[mnemoni_c]

Yea but how many scientists and grants you can sponsor who can find a less compute intensive way to get to AGI?

[namaria]

The bureaucracy is growing to support the needs of the growing bureaucracy. The goal of Microsoft is not AGI or AI or even software. Like any corporation, it's goal is to endure and grow. $100 billion capital integration is right on mission, as long as they can keep from ever writing it off. Deriving revenue from it would be a nice bonus.

[blackhawkC17]

> $100 billion capital integration is right on mission, as long as they can keep from ever writing it off.

Can you explain what writing it off implies in this case?

Or they just write it off I guess..

[namaria]

I mean if it ever gets to a point where they have to reckon this amount as a loss.

[b3kart]

Given the history of the field, I think it’s clear which has a higher probability of success. See http://www.incompleteideas.net/IncIdeas/BitterLesson.html

[sva_]

The essay does indeed argue that it is worthwhile to find more efficient ways of leveraging compute.