In terms of AI, there is an inevitability of technology - in all of human history we have embraced technology and used it.  AI is no different, and ultimately the thoroughbred industry will use AI technology at some stage.  Some say we are using it now, just under a different name. 

Whether AI will be good for the industry, as we know it, is another matter.  It will certainly reduce transport and agistment costs for broodmare owners (if they own their own land and have access to a good AI vet).  It will increase safety for stallions, stallion handlers, and some broodmares - others will be more at the mercy of their owners who may be less competant than stud workers.

In terms of service fee costs, this is a complete unknown.  Supply and demand arguments would lead to the assumption that as the supply of a particular stallion's semen is increased, then his price will drop.  However, stallion books have been steadily rising over the last 50 years, and so have service fees - so this argument is difficult to hold up.
What AI will lead to is a reduction in number of stallions, and therefore those who currently have stallion puchasing power will increase that.  This will lead to more stallion power in fewer hands, therefore increasing the monopoly/oglipoly situation - and reducing competition.  A reduction in competition will not reduce service fees. 

In Aus there are just under 800 stallions.  According to the Stud Book, 90 of those stallions pull in 80% of the service fee income for the industry.  Under AI, that could easily shift to those 90 stallions getting almost 100% of the service fee income.  That sort of power base can't be good for the industry.

I can't see how that will decrease service fees.  More power to a few studs will surely mean the status quo for fees will remain in place.  Perhaps there is a reason that Coolmore hasn't commented on AI - they are in the best position to take advantage of it.