asking "what's your problem with being forced to verbalize agreement with my section of the political compass?" Am I wrong about what you're really asking?
Re: (Score:2)
Oh, please. Just admit that you want to act like an asshole without the normal and natural social consequences. This is getting pathetic.
Re: (Score:1)
Re: (Score:2)
I shouldn't need to explain to you what behaviors are acceptable in public spaces. Even unfortunate kids without proper guidance at home have that figured out by their second week in kindergarten.
No, you know what is and is not acceptable behavior. You just want to act like a shithead without any of the ordinary and natural social consequences.
You can fuck right off with your "free speech" bullshit. You're free to go make your own space where you and your buddies can scream obscenities and smear shit on
Re: (Score:1)
Re: (Score:2)
The problem is that you shit fondlers think "demanding others conform to some political view" means. Not putting up with your racist or otherwise bigoted shit is not "demanding you conform to some political view". It's basic human decency.
Frankly, I don't think anyone who would deny others basic human rights and dignity deserves to be treated like anything other than a shit stain. (Remember the golden rule? It's clear how you would have others to do unto you. I'm happy to oblige.) There is no place for
Re: (Score:1)
Re: (Score:2)
Are you incapable of honest discussion?
Don't bother answering. I know that you're not. You racist shitbags never are.
Re: (Score:1)
Re: (Score:2)
Don't you read your own comments? Fuck off, shitbag.
It'll probably be written in Rust. (Score:1)
I couldn't find a link to any of the code, or where the code was going to be made available, or the rules for writing it. So this is a guess without evidential backing. But it'll probably be in Rust.
Trustworthy? (Score:3, Interesting)
Since I don't consider Mozilla to be trustworthy, I don't see how they are going to create some AI system that is trustworthy.
Re: (Score:1)
Why don't you consider Mozilla trustworthy? All of their products are actually open source and they welcome community contribution.
Are you still crying because Eich resigned to end a massive boycott? Or is it that the Rust community has a code of conduct to keep shitheads like you from acting like shitheads and disrupting the community?
Grow up.
Re: (Score:2)
"Why don't you consider Mozilla trustworthy?"
https://www.fec.gov/data/recei... [fec.gov]
Re: (Score:2)
I'm not seeing the problem. Looks like they're acting in their own best interest and in the interests of their users.
Right-wing policies, as you know, have proven to be terrible for the economy and long-term stability.
Re: (Score:2)
and they welcome community contribution
Horseshit. Mozilla is toxic to the community and actively shuts down all discussion, recommendations for improvements, and in some cases outright insults people who raise bug reports on their products. Every question or query by the community is met with the Apple style "you're doing it wrong" except without any of the sway or authority granted by a devoted fan base. They treat their community with nothing but contempt which is precisely why there's been so many forks of their damn projects.
Re: (Score:2)
That's not been my experience, or the experience of many, many, others. Mozilla goes out of its way to make their communities welcoming for beginners and experienced developers alike.
Usually when I see these kinds of complaints, it isn't because of some unwarranted hostility from the group, but because of the, er, less than appropriate behavior of the person complaining. My advice to you, then, would be to act like a normal person, not an entitled shithead. You'll have a much better experience.
Re: (Score:2)
There's a difference between being trustworthy and merely being better than the alternatives. I'm old enough to remember when Phoenix launched, and Mozilla has done a LOT of stupid and arrogant things since then.
When you have some time, try looking up the Mozilla Lightspeed project. Now that shit will blow your mind.
Re: (Score:2)
I'm old enough to remember when Phoenix launched
Oh, my, you are growing up so fast. You're, what, pushing 30? You've practically got one foot in the grave!
When you have some time, try looking up the Mozilla Lightspeed project. Now that shit will blow your mind.
Um... Why would that "blow my mind"? What a waste of my time you've been.
Re: (Score:2)
The literal definition of prejudice.
in other words (Score:1)
Re: (Score:1)
Every other tech company on the planet is throwing money down the AI wishing well. Why do you only care when it's Mozilla and not Microsoft, Google, or Amazon?
First Thing I Thought Was... (Score:2)
Great, something else for them to fuck up.
Honest AIbak (Score:2)
Trust me!
"trustworthy" "safe" "guardrails" LOL (Score:1, Informative)
People are tired of woketastic globalist racists mediating, monetizing, and weaponizing the content they see according to their own private biases. And I sure as hell won't trust a "safe" and "truthy" AI written by a bunch of people whose political contributions are 100.0% globalist Democrats:
https://www.fec.gov/data/recei... [fec.gov]
The only way forward IMO is that I have a private system that I personally train that runs on hardware I personally own and control. Feed it my emails, my social media content, my pho
Re: (Score:2)
You don’t have to bother with all that training. It already exists. https://en.m.wikipedia.org/wik... [wikipedia.org]
Re: (Score:2)
People are tired of woketastic globalist racists mediating, monetizing, and weaponizing the content they see according to their own private biases. And I sure as hell won't trust a "safe" and "truthy" AI written by a bunch of people whose political contributions are 100.0% globalist Democrats:
Hey look, you nazi's are doing the same thing.
clutching pearls (Score:1)
> OpenAI's text-generating ChatGPT could be prompted to write malware, identify exploits in open source code and create phishing websites that looked similar to well-trafficked sites. Text-to-image AI like Stable Diffusion, meanwhile, has been co-opted to create pornogra
Not possible without true AI. (Score:2)
Me-Too, Or We Could Work on The Boring Browser (Score:4, Informative)