The intersection of "pressure cooker", "backpack" and visits to overseas countries not known for sponsoring low-level terrorism is perhaps not the most obvious criterion for identifying terrorists, so presumably they'd also have to flag most of the ~12,000 people who searched for "bomb-making" last month, for example, and perhaps some more innocuous ones like independent searches for "martyr" and "white house" or "chechnya" and "aeroplane". That's a lot of resource.
Now even if they're pretty selective about who they actually pay a visit to, they're going to return a lot of false positives. No big deal, except that pre-Snowden the programme is supposed to be a secret and even within the subset of people flagging the "possible terrorist" alert they're going to investigate far more journalists (common occupation) than terrorists (rare occupation). Which is a bit of a problem if their modus operandi is instructing police to knock on people's doors and ask them whether they were searching for X on the internet. Was this really PRISM's raison d'etre?
The alternate hypothesis involves a freelance writer famous solely for her number of Twitter followers embellishing her police enquiry story with some very topical themes.
[1]If I worked in intelligence I'd solve that problem by not telling local cops what potential suspects had been doing on the internet...
The use of resources shows that we're serious - so yeah, they waste them. Profligately. Or haven't you flown recently?