Like you said, if you want it to do "something interesting," like building a colony, we're not talking about probes anymore. But even then, it seems like mass producing the would be faster.
Consider - you're clearly going to be getting somewhere faster if you accelerate for 50,000 lightyears and decelerate for 50,000 than if you accelerate for 5, decelerate for 5, build a probe, accelerate for 5, decelerate for 5, build a probe, etc., 10,000 times. In the end you're going to be covering less distance if you go directly (because you're not zig-zagging from star to star), you're going to be reaching a higher maximum velocity (you're going to be accelerating for a longer length of time), and you won't have to wait to manufacture probes along the way.
Consider - you're clearly going to be getting somewhere faster if you accelerate for 50,000 lightyears and decelerate for 50,000 than if you accelerate for 5, decelerate for 5, build a probe, accelerate for 5, decelerate for 5, build a probe, etc., 10,000 times. In the end you're going to be covering less distance if you go directly (because you're not zig-zagging from star to star), you're going to be reaching a higher maximum velocity (you're going to be accelerating for a longer length of time), and you won't have to wait to manufacture probes along the way.