A(Narrow)I is a cheaper form of labor already. I suppose it’s plausible that its General form may not be, but I won’t be betting in that direction.

Assuming we ever get to AGI[1], even if it is 10e9 times more expensive than human labour, ln(10e9) ~ 23 => Within 23*18 months compute will have become theoretically cheap enough to make it equivalently in price to human labour[2].

[1] Arguably, full generality is not required, only generality across the labour domain in question. [2] Of course, the question remains: Who will be able to reap this reward. My bet is on the fabs + chip design.