How Artificial Intelligence Is Changing the Telecom Workforce —
The telecommunications industry will thrive, based on the capability of its service providers to innovate as they move ahead with implementing advancing technologies. Artificial intelligence (AI) and machine learning (ML) are 2 digital forces already impacting how work is performed, whether it’s your favorite beverage being prepared by a robot barista, or virtual assistants handling increasingly larger volumes of requests flooding customer interaction centers.
To date, the role of AI within the telecom industry has been limited to chatbots which automate the routine customer enquiry, extracting the intent to ensure a customer is routed quickly to the proper channel. Telecom providers, however, are increasingly moving towards using AI to not only lower operating costs and improve network efficiency, but to also improve the customer-engagement experience.
For example, by using exploratory data analysis that looks for specific patterns, AI can also detect anomalies in the network or even predict the possibility of a dire event happening. This AI capability can identify suspicious activity related to potential security threats as well.
Such insights will enable telecom service providers to take proactive action and fix problems before their subscribers are impacted. What we see is that AI and ML will become intrinsic parts of more and more enterprise automation initiatives in telecom — beyond back office and network operations — to include the subscribers and the workforce.
Ah, the Possibilities
How can AI and ML transform and improve operations for telecom service providers that serve mobile, fixed, cable, and broadband, markets? Let’s explore some of the possibilities and terms.
Robotic process automation or RPA — offers telecom service providers a way to shift the routine, time-consuming and low-value tasks to software robots. The result is that subscribers with more challenging issues are better served by experienced customer service agents. And, the agents themselves are able to take advantage of their knowledge and trouble-shooting skills — which makes their jobs more interesting.
AI virtual assistants — sometimes referred to as digital assistants or chatbots, is software that uses AI to understand natural language voice commands and to perform tasks. In the telecom customer service centers, AI virtual assistants provide timely assistance with the large volumes of requests that come in, providing better service to customers and reducing costs. Billing inquiries, service requests, and equipment maintenance requests, are a few examples.
Natural language processing — Beyond routine, low-skill, and repetitive tasks and chatbots, natural language processing, or NLP, will become an important part of business workflows. For example, the huge CDR data corpus can be provided to an AI algorithm and an NLP pipeline run on it to identify and extract important insights, such as top grievances and typical requests. This helps ensure that the workforce is correctly aligned to this data, and that AI assistants are trained accordingly.
Optimizing productivity of human agents — AI allows telecom providers to collect, store, and analyze, data and thus achieve real-time behavioral insights from their subscribers. Making the customer profile, along with areas of interest, typical complaints, and past history, available to human agents helps them address the customer’s queries more quickly. This capability translates into an optimized workforce and increased subscriber satisfaction.
Augmenting Employee Performance
When telecom service providers transition to a digital workforce, there are some “must haves” that will allow for successful and faster innovation. The ability to scale, and to have security and governance, is critical to long-term success. And, we can’t stress enough that a plan and a methodology must be in place. Other things to consider include:
• Do appropriate personnel have access to clean and curated datasets to run their ML algorithms? A data catalogue makes curated datasets discoverable with the right checks, governance, and access-related privileges in place.
• Are tools and technologies accessible for data preparation, such as data curation, cleaning, and merging, to make the data consumable by the ML models/algorithms?
• Is there ready access to predefined ML algorithms, preferably with an intuitive drag-and-drop interface that does not require high-level coding skills for building ML models?
• Is the environment easily available for relevant personnel to experiment and run training on? AI and ML typically work better with huge datasets and algorithms may be compute intensive. Hence, the availability of a ML sandbox is a must for a thriving ML culture in any organization.
• What will be the complete model lifecycle management process, such as retraining, inference, scoring, continuous performance evaluation, versioning, and more?
• Are there planned processes to carefully consider consumption of the insight produced by the AI/ML models, as well as integration with existing applications?
The AI Future Looks Bright
What can we expect to see in the future as telecom service providers increasingly incorporate AI and ML?
Customer-support chatbots and virtual assistants are already deployed successfully. Beyond these tools, we will see AI and ML used to divert network traffic to avoid call drops and optimize costs.
There is already much work being done to deploy ML algorithms for network security using anomaly detection and log analysis. AI-assisted customer support and sales have proven successful, and are considered mainstream. These successes allow organizations to explore AI-based content processing and management as well.
Many telecom providers are also exploring using AI-based predictive maintenance to proactively fix problems with communications hardware, hence ensuring highly skilled engineers address the problem proactively, thus streamlining the workforce.
With almost every organization embarking on an AI and ML program today, Persistent Systems sees ML operationalization and ML democratization as critical aspects of this strategy. For organizations wanting to extract maximum impact from AI and ML for their business, services and skills beyond the model development are required. Businesses need services that will enable and support the ML model lifecycle. It’s a unique combination of data scientist, data engineering, and DevOps, which leads to the path to success.