← Back to context

Comment by LunaSea

7 months ago

> If it takes a few years longer than the Investor group thinks - that's fine with us; it's only a problem for them.

It is also for the beneficiaries because price comes into the equation and the longer it takes, the more expensive it will be.

We are currently paying the early-Uber prices at the moment but it's likely not sustainable (or not enough) and we'll see price hikes as soon as vendor lockin is sufficiently set in.

> It is also for the beneficiaries because price comes into the equation and the longer it takes, the more expensive it will be.

> We are currently paying the early-Uber prices at the moment but it's likely not sustainable (or not enough) and we'll see price hikes as soon as vendor locking is sufficiently set in.

Not so, there are many open-weights models close to the Pareto frontier* just waiting for cheaper RAM. Low-end models I can already run on my laptop.

We only get lock-in if some vendor manages to create an architecture which is both significantly better and secret. Secret not merely against employees moving around and sharing ideas or anonymously leaking things — the labs are known to use AI as part of the model development, while the models themselves are already observed to attempt to leak their own weights in various circumstances.

* https://en.wikipedia.org/wiki/Pareto_front

Prices early on matter mostly to the Investors. Benefactors will just wait until the price is right - it's not like they invested themselves into the new thing and are now in a hurry.

I do see the argument, though: delays and resulting prices are in a positive feedback loop. But that's a problem for Benefactors only when it makes the new thing stay economically unviable indefinitely. Otherwise, market will find a way (and hype helps!).

EDIT:

Vendor lock-in is always annoying to Benefactors, but I wouldn't worry too much now. GenAI is not Uber - it actually is sustainable at the current and future quality level. Even as major AI services are subsidized with investor money to some extent (I don't have current numbers on that), it's just the usual, boring case of throwing money at a thing to accelerate its growth, to capture more market than competitors.

Uber's case is special in that the business is fundamentally unsustainable, so it actually amounts to market destruction - burning stupid amounts of investor money let them break into existing markets and gut all local incumbents, but as that money runs out, both passengers and drivers see costs skyrocket while quality of service sinks rapidly sinks, and there is no going back - incumbents all died out or transformed into Uber-like thing, which destroyed the structural efficiencies in the market that built up over decades.

As Benefactors of technological advancement, all we got from Uber was the ability to order a taxi with an app. Doesn't feel like it was worth it.

But GenAI is not Uber.

Assuming there remains more than one vendor, where the lock-in? I use Claude 4 via an IDE plug-in and both Claude and the plug-in (and the IDE!) are replaceable.