It's largely around LLMs, plus some generative image models. That is _where the money is_. Sure, some people are doing some CV stuff that might have some medical applications, but no-one's spending tens of billions of borrowed money on datacenters for _that_.
> or that LLMs do not create value is incredibly misleading.
They're not suggesting that.
Fundamentally, the current economic bubble around 'AI' is based on 'AGI' being achievable, soon. Otherwise, the valuations and amounts of money being spent simply do not make sense. LLMs being somewhat useful will not cut it; the spending implies expectation of dramatically increased capabilities, soon.
>>the spending implies expectation of dramatically increased capabilities, soon.
Even more than that, the spending is a death-race in expectation of:
1) a team reaching runaway superintelligence,
2) that it will be winner-take-all for the first team to get there, and
3) fear that somebody else will get there first.
So, everyone is spending as much as they can beg, borrow, or steal because in their view, there is only first place or being part of the underclass ruled by the other team that gets there first.
This is not about just making useful products/services that can increase productivity for their customers.
It's largely around LLMs, plus some generative image models. That is _where the money is_. Sure, some people are doing some CV stuff that might have some medical applications, but no-one's spending tens of billions of borrowed money on datacenters for _that_.
> or that LLMs do not create value is incredibly misleading.
They're not suggesting that.
Fundamentally, the current economic bubble around 'AI' is based on 'AGI' being achievable, soon. Otherwise, the valuations and amounts of money being spent simply do not make sense. LLMs being somewhat useful will not cut it; the spending implies expectation of dramatically increased capabilities, soon.