Well, it's hard to know what the hell "tight" means without being there. If Switch is right and we steal 9/10 times here, then I'm shoving with 23o, too. On average we win about 9000 for every 7650 we risk (and we don't always lose the 7650). It's not always the case, but I have seen tables that fold to the raiser 9/10 times near the bubble. Of course, shoving every hand is likely to affect your table image and consequently reduce your success rate, so be careful.
With AQ? Say the steal only works 80% of the time. We get 8000 guaranteed for every 2 times we risk 7650. Even if we're only 25% to win the 8650, and 75% to lose our 7650, it's worth it (and that's a fairly conservative assessment of AQ). That's about 4300 won minus 11500 lost. Add in the 8000, and it's still profitable. It's also risky, so I think 80% might be the break even point.
So in this particular case, the table would have to be tight for a shove to be profitable. However, I think an 80% success rate for a big UTG shove at a tight table is downright common. In general, I'd probably shove.