Post

As a software developer, using AI has forever changed the way I work. So far it seems positive.

It's a pick between Claude and Bring Your Own, as I understand it. Which means if you prop up an API endpoint on your own PC, you can just plug in a local model, and otherwise services like OpenRouter can also be plugged in.

So not "open models" then.

Hey Paul. I strongly recommend that you read this essay about “open models.” They can be as exploitative as proprietary models. The huge ethical question concerns capture & use of data for training. “Open” tells us nothing about that. Open =/= ethical. Open =/= progressive.

I've personally been hoping to see some way for the average person to basically replicate the entire bluesky stack without having to actually write all of the software themselves and maintain all of the infrastructure to the degrees required now. I still don't think I've seen that yet.

Bluesky never promised freedom from algorithms, they promised that others couldn't dictate your algorithms for you

So you replace the private could with private AI models, and price out half the world who can't afford tokens to keep up. Doesn't sound like a win to me.

We have given multiple grants to the community, many of which are larger than 5k, and will continue to do so

I don't know about anyone else, but I received a $39k grant from Bluesky to support my work on FedCM. The atproto.com blog is full of similar grants.

Reiterating another replier's recommendation that you read this piece. I think you're fetishizing "open" as somehow magically avoiding all the ethical/theft issues that so many people are so justifiably enraged by. (Aka this high-minded language sounds to many like "a different excuse for theft")

Yes, agreed. Be good to talk to someone about this.

What the American government is doing right now is all anyone needs to see to be radicalized about decentralization.

...but this leaves the door open for people to use open-weight models and experiment.

2nd on open models. As someone that repeatedly got screwed over by Microsoft in the beginning of the "Indie games on Xbox" days, tying your existence to a business outside of your control is a recipe for disaster. Anthropic especially has shown they greatly disrespect developers and open source.

Agree with Paul... It's worth acknowledging that most users care about outcomes, not process; using the best quality model is the only way. Also the "api" for LLMs is literally just prose. There has never been SW where it's easier to switch horses. LLMs are lock-in-proof in a way web 2.0 never was.

I think the risk is that the open models, and often unintentionally, become an excuse to keep using the proprietary hosted models and all the stuff that comes along with them. That they exist is a foil to a lot of complaints about the AI industry, despite few really using them

I am curious how you define "open" models, and I'm really not trying to be pedantic here. Open weight? Open dataset? Are you thinking about the implications of building on what courts may determine is stolen work?

Closed models will not stop improving either. This is Linux vs macOS… or accepting the discomfort of rail travel over airplanes. Being radical in theory doesn’t count.

I think personal computing without personal computers is more ideal; the form of PC’s are alienating and don’t allow us to live embodied lives.

Would love to join. Can bring the day job experience to this.

subscribe. turns out I don't really care about social media, it's the computational sovereignty that I care about

Might be interested in this, I've been working on agentic workflows lately and retaining any sort of security guarantees around them is definitely swimming upstream.

Check out userandagents.com - @burrito.space & @lukeinth.bsky.social have organized a community day next to @localfirstconf.com for the last couple of years. @komorama.bsky.social drops by the chat too.