LWD

joined 2 years ago
[–] [email protected] 10 points 2 months ago* (last edited 1 month ago)

deleted by creator

[–] [email protected] 1 points 2 months ago* (last edited 1 month ago) (1 children)

deleted by creator

[–] [email protected] 7 points 2 months ago* (last edited 1 month ago) (1 children)

deleted by creator

[–] [email protected] 49 points 2 months ago (3 children)

OpenAI has gotten virtually unlimited funding for years. It has first dibs and deep discounts on Microsoft data centers.

And somehow, despite every single trade restriction, multiple random startup companies in China (that don't even know how to secure their own databases) manage to make LLMs that outperform it.

I'm not saying that because Chinese companies are uniquely cool. I'm saying that because this whole AI thing is uniquely stupid.

[–] [email protected] 2 points 2 months ago

I figured as much. Even this line...

M1's capabilities are top-tier among open-source models

... is right above a chart that calls it "open-weight".

I dislike the conflation of terms that the OSI has helped legitimize. Up until LLMs, nobody called binary blobs "open-source" just because they were compiled using open-source tooling. That would be ridiculous

[–] [email protected] 2 points 2 months ago* (last edited 1 month ago) (1 children)

deleted by creator

[–] [email protected] 1 points 2 months ago

I've been able to run distillations of Deepseek R1 up to 70B

Where do you find those?

There is a version of Deepseek R1 "patched" with western values called R1-1776 that will answer topics censored by the Chinese government, however.

Thank you for mentioning this, as I finally confronted my own preconceptions and actually found an article by Perplexity that demonstrated R1 itself has demonstrable pro-China bias.

Although Perplexity's own description should cause anybody who understands the nature of LLMs to pause. They describe it in their header as a

version of the DeepSeek-R1 model that has been post-trained to provide unbiased, accurate, and factual information.

That's a bold (read: bullshit) statement, considering the only altered its biases on China. I wouldn't consider the original model to be unbiased either, but apparently perplexity is giving them a pass on everything else. I guess it's part of the grand corporate lie that claims "AI is unbiased," a delusion that perplexity needs to maintain.

[–] [email protected] 3 points 2 months ago* (last edited 1 month ago) (3 children)

deleted by creator

[–] [email protected] 30 points 2 months ago (2 children)

That's because search engines have reached the stage of enshittification where they no longer need to be good. Instead, they want you to spend as much time there as possible.

LLMs are still being sold as "the better option" - including by the exact same search giants who intentionally ruined their own search results. And many of them are already prioritizing agreeableness over "truthfulness." And we're still in the LLM honeymoon phase, where companies are losing billions of dollars on a yearly basis and undercharging their users.

[–] [email protected] 3 points 2 months ago* (last edited 1 month ago)

deleted by creator

[–] [email protected] 1 points 2 months ago* (last edited 1 month ago) (4 children)

deleted by creator

 

The IRS rules governing nonprofits still required the Mozilla Foundation to beg big to go big: the parent had to go find big grants from Soros, Ford, Knight, MacArthur, and give smaller grants to many. This put it in the lefties-only-no-righty-Irish-need-apply revolving-door personnel sector of NGOs and nonprofits (too many glowies there for me, too). Which meant I had a hostile MoFo over my head the minute I got CEO appointment from the MoCo board...

Of course I can't comment on anything about my exit, for reasons that only the most loopy HN h8ers still can't figure out.

https://news.ycombinator.com/item?id=43251203

 

Context

Senate Bill (SB) 1047 is legislation proposed by Senator Scott Wiener for regulating AI models that cost over $100 million to train. The bill was designed to hold AI companies accountable for potential damages caused by their models.

It gained widespread support from the population of California and a broad coalition of labor unions, AI safety advocates, Hollywood figures, and current and ex-employees of AI megacorporations.

However, many giant corporations including Google, Amazon, Meta, and OpenAI opposed the bill, asking Gavin Newsom to veto it.

