In living systems that intelligence is called differential reproductive fitness, or selection, for short. That is what "useful" means to living things.
One of the problems with the model as stated here
is that the feedback mechanism isn't totally efficient.
That is, we all agree that intelligence is useful for
its "survival value", BUT . . .
an individual who happens to have high intelligence
can still die before reproducing, due to famine,
accident, war, or disease.
or become gay, and fail to reproduce at all.
In the animal kingdom, let us surmise a fish that has
won the molecular lottery and has developed (in a time
of climactic change) a greater resistance to drought
(say, a lungfish, perchance)? The model does NOT talk
about how this protects it from being eaten while still
an infant, before it passes on its genes.
Or what these type of events do to affect the RATE of
evolutionary development.
Not true or false, but incomplete, up to a point.
[Asbestos suit donned.]