The Problem of Time
Sometimes, you can analyze what partner is likely to have for a bid, then see how that fits with your hand and the likely result of some contract.
Given UI, that make make a person more likely to put in this time and effort. The UI will almost certainly bias this analysis.
So, when an AC puts a huge amount of time into analyzing a bidding decision this way, they are probably off-track. First, they are trying to decide if a call is good or bad, not if players will consider or make the call. Second, the player at the table might not have taken the time to do this without the UI (or may not have taken the time anyway). Third, the committee will have the same biases.
There is a problem that a player at the table is likely to spend more time on a bidding decision than players being polled. If this leads to a discrepancy -- players who take time produce one answer and players who don't take time produce another answer -- I am not sure if the discrepancy is relevant to the law. I think it does -- I think the law is referring to what players would do if they were at the table. But the law doesn't actually say that. In any case, it seems to be an unavoidable problem. Also, as noted, the player might not have thought as long without the UI. Also, it could go the other way -- at the table, without the UI, there might be no reason to think about the hand. The person being polled knows there is some problem and hence might have more reason to think about the bidding problem.
NEXT: Practical Issue of Number of Players to Poll