Comment by Havoc
2 days ago
>I want to know where people _genuinely_ think the cutoff is
I don't think there is/should be one at all. If you and I stand in front of a crazy fast growing plant the question on how tall it'll be tomorrow is initially interesting, but a couple measurements in when it's clear the answer is exponentially the relevant question stops being how high is it now / will be tomorrow and more holly shit how long can this keep going & where does this go.
>Except they’re not. Someone is paying for this compute power, and energy.
Certainly, and the questionable subsidizations may disappear but the tech won't. If APIs disappear tomorrow I'll be looking into building a LLM rig. And researchers at unis will continue to do research to advance the tech. And chinese labs will continue. Pace may be somewhat limited but IPOs blowing up stop this tech train
>The cursor founder has said that he ships 10x
CEOs say all kinds of crazy shit. I don't think their promises and marketing is a good reference point for how a technology is actually developing
>usage based to billing
I don't think it's relevant at all for general purpose technologies like this. It'll land where market forces dictate. Maybe some people go bankrupt. Do you know what the billing model on the first electricity grids were? I don't. And I don't care. If it's useful the market will figure out the economics.
>why aren’t major engineering organisations spending all their hiring grown on these tools
They are spending big. Just because they're not committing 100% of their budget right now doesn't mean it hasn't yielded results. It's an evolving situation not a mature technology.
> If you and I stand in front of a crazy fast growing plant the question on how tall it'll be tomorrow is initially interesting, but a couple measurements in when it's clear the answer is exponentially the relevant question stops being how high is it now / will be tomorrow and more holly shit how long can this keep going & where does this go.
What’s happening here is you and I are standing in front of a plant, you’re telling me it’s growing crazy fast, I’m asking you how fast, and you’re telling me it doesn’t matter how fast because it’s going to keep growing this fast. If I ask how tall is it today, you say it doesn’t matter because it’s growing so fast and it’ll be taller tomorrow. If I ask you how tall will it be tomorrow you tell me it’s growing so fast you can’t say.
> I don't think their promises and marketing is a good reference point for how a technology is actually developing
So just to be clear, on one hand the companies are evolving so quickly but on the other hand not as quickly as they’re saying but still so quickly that we can’t quantify it.
> They are spending big. Just because they're not committing 100% of their budget right now doesn't mean it hasn't yielded results. It's an evolving situation not a mature technology.
Ok - so we’re back at the beginning again - it’s so good that companies need to keep investing in it but when I ask how good it is it’ll be better next month so I just need to keep investing in it. But you can’t tell me how good it will be next month because it’s going so well.
We can’t say how good it is today, but it’s great. And we can’t say how good it will be next month or year, but it’s improving so quickly. So what _can_ we say?
You see how crazy that is right? we’re going in circles here so I’m going to leave it at this.
>it’s growing so fast you can’t say.
That's not a fair characterization.
I can and have told you how big the plant is - it can make useful scripts right now.
I'm just avoiding that phrasing because it's just not useful measuring your rapidly growing planet by asking it repeatedly whether it has hit some arbitrary size over and over.
>But you can’t tell me how good it will be next month because it’s going so well.
Neither the companies nor I have a crystal ball. They don't know how good it'll be next month but they need to make the budget decision now. So you make a judgement call on where they think this is going based on trends. Not on a binary has it delivered photoshop yet.
>You see how crazy that is right?
You seem because you expect certainty, perfect future knowledge and binary outcomes in a world that works nothing like that so I can see how it might feel crazy