negative power indicates that power is being extracted from a device (I'm assuming they mean for loads)
With the passive sign convention it's the other way around: negative power is that of sources (not loads) and it means they deliver power, while positive power means power consumed by loads. POSTIMAGE-ADD UPDATE: This is what the first snippet of your book says, i.e. it is talking about (defining) the passive sign convention. There's no contradiction betwen that def and the exercise in the 2nd image; the current source has negative power "-16 W" with the passive sign convention. The remark in parentheses "16 W delivered" is in the common parlance generator-load convention (see below), which while might not be formally defined somewhere in your book is obviously used, at least in that explanation.
With the active sign convention, it's the other way around: sources have positive power and loads have negative power.
Is the sign convention opposite for loads and sources?
Depends which sign convention you are talking about. With active or passive sign convention, the algebraic sum of powers is zero; so yeah they're "opposite" in that sense. These conventions are useful in formal reasoning about circuits (including computer-aided, like SPICE) in which the direction of the current matters. For example, a capacitor can be (at different times) a source or a load.
However with the generator-load convention both powers are positive. The latter is used in common parlance so the average joe user [of electric products] doesn't ever have to go huh over the notion of negative power. Since the power is now always a positive number we sometime need to say whether power is delivered or consumed (when it's not obvious from context). The words "delivered/consumed" are basically a "word convention" instead of a sign convention.