It's pretty much they just gain insanely higher income as you progress through difficulty.

Unfortunately at no level will the AI ever really become intelligent  it improves slightly at higher levels(more early game pressure most notably), but in terms of strategic depth there's no substitute for a human opponent.

The primary way the AI scales, as has already been mentioned, is in terms of gaining huge economic bonuses- an advantage that wanes a great deal in the late game when player's might is limited by fleet supply rather then resources at which point it becomes pretty obvious how bad the AI is at actually commanding battles.  If you survive to the late game against AI opponents with a decent economy/fleet you've pretty much won regardless of their difficulty.

Agressor = attacks you more

Fortifier = builds up defenses instead of attacking you

Economist = builds up economy instead of attacking you

Researcher = Researches stuff instead of attacking you

I've noticed a difference between the 4 roles, to be more specific aggressor is the hardest, economist and researcher is medium and fortifier is easy.

And while this would certainly be ideal, the honest truth is that it's unfortunately not very realistic.  So often the decisions that seem obvious and intuitive to an experienced player are actual derived from several factors in the current situation, rather then just one.  Subtle differences in the situation can warrant entirely difference courses of action.

A simple example:

If an enemy is attacking your gravity well, the decision of whether to leave repair platforms on autocast(or more precisely to chain cast it's heal to try to preserve as many frigates as possible) or to turn off autocast(to conserve AM for heals that really matter) is dependant on a large number of factors such as:

1).  Which side is likely to win the battle with the current ships present and how easily will they win(what sort of casualties should they expect?)

2). How far away are your reinforcements if any are coming?

3).  Does the enemy have the resources to sustain the assault if you manage to defeat this initial wave?

4) Is the enemy primarily relying on AoE damage or focusfire?

Creating a program that takes multiple such issues into account and makes the correct decision is exceedingly difficult as you generally have to do it exhaustively, and even if you do there are always oversights.  Not to mention #1 & #3 are themselves decision points that require intuition rather then a mere observation of the current situation.  

And that is just one of a hundred little skills a player has to get an intuitive feel for.  The honest truth is that Sins is in many ways all about recognizing your opponent's intentions and devising a a counter-strategy.  AI can follow preprogrammed response well enough, but they lack the fluidity to reliable anticipate a human player.

So, I played this game on easy. I have noticed that tutorials are pretty much only 1/3 of the info you actually need to play the game, if excluding fps and fighting games. So I played on easy to get some idea how to play this game. And I got my as* handed to me. Now, I usually do not fail on easy but this was fudiculous. I had one capital ship, had that pretty early, and about 15 other ships of various designs. The first enemy I run into would have beaten them all with just the freaking fighter platform. Instead of doing that the AI presented it's own capital ship that wiped my fleet alone in about 2 minutes. 5 minutes after that it sends 3 capitals to one of my planets that, despite the 5 gun platforms and 2 hangars had zero chance of survival. This was while that AI fought 3 or 4 other wars at the same time. 

So, what is the point if a human can not win against AI even on easy? What is the secret of winning in this game?

hmm, hard to comment specifically without knowing what specific capitalshits were involves here is some general advice:

1). you mentioned you had 5 gun platforms and 2 hangers: that is in general far too much resource to invest in static defenses.  In general it's better to invest money in a fleet or your economy rather then tying it up in static defenses.  Turrets should just about never be built(they are only really used to speed expansion which I'll go into detail on later in this post).  Hanger Bays are very expensive and generally don't get used a to nutni llate game when you have a strong economy and plenty of cash.  Your bread and butter defensive structure is the Repai Bay, which is cheap to build and make a huge difference supporting your fleets.  That said even repair bays should onyl be built at planets you expect an attack- not at every planet.  2 repair bays is generally enough for the early game at any given planet.

2).  Frigates are the workhorses of the early game.  They offer more firepower per supply then capitalships.  That said, even early game 15 frigates isn't a very large force.  you mention an attack of 3 capitalships- that's easily an 8000 credit armada- probably even 6000 credits in frigates could win a battle against such a force.

3). try to start learning the roles of each frigate type.  Corvettes and Long range Missile Frigates both excel at killing capitalships and light frigates are a bit worse but still good, so they may have been a good choice if the enemy attacked with 3 capitalships.   That said, as a rule there are a few natural combinations of frigates, the most notable being a mix of light frigates and corvettes(a Mix of Flak & long range frigates is another such natural combo).  In both these combos each frigate happens to counter the other frigate's weaknesses.  It's a fairly large amount of information to digest, but I would reccomend taking a look at the damage tables in the sticky in the strategy section of the forums- it gives all the needed information on which frigates counter which frigates.

4).  learn to expand rapidly.  As I mentioned earlier this is where turrets shine.  In general you don't need to kill every neutral militia ship with your starting capship and/or frigates when colonizing a planet- doing so is a waste of valuable time.  Kill the siege frigates first(they are the ones that will bombard your newly captured planet), then kill the long range frigates, then move on.  The first thing the newly colonized planet should start building is a turret to mop up the remaining militia ships after you move on.  After the turret finishes mopping up the militia ships it should be scrapped for resources.   As for colonization: you should colonize as soon as possible such that any survivng siege frigates won't be able to bombard the planet dead before they are killed.  It's usually safe to colonize once the last militiasiege frigate has run out of shields.

    There are also a few specific tricks.  For example to colonize a moon or asteroid you only need a single colony frigate.  To do so send the colony frigate to the moon/asteroid and colonize immediately.  As soon as you colonize build a turrent as close to the planet as you can between the planet and the siege frigate militia.  The turret will kill the siege militia before it kills the newly colonized planet.  This means you can have colony frigates solo-colonize any enearby moons/asteroids while your starting capitalship works on clearing out better defended planets.

5). learn how to build a strong early economy.  This is mostly a matter of knowing for each race which econ techs are worth researching and understanding how trade ports work(longest route and all that).

These are all very general tips, though if you could remember which specific capitalships were invovled it would be easier to give specific tips.