This paper explores the concept of mutual (reciprocal) learning as an enabler of the emergence of a collective human-machine intelligence across a smart factory. The interlinking of digital profiles of humans and machines permits the identification and measurement of learning outcomes through participating in and performing of (shared) tasks. To achieve this goal and ultimately to transform today's smart factory into a self-learning factory, the concept model of AUTODIDACT, underlying objectives and research questions related to mutual (reciprocal) learning are outlined. [For the complete proceedings, see ED600498.]