The subtle danger here is the mixture of automatics and manual flying that occurs when you have a sophisticated autopilot and FD but lack of autothrottle. Of course many aircraft are like that - the 737-200 for example. But as we have seen in this accident it is all too easy to concentrate on monitoring the flight instruments (PFD if you will) and inadvertently allowing speed excursions to go unnoticed until too late.
Boeing make a valid point in their NG FCTM that to simplify thrust setting procedures, autothrottle use is recommended during take off and climb in either automatic or manual flight. During all other phases of flight, autothrottle use is recommended only when the autopilot is engaged.
In the CRJ accident it is clear that the pilot's scan fell down and he paid insufficient attention to the thrust settings required for a stable approach. In this case, it might have been better to hand fly the ILS using FD info (nothing wrong with that providing you are competent at hand flying). Scan rate will increase if hand flying and speed excursions more quickly corrected.
Combining manual (thrust levers) with autopilot is not a good idea. Funnily enough only the other day I saw a barrel roll for fun in the simulator where the pilot left the autothrottle engaged throughout the manoeuvre in. As he dived to pick up speed (250 knots MCP) the throttles reduced to idle to maintain 250 knots then upside down the throttles went up to maintain speed then came back again to idle in the final roll out from the dive. It was a classic case of half manual - half automatics, and didn't look pretty.