Sensitivity to assumptions concerning the applications of computers to education is needed, as educational practices are strongly shaped by both group and individual assumptions. An educational practice will not necessarily be improved if done by a computer. A consequence of searching for computer uses has been a tendency to build from computer strengths, not student weaknesses, and specific assumptions by designers concerning human behavior have created unexpected effects. Most importantly, educational research may not have a positive influence on computer uses in education, as generally assumed. The existence of computer assisted instruction (CAI) is made tenuous by problems of poorly developed materials, limited availability of properly developed materials, locally produced materials, and widespread copying of materials. Therefore, not only is it questionable to assume that educational research can have a positive impact on technology implementation in the schools, but some effective methods such as programmed instruction and CAI can be almost completely negated by premature and inadequate implementations within the educational system. Videodiscs and complex packages such as word processing can help maintain centralized control over computer uses. Prediction of eventual educational roles for computers is difficult, and effective roles should be promoted. (LMM)