I am in a current game where basically everyone and I had a pact of secrecy against Alexander only to find that when Alexander actually decided to attack me that no one would assist me, and no one went to war with him at all.
Does Alex have your military badly outnumbered? If so, the AIs may be waiting to see how the war goes before jumping in. If you kill enough units, he may take a DOW from other civs once he is sufficiently weakened to be an attractive target. Until then, the other AIs probably figure that you're going to be defeated, and do not wish to antagonize Alexander fighting a losing battle. Under those circumstances, the best response for the other AIs is to dogpile you, in the hopes of winning an easy city or two before Alex takes them.
The AI seems to do an effective job of subscribing to realpolitik. It follows Kissinger's maxim that nations do not have friends, they have interests. The AI will backstab you if the opportunity presents itself, and it will not willingly throw itself in front of the oncoming train of a dominant civ.
The one major flaw with the way this is implemented in CiV is that peaceful play is not rewarded. In real life, war is a rare event. In part, this is because nations have broader interests and richer diplomatic options. War is also rare because it is more costly in real life. In principle, strong AIs (and humans) should prefer to achieve their aims peacefully wherever possible, should demand those ends of weak civs, and should receive those demands within reason.
Unfortunately, the unit promotion system means that it can be strictly profitable to fight a war, even if you accomplish no ends except improving some of your units. Even minimal losses are perfectly acceptable if the result is a few highly promoted, unusually durable units. This promotes warmongering, which in turn causes the programmers to tell the AI to be hyper-aggressive. Since the AI presumably calculates power from units and population (which represents the potential for units), you're disproportionately likely to be targeted early on the higher difficulties. Plan accordingly.
The diplomatic experience must also be limited because the AI is bad at calculating the relationship between its desired ends and the expected costs of attaining those ends. The more options you give an AI that cannot make this calculation, the more opportunities players have to exploit the AI. (People have found plenty of ways to exploit the few options the AI has already.) This is probably a large part of why diplomacy is so limited in this version of Civ.
But this AI makes much better decisions about determining who to fight when. If you're not getting assistance against a dominant power that everyone hates, odds are that the other AIs don't think you stand a chance.
It seems Firaxis took the lazy route of leading the player to a trial & error thing, finding out the hardway through numerous sessions.
No, this is just a game that was rushed into release to make the publisher's financials look the way the CFO wants them to look. Believe it or not, top executives of corporations don't maximize their own earnings by maximizing the company's profits, due to the way that Wall Street evaluates companies. This leads to all kinds of silly behaviors, such as rushing games that aren't ready to market.