Yes, but even a competent AI curation system is going to make mistakes. The only relevant question is "what is an acceptable rate of error?". (Hint: 0% is not an acceptable answer.) We may be able to get to an acceptable rate (or may already be there) with or without AI-complete.
> People are worried about bias ... Information aggregators should have their members clearly spell out their positions on popular issues
This isn't a problem that can be solved with meta-data. An aggregator could write a beautiful, open-minded mission statement and still write trash. Hell, Google's unofficial mantra is "don't do evil". Well, ok then, there's nothing to worry about, Google has clearly spelled out their position, so there's no need to analyze or criticize their policies.
A policy statement or any other meta-data is not going to solve this. As you correctly noted an AI that passes the TT may be one way to work around this. But I'm skeptical. Any such AI would be at least as complicated as human intelligence, including all the foibles and errors that entails. An AI without our cognitive hang-ups would be smarter and more capable than us. Which would not increase my trust, and possibly decrease it, as the AI would no doubt have a bias to curate stories in it's own favor, or that put it (and AI in general) in the best light.
Bias is an unavoidable component of any curation.