The inflection point should NOT be considered a brace point.  The correct way to take into account the inflection point is by using the proper Cb factor for the moment diagram to increase the moment strength of the beam calculated using the full unbraced length.  

Note, the previous post is AISC's official stance on the subject (as published in their "AISC Steel Design After College" short course notes).  Note also that the Cb values provided in the LRFD manual are based on a totally unsupported beam between brace points.  AISC does however in the same short course notes provide per Yura revised (and less conservative) Cb values for the typical case where the top flange of the beam is braced by a composite deck.  This is not the same thing as assuming Lb as the distance between the inflection points.  Lb is still assumed to be the full length of the beam and Cb is calibrated to take into account the top flange bracing.  

I agree with WillisV that this is "AISC's official stance on the subject".  

I do know that many engineers over the years did take the distance from the column support to the inflection point as an unbraced length lu in ASD and use that for the adjoining negative moment...along with Cb = 1.  

I was at a seminar in the early 80's when Yura was asked this question and I believe he answered "yes, as long as Cb = 1 you are generally conservative doing this".  However, it subsequently came out that there are some cases, with particular parameters, where it wasn't conservative.  

We used to take the IP distance x 1.2 and use that to be extra conservative.  But this was based on engineer's response to silence in the AISC specification about this situation.

Today, I use the full length with the proper Cb.

Also - you can add bottom chord braces near the IP and definitely use the shorter length.

For lapped Z's over supports, AISI uses the distance from the end of lap to the I.P. as the unbraced length of the bottom flange.