Consider the common situation where the algorithm is a machine learning model. Nobody knows how it works; the best you can hope for is a bit of documentation for how the algorithm came to be created.
I don't think that is quite the answer. I think you would explain that the algorithm considers the past data on x to make decisions based on x. So then if you say 'we consider factors like your name in order to work out ethnicity and then use that to determine your credit risk' you get a big fine.
If you say 'we use data about the previous likelyhood of people in your profession and age group having an accident in order to analyze the risk....in order to price your insurance' then you are on safe ground
Serious answer, those people are not in the business of responding to legal enquires from customers. Even if they did it is still an example of "burdensome".
1) You write it once and put it in your privacy policy
2) Enquiries would be from people who are a) concerned that the decision you made was unfair and would like it reviewed by a human. b) would like an explanation of the decision.
Personally I welcome this. Sitting in a bank and being refused a mortgage because 'computer says no'[0] with no recourse or reason is tough.
If you are profiling for marketing reasons I wonder what kind of enquiries you are expecting?
Can you provide some more color here? Without concrete examples this statements is pointless.