ya, I have noticed that as well -- too many times to tell. The AI's cost/benefit analysis seems absurdly skewed in a few situations. When invading their planets, they never force a pyrhic victory on you in order to make you pay in blood for that planet, which simply by the act of delaying my advance, they would suerly benefit from staying to defend.
Another key instance of AI feebleness comes over on their attacks, when they get to a point in a battle where they think they are bound to lose, they are almost 50% wrong in their decision to phaze-jump away -- particularly in the case when a SB is present (and therefore they lose all antimatter and 30% remaining hull points).. Even if I have a phaze jump inhibitor and can effectively DESTROY nearly every one of their ships if they choose to flee, they will never stick around to AT LEAST knock out the inhibitor, and certainly never focus on a few key structures/cap ships before annihilation -- they invariably just flee with abandon.
Finally, it would be nice to see a differentiation between cap-ships and frigs/cruisers -- if I am about to lose a battle but still have a significant fleet remaining to possibly evacuate, I am likely to leave behind the frigs/cruisers (excepting maybe strikecraft hosts) in order to ( a ) knock out remaining weak enemy structures/ships and ( b ) provide cover/distraction to ENSURE the caps get out and I don't lose their valuable levels worth of EXP. Their code presently does not appear to make the strategic distinction of ship-importance, and so effectively undervalues the Capital ships to the AI's long-term detriment (and so, by mid-late in the game, I am typically the only one with a capital ship above level 5, and I have NEVER seen an AI that confronted me late-game with more than 1-2 above level 8.