Recently, conversational generative AI chatbots have taken democratizing AI to the next level. They can now automate routine tasks and generate creative content that is nearly unparalleled. This has thrown many user communities into a professional, ethical, and situational dilemma. Many ongoing initiatives are already using generative AI to create logical, relational, and process-oriented content that is valuable to end users. From a network perspective, generative AI can be used for a wide range of activities, including compiling reports, automating the network, building tools to resolve network outages, optimizing business processes, and many more. Generative AI can manifest as a tool that network planners/operators use as an outsourced aid or as an in-house tool that can be used for complete network ops. In simple cases, generative AI can be used to respond to outage tickets, connecting users to commonly experienced problems. Generative AI can, in this case, become the backbone of an auto-response system, communicating with users of a network on one side and the operations team on the other while precisely and in a timely fashion identifying failures in network behavior. Over time, generative AI can start to run networks autonomously – where it can detect faults, gather customer feedback, create reports, and take action on those reports with minimal or no human intervention. Imagine a network that can run itself, diagnosing faults, responding to customer requests for bandwidth, and even responding to requests from other generative AI instances efficiently and effectively. When generative AI identifies issues in a network, it can execute the DevOps process by creating its own patches or code snippets to resolve the issues, making the network more efficient, resilient, and restorative. Similarly, generative AI can be used to automatically generate Request for Proposal (RFP) documents by identifying the network's needs and matching them with available technologies.
Some of the above use cases may seem like a SciFi movie ensemble, but these are all aspects of the network that can be impacted, albeit in small increments over time. The question is, which parts of network automation and network operations can be handled by generative AI, and what is the path to get there? We propose a 2.5-hour workshop at OFC 2024 that will discuss these and similar topics with industry and academic experts. The topics to be discussed will include, but are not limited to:
Generative AI for network operations
Generative AI for failure detection and resolution
Challenges in generative AI for business continuity
Similar to the first bullet] Adapting generative AI for network ops
Using generative AI framework for network service development
Can we trust generative AI for network ops? What safeguards can be put in place?
Legalities and boundary conditions on the use of generative AI from the perspective of data integrity, privacy, and anomalies.
Ashwin Gumaste, Infinera Corp., United States
Anurag Sharma, Google Inc., United States
Ricard Vilalta, CTTC, Spain