It will reflect approximately* the time a user's browser would spend waiting for AJAX, assuming the load on your servers is the same as the load the user encounters, and assuming that the user has a CPU and RAM configuration comparable to the one on the computer running the test, and assuming the user is on the same speed of connection you're using to test, with the same amount of network traffic.
*Data is only approximate due to the polling frequency as discussed by user246
There's some more information in this blog post about taking averages of response times to account for variance in any of the above:
In standard benchmarks, it is common to see 90th percentile response times used. The benchmark may specify that the 90th percentile response time of a transaction should be within x seconds. This means that only 10% of the transactions have a response time higher than x seconds and can therefore be a meaningful measure. For web applications, the requirements are usually even higher – after all, if 10% of your users are dissatisfied with the site performance, that could be a significant number of users. Therefore, it is common to see 95th percentile used for SLAs in web applications.
A word of caution – web page response times can vary dramatically if
measured at the last mile (i.e. real users computers that are
connected via cable or DSL to the internet).
There's also some information here about how to measure response time:
For the analysis of server-side problems measuring at the server-side
is enough. We however have to be aware that this does not reflect the
response times of our end users. It is a purely technical metric for
optimizing the way we create content and service requests. The
prerequisite to meaningful measurements is that we separate different
transaction types properly.
So really, the question is what are you measuring? For server-side optimization, you probably want to isolate each AJAX request, run that with load using a load-testing tool, and see at what point it slows down beyond your acceptable standards as described in your SLA. Client-side optimization can only go so far when you don't control the client.