What's wrong with Open Source AI?

The gap between open and closed AI has gotten suprisingly small. Open-weight models are catching up with the frontier ones, local inference keeps getting better and we have more ways to build, run and expriment with models ourselves. So what are we actually doing with all of this??

solitary beneath the rain — xeriesjame

If you spend enough time on my side of the internet, we’re kinda just vibing. People are building things openly, sharing what they’re working on, writing on their little blogs and their little tweets. It reminds me of the 2000s internet of forums and personal websites.

It has never been this easy to build things on your own computer. You can have way more control over what you’re using than you ever had before.

Scroll through my feed long enough and you will start becoming optimistic about what’s coming next.

And the people representing this world and these ideas become hard to find. There aren’t many of them, and for those most visible personalities are controversial ones.

(The few exceptions are not much more encouraging. If you look at local AI for example, yes some figures brought a lot of attention to open models and local AI, but when you talk to them or just are in the field you realize that’s its all hype and money that drives them and much less the technology. Also idk you can tell by their content right?! What im saying here is i wouldn’t trust these people long term, give them more power.. anyway..)

Representation are important, this matter more than i think people realize.

We like to pretend our opinions and interests form independently but a ridiculous amount of them come from what we see other people doing. We know this well enough to have built entire slop advertising industrie around manufacturing that influence.

And yet when it comes to open models, we mostly just hope people will find their way to them.

So now look at people outside of X AI bubble. A lot of people don’t like AI, they see it mostly as a theft and a threat to their jobs. Look on Instragam, it’s a shame to use AI, how dare you?

last selfie beneath the downpour (homage to @xcopyart) — xeriesjame

That fear gives a lot of power to whoever gets to explain what AI is and what people should be afraid of. Some have understood this already. Look at how widely Jacob Coxon’s message has spread. It was a smart communication move by people the people behind it, I now hear so many people outside the AI bubble say that AI is going to kill us soon.

China is a very interesting contrast here. I saw a survey last week saying that 93% of the chinese believe AI will help their country. In the US, it’s only 36%. Why do they have a good storyline around AI and we don’t? ;) ;)

Meanwhile, the case for open AI barely reaches those same people.

You don’t have to convince someone to love AI to make them care about who controls it.

I don’t want companies to have my data. I don’t want them deciding what I’m allowed to do with a model. I don’t want access to intelligence to permanently depend on a subscription to a closed system. I don’t want to be part of the “permanent underclass.” (omg i hate this expression so much)

That’s a story people can understand.

There is already plenty of money going into open source models, inference and tooling. You don’t even need expensive hardware, you can just get a GLM subscription and start using open models rn.

So who’s investing in making people actually care?

i saw trojan horse weeping in the downpour — xeriesjame

Who’s funding creators to build with open models, write about them, make videos about them and push these ideas outside our little X bubble?

We’re funding the development of open AI. I’m less sure who’s funding its place in the culture.

I love that internet.

I’m just not sure it’s enough.