Due to technological developments, the future of education has often been discussed with dramatic visions. In the 1960s, it was debated whether computers would replace teachers in classrooms. In the 1970s, more concrete visions came up with expensive solutions in form of large mainframe computers with interactive terminals connected to digital learning libraries. In the 1980s, it was expected that the emerging personal computers would individualize education. Around the year 2000, the Internet came along for the public with Wikipedia and other information sources that again questioned the importance of books and teachers. The next step have been mobile computer systems that basically enabled contextualized authentic learning outside of school with ubiquitous Internet-based information access and even digital information augmentation of the physical world. And today, many believe that it will be generative artificial intelligence systems that will fundamentally and rapidly change the world of education. This contribution reflects this history and likely future of digital educational technologies to discuss why some of them succeeded and others failed. Through a generalization of digital learning technologies, the strengths and weaknesses of such technologies in supporting learning processes in an increasingly complex world are characterized. From there, promising computer capabilities for future educational technologies are combined, leading to what we will call Knowledge Media Machines and Digital Knowledge Building Environments mainly based on post-constructivist theories. Finally, the prototypical system Ambient Learning Spaces (ALS) is outlined that has been developed and evaluated in a research project for over more than a decade.