ABSTRACT
While the discussion of AI in the 1970s and 1980s explicitly referred to human cognition as a benchmark for the success of prototype implementation, the relation and analogy to biological systems is virtually absent in the current development and assessment of neuro machine translation (NMT) systems. Rather than representing the mechanisms of the human mind, MT systems are today more frequently considered automatized aides that extend human cognition. The chapter points out how this conception developed over the past decades and sketches the development of interactive machine translation (IMT) and its relation to classical and connectionist theories of artificial intelligence (AI). It suggests that extended and embedded theories of cognition may be instrumental to conceptually underpin and ground IMT in a wider theoretical framework.
