First, nobody is proposing that AV use will ever be FREE. Why didn't they charge a fee for every trip? Making a new service free when the current service costs money is a poor way to assess how the novelty will change car use.
A better study design would have estimated the cost per mile of future AV use, then charged participants accordingly.
Second, no AV comes with a human who can run errands for you, like enter a grocery and push a cart around buying goods. At most, future AVs will only drive up and wait for a preordered purchase to be loaded aboard.
If the study's AVs included any service more than moving passengers around, it crossed well outside the foreseeable use for AVs, especially short term. It's at least as likely that companies like Amazon will offer the same delivery service much more efficiently at lower cost to the customer.
It sounds like the virtual AVs in this study delivered more than real AVs ever will.