Mozilla's statement

On August 29, Mozilla joined the corporations to endorse a veto, publishing its own statement:

Mozilla is a champion for both openness and trustworthiness in AI, and we are deeply concerned that SB 1047 would imperil both of those objectives. For over 25 years, Mozilla has fought Big Tech to make the Internet better, creating an open source browser that challenged incumbents and raised the bar on privacy, security, and functionality for everyone in line with our manifesto.

Today, we see parallels to the early Internet in the AI ecosystem, which has also become increasingly closed and consolidated in the hands of a few large, tech companies. >We are concerned that SB 1047 would further this trend, harming the open-source community and making AI less safe — not more.

Mozilla has engaged with Senator Wiener's team on the legislation; we appreciate the Senator’s collaboration, along with many of the positive changes made throughout the legislative process. However, we continue to be concerned about key provisions likely to have serious repercussions. For instance, provisions like those that grant the Board of Frontier Models oversight of computing thresholds without statutory requirements for updating thresholds as AI proves safe will likely harm the open-source AI community and the startups, small businesses, researchers, and academic communities that utilize open-source AI.

As the bill heads to the Governor’s desk, we ask that Governor Newsom consider the serious harm this bill may do to the open source ecosystem and pursue alternatives that address concrete AI risks to ensure a better AI future for all.

Source: Mozilla (PDF).

Gavin Newsom vetoed this bill on September 29th.

 

Mozilla recently removed every version of uBlock Origin Lite from their add-on store except for the oldest version.

Mozilla says a manual review flagged these issues:

Consent, specifically Nonexistent: For add-ons that collect or transmit user data, the user must be informed...

Your add-on contains minified, concatenated or otherwise machine-generated code. You need to provide the original sources...

uBlock Origin's developer gorhill refutes this with linked evidence.

Contrary to what these emails suggest, the source code files highlighted in the email:

  • Have nothing to do with data collection, there is no such thing anywhere in uBOL
  • There is no minified code in uBOL, and certainly none in the supposed faulty files

Even for people who did not prefer this add-on, the removal could have a chilling effect on uBlock Origin itself.

Incidentally, all the files reported as having issues are exactly the same files being used in uBO for years, and have been used in uBOL as well for over a year with no modification. Given this, it's worrisome what could happen to uBO in the future.

And gorhill notes uBO Lite had a purpose on Firefox, especially on mobile devices:

[T]here were people who preferred the Lite approach of uBOL, which was designed from the ground up to be an efficient suspendable extension, thus a good match for Firefox for Android.

New releases of uBO Lite do not have a Firefox extension; the last version of this coincides with gorhill's message. The Firefox addon page for uBO Lite is also gone.

 

sigh

 

Gary Vee is a notorious ~~grifter~~ NFT salesman with a checkered past.

Webacy is a cryptocurrency wallet "technology layer" that "provides security features" like password backup, "digital wills", etc.

 

On Valentine's Day 2024, Mozilla came out with a piece critical of AI chatbots titled "Creepy.exe: Mozilla Urges Public to Swipe Left on Romantic AI Chatbots Due to Major Privacy Red Flags."

But before they found red flags, back in 2019, Mozilla promoted a workshop on a creepy, rainbow-washed, chatbot ecosystem where people identified as "queer" were required to bare their most intimate sexual thoughts.

From the post:

your... interactions will be recorded... you will occasionally be prompted with random survey questions

What kinds of questions did they randomly ask the people who would "queer the AI"? Creepy stuff like

Have you ever sexted with a stranger?
Have you ever sexted with a machine?
Do you remember the first time you were aroused by language?
Do you think an artificial intelligence could help fulfill some of these... needs?

The workshop providers guided people into establishing an intimate, sexual connection with the chatbot they could create.

How might we build trust with an AI?
How might we give it its own sense of desire?

