Just wanted to point out an implicit and not necessarily correct assumption, leading to poor-quality advice:

Suppose you know a not-very-smart person (around or below average intelligence)

It seems that you assume that intelligence is one-dimensional. In my experience, while there is a correlation, most people are smarter in some areas than in others. For example, a mathematical genius may be incapable of introspection and have to interest in rational thinking outside math. Let's take your example:

S/he read about rationality, has utilitarian inclinations, and wants to make the world better. However, s/he isn't smart enough to discover new knowledge in most fields, or contribute very much to a conversation of more knowledgeable experts on a given topic. Let's assume s/he has no exceptional talents in any area.

First, an "average person" does not read about rationality and has no "utilitarian inclinations". They often do want to make the world better if socially conditioned to do so by their church or by the TV commercials showing a sick child in the 3rd world whom you can save for a dollar a day or something. So, the person you describe is not "average".

Second, this "average person" might be (and likely is) intelligent in a way that does not show up on the IQ tests: he or she might be unusually good at running a corner store, or being a great parent, or whatever. Some of the talents may be latent, because they had no chance of being manifested. I would still call it "intelligence" by Eliezer's definition: ability to optimize the universe, or at least some small slice of it.

As a consequence, your advice is suspiciously indistinguishable from the one you'd give an "LW-smart" person. My inclination would be to find this person's area of aptitude and offer custom advice that plays to their strengths.