Toasterson @[email protected]

#Illumos, #golang, #rustlang and other development things.

@OpenIndiana Maintainer

[Toasterson @[email protected] @TheRealToaster]

To make it easier for people to find all my profiles I made a quick linktree :)

https://t.co/aXXl03I1bg

Checkout all the socials and support my work on GitHub Sponsors, Patreon or Paypal

[Toasterson @[email protected] @TheRealToaster]

@maria_rcks Have Grok Bot like bots :)

[Toasterson @[email protected] @TheRealToaster]

@anniealtman108 Make it so that there is nothing they can prevent from coming out by not being alive.

[Toasterson @[email protected] @TheRealToaster]

@nmatt0 Zones: yes

[Gregory M. Kurtzer @gmkurtzer]

This is where open source has historically taken over, when everyone collaboratively joins and works together to change the world.

https://t.co/FEBClTRRv5, you should join, just sayin’…

[Gregory M. Kurtzer @gmkurtzer]

What if....

America continues to face backlash from the community against AI, data centers aren't supported, AI becomes super regulated so there is very low US competition, tier-1 AI labs don't have successful IPOs and are unable to continue at the pace we need...

What happens?

Companies globally must remain globally competitive and thus will become dependent on the only option that will exist, which will be the Chinese open-weight models.

Is that bad?

Open-weight models are actually not open-source, and this matters considerably because even though they are licensed with open source licenses (Apache, MIT, etc), we don't know what is inside of them. We don't know how they were trained. We don't know what their alignment is or will be.

This is critically important because they are black boxes that will have zero accountability.

It's become the freeware of the 2020s, but worse, because people are enabling them with harnesses without appropriate controls and no US based company to be held accountable to.

Considering how dependent AI has already become to the success of much science, medicine, research, energy, cyber-security, and software development (among others), the US loosing control and leadership of AI is actually very dangerous.

Is there a fix?

I believe there is, let's go back to talking about open source.

Historically open source has been the fix to transparency, ethics, massive growth, collaboration, and safety.

OpenWALDO is the first (and only) truly open source community for AI where anybody can join, collaborate, audit, and contribute to the entire AI corpus, tooling, and compose definitions for models.

https://t.co/nABEi3FYz6.

I've seen what a community of amazing people can do when they come together. It is what pivoted my own career over 25 years ago from biochemistry to Linux. And it is exactly what AI needs now.

Come and join the open source community. Learn about AI training, inference, infrastructure, and (not to sound cliche, but...) be part of the future!

@OpenWALDO

[Toasterson @[email protected] @TheRealToaster]

@kimmonismus And when we accepted each one of those we advanced and became more capable. But not because we accepted them as limits but as challenges.

[Toasterson @[email protected] @TheRealToaster]

@0xBakeer Happy too. I experimented during Opus 5 times already where we had this situation before. Local setups need work to be "good" its kinda like neovim come to think of it.

[Toasterson @[email protected] @TheRealToaster]

@divya_venn Yes, but then everybody had the engine. For training thats still needed. More training efficiency is mandatory!

[Toasterson @[email protected] @TheRealToaster]

@_winter_wonders This is why we need OpenWALDO and a Wikipedia style approach. Make the commons happen again. We slacked on it for a decade. Time to push the commons again.

[Toasterson @[email protected] @TheRealToaster]

@vineerpasam He uses AI in most of his streams these days. But he is good.

[Yann LeCun @ylecun]

Altman, Hassabis, and Amodei are the ones doing massive corporate lobbying at the moment.
They are the ones who are attempting to perform a regulatory capture of the AI industry.
You, Geoff, and Yoshua are giving ammunition to those who are lobbying for a ban on open AI R&D.

If your fear-mongering campaigns succeed, they will *inevitably* result in what you and I would identify as a catastrophe: a small number of companies will control AI.

The vast majority of our academic colleagues are massively in favor of open AI R&D. Very few believe in the doomsday scenarios you have promoted.
You, Yoshua, Geoff, and Stuart are the singular-but-vocal exceptions.

like many, I very much support open AI platforms because I believe in a combination of forces: people's creativity, democracy, market forces, and product regulations.
I also know that producing AI systems that are safe and under our control is possible. I've made concrete proposals to that effect.
This will all drive people to do the Right Thing.

You write as if AI is just happening, as if it were some natural phenomenon beyond our control.
But it's not. It's making progress because of individual people that you and I know. We, and they, have agency in building the Right Things.
Asking for regulation of R&D (as opposed to product deployment) implicitly assumes that these people and the organization they work for are incompetent, reckless, self-destructive, or evil. They are not.

I have made lots of arguments that the doomsday scenarios you are so afraid of are preposterous. I'm not going to repeat them here. But the main point is that if powerful AI systems are driven by objectives (which include guardrails) they will be safe and controllable because *e* set those guardrails and objectives.
(Current Auto-Regressive LLMs are not driven by objectives, so let's not extrapolate from their current weaknesses).

Now about open source: your campaign is going to have the exact opposite effect of what you seek.
In a future where AI systems are poised to constitute the repository of all human knowledge and culture, we *need* the platforms to be open source and freely available so that everyone can contribute to them.
Openness is the only way to make AI platforms reflect the entirety of human knowledge and culture.
This requires that contributions to those platforms be crowd-sourced, a bit like Wikipedia.
That won't work unless the platforms are open.

The alternative, which will *inevitably* happen if open source AI is regulated out of existence, is that a small number of companies from the West Coast of the US and China will control AI platform and hence control people's entire digital diet.
What does that mean for democracy?
What does that mean for cultural diversity?

*THIS* is what keeps me up at night.

[Toasterson @[email protected] @TheRealToaster]

@theo Aren't the Reasoning traces encrypted?

[Toasterson @[email protected] @TheRealToaster]

@Tech_girl It's evolving process. But you simply take the manager role for now. Local models can be moved to other more autonomous processes and constructs like Sociocracy or Holocracy. But hosted ones aren't trained for that.

[Toasterson @[email protected] @TheRealToaster]

@Tech_girl If I do I separate into dedicated controllers. And give orders to handle things plus global memories on my VM's processes they can follow. All time I spend on them is macOS limitations where I am waiting to get Linux/illumos hardware delivered

[Toasterson @[email protected] @TheRealToaster]

@coffeecup2020 Space weight editing to the rescue then :)

[Toasterson @[email protected] @TheRealToaster]

@anniealtman108 People hearing real singing being confused if people are ok XD

[lyra @_lyraaaa_]

editing individual neurons and attention heads to graft a conditional personality trait into qwen3.5-4b

it acts fully normal until you bring up javascript, and then it hard refuses. this method generalizes and variants work on different architectures too

blog post soon!

[Toasterson @[email protected] @TheRealToaster]

@peterrhague Then we shall be the Firstborn