The question is not whether condoms reduce risk; it's whether over a reasonable period they reduce cumulative risk to an acceptable level. The study you cite implies that under best-case circumstances, millions of condom users in Africa can expect condom failure.
If "The risk of contracting AIDS from vaginal intercourse *without a condom* is extremely low" shouldn't someone tell the millions of AIDS cases in Africa? They'd be so relieved to know.
People who "engage in anal sex or sex with a partner known to have HIV" might as well dispense with the messy preliminaries because over the long run they're bound to become infected. As an advocate of the gay lifestyle you're no doubt aware that in many gay circles infection is actually sought after. With or without your advice, those having regular sex with infected partners, especially practitioners of gay sex, are bound to achieve this death wish.