Openness needed for AI Safety, the battle for the future of AI

[ygolo]

For the smaller groups, communities, and individuals who would not want cronyism to take root in a technology that could be "more transformative than electricity":

Joint Statement on AI Safety and Openness

We are at a critical juncture in AI governance. To mitigate current and future harms from AI systems, we need to embrace openness, transparency, and broad access.

open.mozilla.org

 

[ygolo]

Note, unlike the future of life AI open letter which excluded most people in understanding what was said, stating things in a weird sci-fi language, the language of this open letter is clear and inclusive.

As signatories to this letter, we are a diverse group — scientists, policymakers, engineers, activists, entrepreneurs, educators and journalists. We represent different, and sometimes divergent, perspectives, including different views on how open source AI should be managed and released.

Anyone can be an activist.

I've already signed.

If you aren't already an expert (very few of us are), I think the paths to learning and growing skills outside of large institutions (and increasingly only large, unelected corporations) need to be kept open.

 

[ygolo]

AI godfather warns Sam Altman, Demis Hassabis want to control AI

The real risk of AI isn't that it'll kill you. It's that a small group of billionaires will control the tech forever.

www-businessinsider-com.cdn.ampproject.org

 

[ygolo]

Think about Amazon's, Microsoft's, and Google's capture of cloud(mainframe 2.0), Apple's and Google's capture of mobile, and Facebook's and Google's capture of advertising technology (what social media and search really are).


Imagine if Microsoft (OpenAI), Google, and Amazon (Anthropic, Titan) capture all foundation model building from now through the future, and NVIDIA captures all hardware from now through all the future.

The lobbying and Sci-Fi doomerism seems to have led to successful rent seeking by Big Tech of the government and almost total regulatory capture (again) by Big Tech.

 

[ygolo]

 

[ygolo]

 

[ygolo]

I've known this for a while, but it's good people are talking about it.

 

[ygolo]

There's a form of AI that's been primary used for exploring the public rather than being used by the public, and that's reinforcement learning, specifically q-learning.

A search for consumer facing q-learning gives only uses that drive down costs, replacing workers, and drive up revenue by exploiting consumers.

https://www.perplexity.ai/search/What-consumer-facing-qH0tS28wTeqE6XOFvjy0ug

 

[ygolo]

Terence Tao, mathematician: ‘It’s not good for something as important as AI to be a monopoly held by one or two companies’

The Fields Medal winner is attempting to solve one of the Millennium Problems, with a reward of $1 million, but he also applies his analysis to topical enigmas such as the Venezuelan election and the advance of artificial intelligence

english.elpais.com

 

[ygolo]

https://securelb.imodules.com/s/1717/giving/interior.aspx?sid=1717&gid=2&pgid=22315&cid=35778&dids=276&bledit=1&sort=1

Make your donation now - Wikimedia Foundation

Donate to the Wikimedia Foundation, the nonprofit that hosts Wikipedia and other crucial free knowledge projects. Each year, the generosity of the 2% of readers who donate allows us to expand the reach of Wikipedia and its sister projects. Our goal is to ensure knowledge remains freely available...

donate.wikimedia.org

Open Access week this week. I thought people should know. I think of it like NPR, but for text instead of radio.

 

[ygolo]

Why open-source AI models are good for the world

Their critics dwell on the dangers and underestimate the benefits

www.economist.com

 

[ygolo]

The Open Source AI Definition — by The Open Source Initiative

A multi-stakeholder process to define an “Open Source AI”.

opensource.org

 

[ygolo]

More-powerful AI is coming. Academia and industry must oversee it — together

AI companies want to give machines human-level intelligence, or AGI. The safest and best results will come when academic and industry scientists collaborate to guide its development.

www.nature.com

 

[ygolo]

The DeepSeek R1, rule based reinforcement learning (which a lot of people call "self-play") becoming open source should be the nail in the coffin about closed source being safer humanity.

But we have to deal with what we have:

Navigating Global Regulations and Open Source: US OFAC Sanctions

Learn how U.S. OFAC sanctions impact global open source collaboration. Understand compliance, regulations, and strategies to navigate these complex legal frameworks responsibly with insights from The Linux Foundation.

www.linuxfoundation.org

 

[ygolo]

Why the 'spirit' of open source means much more than a license | TechCrunch

Arguments about what is and isn't open source can be resolved by deferring to the OSI's definition -- but there is much more to it than that.

techcrunch.com

 

[ygolo]

Open Euro LLM

A series of foundation models for transparent AI in Europe

openeurollm.eu

Hopefully, by "truly open," they mean open-sourcing the training data as well as publishing their techniques for reproduction, code, and weights.