No, you're right, Biff. It is a recommended practice (studies by NASA have proved this) to always handfly the aircraft if flying in icing. You can detect ice and remedy the issue a lot sooner than if the autopilot is flying the plane.
As an IFR pilot of a single-engine GA (general aviation) aircraft who flies in the Northeast US, let me offer a dissenting opinion based on experience:
Regardless of whether the aircraft is known-ice (certified to fly into forecast icing conditions) or not, a proficient IFR GA pilot about to launch on an IFR flight will already know, thanks to proper preflight planning, that icing conditions might or will be encountered. Hence, if icing conditions are suspected during the flight, the pilot must routinely scan the leading edges of the wing or horizontal stabilizer, or scan some other protrusion in order to see the first sign of ice build-up. This is just as important as scanning the instruments for proper altitude, attitude, speeds, and heading. At the
first sign of ice build-up, the pilot must then respond by executing his/her already conceived plan to escape the ice build-up.
With this in mind, using the AP if the aircraft is so equipped is actually preferable since the AP will fly the aircraft and free up the pilot to both monitor the AP's performance (heading, alt, speed, etc) and scan for the first signs of ice formation.
Any IFR GA pilot who waits for reduced performance (aka feeling it in the hand flying of the aircraft) to confirm that ice build-up is occurring is just asking for an ice-induced episode like that which seems depicted in this audio clip.