I'm inclined to think I just added some honesty.
When they've used it in their 'research and development to better understand your preferences and offer you better services' or whatnot, do you seriously believe that isn't to interact with ad targeting and all the other ways they use the models?
Aspersions aside you have a point about the risk for confirmation bias, but that goes for conserving overly credulous attitudes as well.
Everyone can make up their own minds about how to fill the blanks in how Google really handles this data but the track record isn't great for straightforward honesty there.
