Academics, strategy pundits and entrepreneurs like to talk about the transformative power of artificial intelligence (AI) and the need to accelerate implementations. AI has been called the “new electricity.” But service provider operations teams tend to tell a different and more nuanced story.
While no one disputes the long-term potential of these technologies, what’s becoming clear is that the complexity of advanced AI, such as neural networking, is difficult to fit into their current paradigms.
The pragmatic way to adopt these technologies is to define a clear use case, start with what you already have, and layer on these technologies in such a way that operations teams can augment existing architectures. A phased approach allows room to focus on learning and enhancing rather than replacing existing network operations (NetOps) practices, including root cause analysis, in order to evaluate costs and benefits. Key to this process is focusing on a better-together approach to provide evidence-based demonstration of value. In this paper, we will introduce neural networking, explain why it is a good fit for today’s networks, and provide use cases from our experience in deploying these technologies in 5G and cable environments. It is more than possible to experiment with this technology, while evolving, rather than reinventing existing skill sets and processes.