There is an internet meme in Japan called “Odrokiya(驚き屋)” who are overly surprised by the release of OpenAI and overstate the usefulness of AI.
Should we think of AI as being hyped by such people?
I’m not sure what you’re actually trying to ask here.
Are ai advancements being OVERLY hyped? Yes. That doesn’t mean there’s nothing to them, though.
Are ai advancements being hyped by paid ai hype shills? My guess would be yes. Again, that doesn’t mean there’s nothing to them.
Will the max usage end up being as high as current hypers are hyping? Unknowable. There is SOME amount of false hype, but also SOME amount of meat to the hype, and SOME amount of continued improvement in ai abilities. If ai abilities (not just underlying model strength, but also the ability to harness models in real use cases) stopped increasing today, then the total realized gain from AI would fall short of existing hype. If ai abilities continue to increase, then who knows when the hype bubble will pop.
Either way, when the hype bubble DOES pop, it will fail to reach the usefulness of peak hype, but that “when” could be tomorrow or could be after 10x improvement in actual usefulness.
I think it's being overhyped by those who fall into either of these two categories;
- financial incentive to do so
- feel the barrier to entry for "coding" is low enough for them to "build"
I think there's a fine distinction between the terms overhyped, and the reality of deceptive advertising, just as there is a fine distinction where 5% of the economy can impact 90% of the economy.
Most of the economy is entry level positions. You sieve these positions to get mid level talent. You sieve them again to get senior level talent.
What will naturally happen when a business sector that is a natural labor multiplier but only about 5% of the economy, multiplies that labor with the outcome being entry level positions are replaced.
5% affecting 90%. Will it take over completely, not at the time we talk about this. We know people age and die eventually.
What happens when there is no economic incentive because there are is no economic demand for entry level positions which would be needed to get the remaining 10% of mid and senior positions as an aggregate?
Its a sequential pipeline. No one going in, means no one coming out. The pool shrinks over time until it collapses. No AGI needed...
The mere expectation to business leadership that labor can be replaced by using AI, regardless of outcome because debt financing decouples these, ensures you've already integrated before you can see the problem, and by the time you see it there is nothing you can do to solve it. Its a classic cascade failure, like a dam breaking, or snow pack that becomes an avalanche.
In other words, these are the problems where lives are lost at least once, in order to find the co-factors that objectively show the cascade failures so you can react and adapt in time.
I think in general it is not hyped, but the current wave of it is. Right now the cutting edge is LLMs, transformers and architectures around that. I think it's possible it will reach a limit in the next year or two, and even though that will change a lot of things it won't be a completely autonomous agent that can exist in the world like a human.
idk about the meme but the posts need to stop, they are definitely defeating the purpose of natural usage, causing mass effects that are beyond their control.
Are ai advancements being OVERLY hyped? Yes. That doesn’t mean there’s nothing to them, though.
Are ai advancements being hyped by paid ai hype shills? My guess would be yes. Again, that doesn’t mean there’s nothing to them.
Will the max usage end up being as high as current hypers are hyping? Unknowable. There is SOME amount of false hype, but also SOME amount of meat to the hype, and SOME amount of continued improvement in ai abilities. If ai abilities (not just underlying model strength, but also the ability to harness models in real use cases) stopped increasing today, then the total realized gain from AI would fall short of existing hype. If ai abilities continue to increase, then who knows when the hype bubble will pop.
Either way, when the hype bubble DOES pop, it will fail to reach the usefulness of peak hype, but that “when” could be tomorrow or could be after 10x improvement in actual usefulness.