No individual AI solution serves as a universal remedy

Artificial Intelligence (AI) and Machine Learning (ML) are top-of-mind for decision-makers in the IT space, and rightly so. Generative AI (GenAI) is one of the most transformative technologies to enter the world in years – arguably since the launch of the first iPhone. There is no doubt that AI/ML will significantly change the way we live and work.

This is true at the network level as well. Communication service providers (CSPs) around the world are taking a hard look at how they can incorporate GenAI to improve automation to support their continued efforts in delivering better services and smarter operations. Leaving GenAI out of their strategy would only ensure they’re left behind.

But conversely, GenAI cannot be the only method or consideration.

For some time now, the networking industry has relied on rules-based automation – and despite the emergence of AI, it will remain a vital part of the solution. Simple “if this, then that” rules may sound primitive compared to AI, for sure, but they are straightforward and effective. The reality is that AI, when left to its own devices, can potentially infer something unexpected and get it wrong. CSPs may obtain a better outcome because rules-based automation is more predictable - often providing the implementation a lot faster, as well.

Meanwhile, GenAI may be the AI du jour, and it absolutely has a role to play, but not as a conductor of network operations. GenAI is great at proposing code to go on top of Software Development Kits (SDKs), for example, or enhancing the customer experience by translating issues within the network into easily digestible language.

This is why it is important to utilize a blend of AI and automation approaches. Applying the right technology, not simply the latest technology, is of vital importance – and other AI applications along with rules-based automation have a role to play, particularly in workflow-based scenarios.

Embracing a broad set of AI-based capabilities is imperative. AI functionality can be leveraged across the entire automation lifecycle, spanning planning, orchestration, assurance and analytics. For this reason, pre-packaging AI can solve specific use cases. One example is the use of predictive analytics to identify future failure scenarios, such as predicting potential network health issues in optical cards.

AI could also be used to build relationships between logical network layers, discovering cross-layer traffic paths and, in doing so, building a richer multi-layer network model that allows faster correlation of events across the network. This avoids the instances of multiple network operations teams chasing the same fault at different technology layers, which could reduce the cost and time required for network troubleshooting.

Alternative forms of AI could be leveraged for anomaly forecasting, silent fault detection, and identifying resolution urgency and priorities issues by business importance. And cross-layer circuit stitching can be implemented for orchestration and inventory use cases. Ultimately, AI can help CSPs achieve their vision around closed loop automation, driven by declarative intent and spanning across the entire automation lifecycle.  

And to make the most of AI, we need an open ecosystem – that is to say, we need to not be locked into one vendor or application provider, and instead leverage best-in-breed AI from multiple players. Network operators should also strongly consider implementing AI guardrails and human-control / feedback loops. For key functions, the control can’t be left under the control of AI alone, if at all – lest it starts automating out of scope.

If done correctly – sourcing best-of-breed technologies from an open ecosystem, and embracing a blend of AI and automation – we’ll soon see specialized GenAI applications in the networking industry. These can include Dashboard Creators leveraging GenAI, SDK genes in which GenAI can create service template code based on text or voice-based queries, and even workflow builders created by GenAI.

But it’s important to remember that, while AI and GenAI can play a significant role in the networks of tomorrow, so too can ‘traditional’ technologies, such as rules-based automation. It’s incumbent on CSPs to determine where these technologies add the most value and the least risk, and which make sense for each individual use case. A comprehensive strategy needs to be developed, not one where all eggs are put into one basket.

Kailem Anderson is Vice President of Global Products & Delivery for Blue Planet, a division of Ciena. His responsibilities include global ownership of Product Management, Engineering, Delivery, Support and Partner Ecosystems for Ciena's Software business (Blue Planet). In this capacity, he leads a team of more than 700 employees, to drive the vision, direction, execution, delivery and profit & loss of the Blue Planet suite of products into the Telecommunications market segment. Well-established in the networking industry for over 20 years, Kailem has held various leadership positions at Cisco, IBM, and Microsoft, where he focused on introducing new technologies to market in the areas of networking, security, data center, automation and SaaS. Kailem holds a bachelor’s degree in Engineering and an MBA from the University of New South Wales and resides in Sydney, Australia.