Are insulators and conductors arbitrary categories? I have seen charts showing the transition from insulator to semi-conductor is at $10^{-8}~\frac{\text{S}}{\text{cm}}$ and between semi-conductor and conductor is $10^{3}~\frac{\text{S}}{\text{cm}}$. Are these switch over points arbitrary man-made conventions or is there a physical differences that sharply divides insulators, semiconductor and conductors?
I've been telling myself for example that all materials conduct electricity, it's just that some do it radically less well, up to 18 orders of magnitude less well. I'm wondering if I'm correct.
 A: Another way of distinguishing conductors and non-conductors or insulators is with band gap - for good conductors the fermi level of electrons is inside a band - semiconductors have a small band gap and good insulators have large band gaps...
Electrons in solids lie in energy bands, whereas in atoms and molecules they have generally sharp levels.
If you have a partially filled band then electrons can reorganize themselves within a band and easily allow a current to pass if a voltage is applied.
If bands are completely full no reorganization is possible, but above the highest filled band will be an empty band. - the filled is the valence band the empty one is the conduction band. If the gap between these is small then electrons from the valence can move to the conduction and then electrons can reorganize themselves a bit in both levels to allow current to flow.
If the band gap is large then it is hard to move electrons up and get a current to flow.
A: As you have expected, there is no sharp divide between the groups.  The divide is man made.

Since all conductors have some resistance, (except superconductors -
  follow this link to find out more) and all insulators will conduct
  some current if they are forced to, this means there is no absolute
  dividing line between conductors and insulators. Since some metals are
  very, very good conductors with only a very small resistance, and some
  non-metals are very, very good insulators, the terms are convenient
  when we are dealing with the usual voltages encountered in a
  laboratory. (source)

In practice the definitions are helpful because the dynamic range between good conductors and good insulators is very large, and there are not very many situations where the middle ranges between conductor, semiconductor, and insulator prove useful.
Needless to say, those who work with high voltages (hundreds of kilovolts and above) will define the boundary between insulator and not-insulator very differently than a hobbyist tinkering with 5-10V.  At higher voltages, we find the resistance of insulators starts to have more of an impact on our design, and we become more picky about how high that resistance needs to be before we call it an insulator.
A: An empirical answer:
Metals (often copper) can be used as insulating support structures in superconducting magnets.  Compared to ~0 resistivity of the coil, the resistivity of metals makes for very good insulating properties!
Or, going in the other direction, a 50-watt VandeGraaff-type power supply may output 500KV at 100uA.  Such a supply has an internal series resistance of 5e9 ohms.   So, a million-ohm resistor placed across the generator's output will act as a short circuit, pulling the 500KV down to 0.02 percent; 100V.   Or, we could use some long strings of 1-megohm resistors as wires.  They'd serve as excellent conductors; having little effect on the 500KV output voltage, even at the maximum supply current.  For this VandeGraaff machine, objects having resistance under ~500 meg-ohms are "good conductors."
So, obviously the definition of "conductor" and "insulator" varies depending on the system we're dealing with.  As a rule of thumb, the line separating the two is roughly the same as the internal resistance of any "power supplies" present.
A: As is often the case, the answer to this is actually slightly context-dependent. For many everyday purposes, the answer of Cort Ammon that it is all a matter of degree is correct.
However, one other context worth mentioning is when condensed matter physicists speak of whether a particular state of matter is a "conducting" (or "metallic") state or an "insulating" state. In this case there is a precise distinction: a conducting state is one that is gapless, meaning that as the size of the material tends towards infinity (the so-called "thermodynamic limit"), the energy difference between the ground and first excited state of the material also tends towards zero*. One can use this to classify various states of matter unambiguously and study transitions between them.
For many typical materials the classification according to this scheme will line up with the "everyday" classification, but in principle there could be exceptions. One could imagine, for example, a material that has a persistent gap but that gap is much smaller than the scale of thermal excitations, so at room temperature it is effectively conducting.
See also this blog post for some nice entry-level details on this classification.
*slightly more precisely, the spectrum near the ground state becomes continuous. The distinction becomes important in Anderson insulators, which are insulators but have a pointlike discrete energy spectrum.
