Acceleration of 4IR Driven Digital Transformation Through Open Source: Methods and Parallel Industries Knowledge Reapplication in the Field

Abstract

Abstract In the age of open source, oil and gas companies that invest in licensing proprietary platforms to build artificial intelligence (AI) solutions will have to closely guard against obsolescence. While the mathematics behind AI has existed for decades, its rapid adoption in virtually all sectors has been powered by open source, where a global community of academicians and data scientists are continuously developing and improving the mathematics behind the predictive and prescriptive solutions of the future. We posit that the existence of AI software development, which we describe as data driven analytical solutions providing predictive and prescriptive answers to questions, is solely due to the academia's focus on research and its pedagogical goals driving the dissemination of the research to the open source community to move the research to development. Indeed, all data science coursework is steeped in training on open source platforms and tools. When models do not depend on platforms, organizations can deploy flexible and agile software that has the ability to leverage better methods that come out in the future. Borrowing ideas from other industries we can see that just as Wordpress and its open source community of designers and developers revolutionized how websites are created, software that can quickly deploy the best available technology (and be mobile) while maintaining the security of the data is poised for user adoption and success where legacy systems and traditional dashboards fail. While the mathematics of physics does not change our confidence in its accuracy, in comparison, the mathematics of an AI algorithm is an approximation. This mathematics is being continuously improved though rarely achieving 100% accuracy. Algorithms are built upon open source AI libraries that are continuously improving. Similarly, the methods and techniques for developing the AI algorithms in oil and gas are continuously improving through two main sources, academia and adjacent industries reapplication of AI methods. We provide examples of open source technologies already part and parcel of all data scientists' repertoire of development, who are currently working in oil and gas. We further showcase reapplication of solutions from other industries, such as, medical image data analysis to seismic data processing and subsurface characterization.