I have been enjoying playing Ubisoft's new blockbuster title.
I like many of the the gameplay dynamics - buttons use is fairly intuitive and easy to use. But like many others, I have noticed that the controls are often times too easy! This was explicitly done by the developers to attempt to create a more intuitive interface and I understand their motivations.

1) They want to allow a broad spectrum of players to enjoy their game! And,
2) they don't want the difficulty of a task to get in the way of the pacing(story telling) of the game.

However, I often feel cheated by my character's skills and this left wondering ways to keep the simple intuitive interface but at the same time allow players to feel a sense of accomplishment when they do things that would be difficult in real life. One solution would be to use reinforced learning.

Before we start let me be clear about three definitions.

The player: is the user - the person holding the controller(you).
The character: the in-game person, object whatever that you control.
Reinforced learning: the ability to perform a task better with practice. This makes hard tasks easier the more you do it.

Now, let me outline a situation and explain how this could work.
During on of the early missions of Assassin's Creed, your character 'Altair' has to walk along beams to cross a river. The mechanics of the crossing is to line up your character with the beam and press forward. For those of you who have played Zelda: Twilight Princess, it is the exact same. The problem with this mechanic is that it is insanely easy. You begin to dread crossing beams because as long as you keep pressing forward, you never fall. I can't remember how many times I have said "Damn F****** crossings. what a waste of time." To me anyway, this "task" has no redeeming features to the gameplay or to the story of your character.

How do we fix this?
Here is one option that come to my mind.

Give the character trainable skills. In this case balance.
Crossing beams will still require very little skill at first but the player is required to do a little of the balance work. As the player successfully completes crossings, we use reinforced learning on the characters skill so that character now guides itself and needs less input from the player. So minimally, a simple crossing can be done by anyone at any skill level but at least you feel like you have learned the skill. To keep things interesting we allow the player to attempt faster crossings that still requires input. This should keep players engaged and challenged but more importantly, it is your choice as the user to learn/perfect your skills. If you really care about moving across beams quickly then you will spend the time trying to perfect the skill. But, the important aspect is you don't need the advance skills to progress through the game. This allows the game developer to be assured that any player with any skill level will be able to enjoy their time and move forward in the storyline. Just as important, when the player is faced with important tasks, they may take a little more care in their movements as to not frustrate themselves with the negative consequence of falling from a beam. One way to look at this is as automatic RPG skill selection without the annoying and overwhelming selection screens.

Here I have outlined only one skill. We can really have skills affect a number of different gameplay elements. Auto Targeting, leaping, climbing....

The best part of this approach is that algorithmically from a programmers point of view, this type of learning is very easy to program.