The only thing will truly give you enough context to evaluate a candidate, imo, is their process and ability to explain it. It also depends on the responsibilities of the position. UX is such a broad term that it has different tasks associated with it depending on where you are. Is the position more research focused? UI focused? etc.
If there is a more functional spin to the position (i.e. generate prototypes from preconceived wires), then you'll want to see someone perform that function. But if the position is one where they will research a problem and design a solution to it, then the process more be of more importance.
As a very quick assessment, I usually pose a problem to be solved, typically one that has just been worked through with the team. I give a high-level overview, some context of the users/goals, and then ask them to walk me through how they would solve the problem.
For example, recently I gave a candidate an idea from sales to alter our search page and results. I gave context to our users and how/why they typically use the site and then asked her to show us how she would evaluate the request and, if valid, how she would implement it. The artifact they create isn't as important to me as to show how she evaluated the problem/idea and arrived at her solution.
This article gives insight to how Moment hires designers, focusing on people:
of diverse backgrounds
We actively seek designers from varied backgrounds to enrich our way of thinking about the work we do...The best work happens when you bring all these bright and opinionated people together to work on solving client problems.
that value collaboration and critique
As you can imagine, collaboration and critique are difficult skills to evaluate in an interview. We look for evidence of effective teamwork, and want to hear about the roles that people gravitate toward in different team experiences.
and who can tell the story of their work.
A successful project story should leave your interviewers feeling interested and engaged in your solution, and should provide enough rationale to convince us that your solution is a good one.
Jonathan Baker-Bates answered this question quite succinctly when he wrote:
Finding out whether a candidate has the above qualities takes you about an hour, during which time you will have hardly any opportunity to look at their portfolio. This is a good thing because most UX porfolios are impossible to critique anyway, even if you are a UX professional. So forget that. Concentrate on the person, not whether they can make a site map or a wireframe. Drunk monkeys can make decent-looking wireframes, believe me. And I've seen them do it.
In short, focus on the qualities that you want your UX person/department to have and seek those out, not necessarily creation of an artifact.