Even the consenting participants in the workshop complained about the AI's creep factor:

it feels like the A.I. is gas-lighting you. Seems like a noncommittal sexting bot. It should at least be clear about what it’s trying to do.

The startup that Mozilla fostered for this panel ended up crashing and burning, but its creepier, worse brethren live on inside of Firefox 130, displayed as first-class options within Mozilla's chatbot options. I just thought it would be fun to take a trip down memory lane to see how many creepy red flags AI companies could get within Mozilla's view without ever concerning them.

 

Now that Google and Microsoft each consume more power than some fairly big countries, maybe it's time for 2024 Mozilla to take heed of 2021 Mozilla's warnings.

 

There seems to be minimal information about this online, so I'm leaving this here so cooler heads can prevail in discussion.

Link to filing: https://archive.org/details/jyjfub

Notable portions:

Teixeira was hired as Chief Product Officer and was in line to become CEO.

Mr. Teixeira became Chief Product Officer (“CPO”) of Mozilla in August, 2022. During the hiring process, Mr. Teixeira had conversations with executive recruiting firm, Russell Reynolds Associates, that one of Mozilla Corporation’s hiring criteria for the CPO role was an executive that could succeed Mitchell Baker as CEO.

Also, shortly after being hired, Mr. Teixeira had conversations with Ms. Baker about being positioned as her successor.

After taking medical leave to deal with cancer, Mozilla swiftly moved to replace CEO Mitchell Baker with someone else.

Shortly before Mr. Teixeira returned from leave, Mozilla board member Laura Chambers was appointed Interim CEO of Mozilla and Ms. Baker was removed as CEO and became Executive Chair of the Board of Directors.

After returning, Teixeira was ordered to lay off 50 preselected employees, and he objected due to Mozilla not needing to cut them and their disproportionate minority status.

In a meeting with Human Resources Business Partner Joni Cassidy, Mr. Teixeira discussed his concern that people from groups underrepresented in technology, like female leaders and persons of color, were disproportionately impacted by the layoff.

... Ms. Chehak verbally reprimanded Mr. Teixeira, accusing him of violating [a] non-existent “onboarding plan” and threatening to place Mr. Teixeira back on medical leave if he did not execute the layoffs as instructed.

Mozilla's lack of inclusivity was a known problem

In February 2022, Mozilla commissioned the firm of Tiangay Kemokai Law, P.C. to assess its performance in providing a diverse, equitable, and inclusive workplace culture.

The report delivered in 2023 from Tiangay Kemokai Law, P.C. states in part: “MoCo falls into the Cultural Incapacity category based on leadership’s inadequate response to the needs of a diverse culture or else the need to create a more diverse, equitable, and inclusive culture, which is reflected in current systems, processes and procedures, policies and practices, or the lack thereof, and are incongruent with MoCo’s stated values and goals.”

Steve Teixeira has been put on leave.

On May 23, 2024, Mozilla placed Mr. Teixeira on administrative leave.

Mr. Teixeira requested a reason for being placed on administrative leave.

Mozilla did not provide Mr. Teixeira with a reason why he was placed on administrative leave.

Mozilla cut off Mr. Teixeira’s access to email, Slack messaging, and other Mozilla systems.

Mozilla instructed employees not to communicate with Mr. Teixeira about work-related matters.

Upon information and belief, an investigation into Mr. Teixeira’s allegations was finally conducted in late May 2024, but Mozilla did not do so under its internal policies and procedures regarding managing complaints of discrimination. Mr. Teixeira was not contacted to participate in the investigation into his complaint of unlawful treatment.

Coverage online so far

~~I say "alleged" because there appears to be no consensus on the veracity of this document.~~

Update: this appears to be confirmed.

This has received no "news" coverage besides one angry loudmouth (Bryan Lunduke) whose entire commentary career has been shaped by his political beliefs, regardless of truth.

view more: ‹ prev next ›