[aspenmartin]

This makes sense but I am not comfortable with the open source picture either. It depends on the use case and long term strategy. If you're handling customer support tickets or something, beyond some capability I can't imagine the ROI would be worth needing the absolute frontier. You are then in a wonderfully lovely place: absolutely full control over the model and inference stack (if you want).

But

- Plenty of businesses are stuck between a rock and a hard place: make 3p models load bearing, or sacrifice performance to the competitors that are willing to swallow that risk

- I just cannot imagine a viable equilibrium where OSS models compete on capabilities with the frontier without a hidden payer and shaky economics. Quant funds, some sovereign AI effort, cloud business, sanctioned distilled frontier models; all of these are demonstrably viable vehicles for OSS development. But the optimal position for these purposes is not to be the best, it's to be good enough (which I think is the position they find themselves in). I can imagine temporary points where open models pull ahead but not sustainably.

I agree there are many many use cases where the optimal choice is to reach for open weight models (I assume yours is one of them).

But the economics of frontier model development necessitates that these models are behind and thats a problem for other cases.

[digitaltrees]

I am not sure I understand your arguments. You seem to be taking both positions. I suspect that is because there are a few confounding variables and diverse needs AI solves and that makes a cohesive analysis hard to reach. I have a clear view that cuts through the complexity, whether you find it useful or persuasive is another matter but I’ll lay it out.

1. All intelligence has value. A Harvard MBA or Stanford CS engineer have value. Haiku has value, Opus has value.

2. Sometimes it makes sense to buy, build or rent capabilities. But lack of ownership of something your business requires to operate is an existential risk which requires ownership.

3. Current OSS models are good enough for substantial automation and productivity gains with the right orchestration and harness. Especially with domain experts in the loop.

4. Frontier models while better, introduce existential risk; so they are a temptation that should be ignored if you care about your business being able to continue to exist.

Let’s invert this conclusion. Imagine you build a workflow around Fable 5 and the workflow going down would harm your business. Well the government has demonstrated ability and willingness to impose a multi day / week / month interruption in your business with no legally required notice, right to appeal or compensation. And further demonstrates a willingness to give preferential access to your competitors.

For these reasons, the only viable path is a model you have on servers you control and can replace if needed. That means cloud providers are fine if you have the model archived somewhere.

[aspenmartin]

I agree with everything you're saying. And I would also agree if you can run your business adequately on OSS models you should absolutely do that for exactly the reasons you say. All I mean is: frontier models will always have their place and I don't see a way for OSS to ever change that picture. I also worry that the economics that makes meaningful open weights model development possible may break someday but I don't really see a mechanism for this at this point.

I don't actually have a use case in mind where you would _need_ or get a substantial competitive advantage of the absolute frontier in some sort of workflow or whatever load bearing DAG there is for your business. But if there are cases like this, that will suck.

[digitaltrees]

Red hat did it for Linux. Someone needs to do the same for AI. Make money on the implementation and maintenance not the api tokens. Tech companies have gotten fat and Happy on recurring revenue. But they are abusing the trust users have placed in them. I feel a back lash in a way I’ve never experienced.

I think AI labs are assuming that if they build and serve intelligence every one will want it. I think the last month has resulted in a lot of people saying “f u I would rather fail in life than serve you”. I personally went from feeling excited and optimistic about building amazing things on top of the models to hell bent on being self reliant and lots of my friends are doing the same.

[aspenmartin]

I think the problem is whether you're for or against local models, the training strategies etc are roughly the same.

> I think the last month has resulted in a lot of people saying “f u I would rather fail in life than serve you”.

I agree a lot of people are saying this and doing this but this won't last long. People do not want to suffer or fall behind their peers. At some point not using AI for anything will not be a feasible option. Unless you operate from an incredibly uncommon and privileged financial position, you don't really even have a choice.

[digitaltrees]

The thing the overlords are forgetting is they need customers and users.

You are right that the temptation to use the asymmetric advantage AI gives will be intense. But banks felt that temptation in 2008 in with cheap capital and 40x leverage. Some banks resisted. Which ones do you think survived in 2010? Business and technology moves in cycles. Optimize for longevity and let your competitors go out of business

[eueeu]

You’ve been boosting frontier model development for months and talking aloud about escape velocity lmao.

Take your ball and go home

[aspenmartin]

I don't have a ball, where's my ball? Did you take my ball? I want my ball.