I think it's hard to tell what the growth rate will be because the conversation is confused by hype. And part of that confusion is people not being clear about whether they're talking about the present or future.
You can *bet* on the future, but it's hard to have valuable conversations about it, because it's just comparing opinions. In-depth articles are better written about the present.
It’s impossible to tell how long the growth rate will be like this. That’s why the disagreement about the future. "If it lasts for a month…" "if it lasts for 5 years…" they are both right from different priors.
I think the growth rate itself has been somewhat exaggerated due to things like people pointing to interesting open source demos and saying "see, it works!" as if they were finished products. Screenshots can be misleading too. Also a lot of announcements at once, but announcements aren't the same as shipping and they may be projects started last year.
It's clear that AI has everyone's attention and we're seeing lots of activity, but an "overnight" success can still take months or years to build. It was a bit less than three years between GPT3 and GPT4, for example.
I don't even recall HERE maps, but Google Maps is a great reminder of how profound a change AI applications have. Maps wasn't as profound of a change as many people claim LLMs will provide, even if LLMs are "only" as profound as Google Maps was, that's still a change that will touch many aspects of society.
I think it's hard to tell what the growth rate will be because the conversation is confused by hype. And part of that confusion is people not being clear about whether they're talking about the present or future.
You can *bet* on the future, but it's hard to have valuable conversations about it, because it's just comparing opinions. In-depth articles are better written about the present.
It’s impossible to tell how long the growth rate will be like this. That’s why the disagreement about the future. "If it lasts for a month…" "if it lasts for 5 years…" they are both right from different priors.
I think the growth rate itself has been somewhat exaggerated due to things like people pointing to interesting open source demos and saying "see, it works!" as if they were finished products. Screenshots can be misleading too. Also a lot of announcements at once, but announcements aren't the same as shipping and they may be projects started last year.
It's clear that AI has everyone's attention and we're seeing lots of activity, but an "overnight" success can still take months or years to build. It was a bit less than three years between GPT3 and GPT4, for example.
Reminds me of the disrupting innovation S-curve of growth mentioned in the Innovator's Dilemma.
I think of HERE Maps vs Google Maps and how the latter overtook the former over time.
I don't even recall HERE maps, but Google Maps is a great reminder of how profound a change AI applications have. Maps wasn't as profound of a change as many people claim LLMs will provide, even if LLMs are "only" as profound as Google Maps was, that's still a change that will touch many aspects of society.