The UCLASS (or Unmanned Carrier Launched Airborne Surveillance and Strike) precursor from Northrop-Grumman, the X-47B, has recently gone to sea aboard USS Truman for testing purposes. The Navy is understandably excited; given the place that drone systems have taken in the US military and in the public mind the lack of sophisticated carrier-based UAVs was becoming aggravating. Moreover many in the US Navy view UAVs like the X-47B as the future of naval aviation, capable of carrying substantial payloads greater distances at lower cost than an F-35 and without the problems associated with sending US pilots into harms way.
Unfortunately I don't quite share the enthusiasm for high-end military UAVs. Operating in de-conflicted airspace like Yemen or Pakistan's Federal Autonomous Tribal Areas presents relatively modest challenges to drone operators, but even modest air defenses make life nasty, brutish and short for American UAVs. Against the high-powered opponents that the X-47B's stealthy airframe suggests it's successors are meant to deal with such defenses will be magnified many fold, and will included electronic warfare capabilities not used against the US in recent conflicts.
Signal loss for drones obviously causes problems, given that they rely on remote operators for essentially all vital instructions while in flight. Artificial Intelligence (AI) can allow for some degree of autonomy; most modern airliners have auto-pilot functions that can land the planes more comfortably than jittery, flesh and bone pilots. Even aside from the legal issues (or Skynet concerns) of killer robots, AIs become much less capable in unstructured situations like, say, a modern battlefield. When the area is littered with a multitude of different targets, threats, "things that you're not allowed to blow up," and stuff that exists in grey zones between those categories AIs tend to run into serious problems, especially in the vital area of rapidly making priorities and performing actions to carry them out. If real-time strategy video games get developed where units can reliably find their way across a map without getting hopeless lost we should get concerned about the onset of truly autonomous battlefield AI. Since such a day is likely to come somewhat after the heat-death of the universe, UCLASS might not be quite the futuristic jewel the Navy hopes for.