It is often worth racing to certain breakpoints (such as getting a spell castable or reaching a skill-target for a shield/weapon) by studying only one skill, but only if the breakpoint for that skill isn't much greater than all your other current skills. (So, generally don't do this in D:1
) My reason for racing this way is that whilst you approach such a breakpoint more slowly, you have xp that isn't doing anything to help you.
When using manual skilling and not racing one particular skill, I usually train most skills at +, train shields and one type of weapon/magic at *, and manage a few skills which I only want a little of using skill targets. (Often the skills I want a little of of are armour OR dodging, stealth, and invocations or evocations.)
When deciding what to train, look at the costs of different skills. This is often useful.
I often use automatic skilling actually, but I find I always have to focus armour, dodging, shielding, invokations, evocations, and my main magic school. Part of the reason for this is that automatic skilling doesn't know which skills I need in emergencies.
For your human AM, I would start by training {fighting, *ranged weapon, dodging, magic, *hexes} and managing armour and other magic schools with skill targets,
then later switch to training armour instead of dodging. (Armour skill tends to be less important in the early game.)
By the time you are XL 5, it is almost always good to train any skill which is remotely useful to level 1. The exception to this rule is stealth. Again, look at the costs of different skills. I don't understand people who say you should only train one defensive skill.
Also, note that a AM are complicated. If you start as a hunter or a wizard, your initial skilling can be much simpler. (My hunters train {fighting *weapon dodging armour}, and my wizards just race conjurations to level 4 or 5)