Great read. People naively say AGI, but there are a lot of things to consider and overcome, and we aren't sure if the AGI is the thing we're expecting.
Cogent observations as usual, Rob. How do we even identify machine intelligence beyond what we already understand? It seems that intelligence in machines may be difficult to identify and even measure, since we know so little even of human intelligence. Back in the early days of evolutionary algorithms we spoke of goals. As models become increasingly complex, how do analyse these beyond what we've programmed? Are there emergent properties? Do these contain intelligence? What are their implications?
Your perspective on economic returns is refreshing. Keep it up!
Point 5 seems underappreciated to me, especially by so-called e-acc types. These people have never worked in a conventional company, which of course comprises most of corporate America. It is hard to appreciate how risk averse corporate America is, and how its adoption curve for any new technology is radically different from that of the early adopter cohort in which you find e-acc types.
Great read. People naively say AGI, but there are a lot of things to consider and overcome, and we aren't sure if the AGI is the thing we're expecting.
Memo to myself: https://glasp.co/kei/p/9a539065cfc5dc503fcd
Cogent observations as usual, Rob. How do we even identify machine intelligence beyond what we already understand? It seems that intelligence in machines may be difficult to identify and even measure, since we know so little even of human intelligence. Back in the early days of evolutionary algorithms we spoke of goals. As models become increasingly complex, how do analyse these beyond what we've programmed? Are there emergent properties? Do these contain intelligence? What are their implications?
Your perspective on economic returns is refreshing. Keep it up!
Point 5 seems underappreciated to me, especially by so-called e-acc types. These people have never worked in a conventional company, which of course comprises most of corporate America. It is hard to appreciate how risk averse corporate America is, and how its adoption curve for any new technology is radically different from that of the early adopter cohort in which you find e-acc types.