If you need the construct for a quick example to play with, use the : operator.
But if you are creating a vector/range of numbers dynamically, then use seq() instead.
Let's say you are creating the vector/range of numbers from a to b with a:b, and you expect it to be an increasing series. Then, if b is evaluated to be less than a, you will get a decreasing sequence but you will never be notified about it, and your program will continue to execute with the wrong kind of input.
In this case, if you use seq(), you can set the sign of the by argument to match the direction of your sequence, and an error will be raised if they do not match. For example,
seq(a, b, -1)
will raise an error for a=2, b=6, because the coder expected a decreasing sequence.