The current master thesis researches the problem of implementing automation in Norwegian newsrooms. The technology of Natural Language Generation, powered by artificial intelligence, allows partial automation of mundane and repetitive tasks, while freeing journalists to do more creative and challenging work, like data analysis and interviews. Along with the benefits in speed and accuracy, automated journalism allows creating innovative news products using personalization algorithms. Looking at the phenomenon from the media innovations perspective, the thesis answers the questions: (1) How is automated journalism currently being implemented in newsrooms in Norway? and (2) What is its potential as an innovation? The study is based on empirical data from 11 in-depth interviews with journalists, system developers and scholars working with automated journalism in Norway, Sweden and Germany. Since the topic is so new, it was important to put Norwegian experience into the European context. The findings show that automation is suitable only for certain types of tasks in journalism, and it is beneficial only for specific types of newsrooms. The ethics of algorithms is still an important issue to explore, as computer reasoning is different from that of human and its outcome is difficult to predict. In the thesis I argue that journalism, although transformed by computational tools, still stays strong as a profession (at least for now) and mostly benefits from the introduction of automated text generation software. The possibilities of automation in newsroom along with the ethics of it should be thoroughly discussed by media scholars and made openly available for the news professionals.