The transformation of Mobile Telecoms with AI - David Copperfield takes the stage!

As the product owner of a mobile network planning software, I remember an ex-CEO of network operator saying “the ultimate future of mobile network planning would be ‘you think of rolling out a site…and a site gets rolled out’”. At the time, this seemed as fantastical as a  David Copperfield magic trick. However, the rapid advancements in AI, particularly generative AI, are transforming this vision from magic to reality.

The Magic of AI in telecoms

The excitement and apprehensions surrounding AI are palpable. However, the mobile telecom sector has well and truly commenced its transformation journey by incorporating AI in numerous aspects of its operation from customer care to network operations.

The AI magic box

Whilst generative AI and tools like ChatGPT dominate headlines, AI encompasses a number of branches including Machine Learning (ML), Deep Learning (DL), Generative AI (GenAI), Computer Vision (CV), Robotics and Natural Language Processing (NLP). The diagram below captures these categories and their interactions. Each of these branches enables the creation of unique solutions to the diverse challenges and opportunities in mobile telecoms.

Transforming mobile telecoms

Mobile telecom networks are huge and complex. Operators, vendors and standards bodies like 3GPP and the RAN Alliance are exploring numerous use cases that integrate the different branches of AI within these highly complex networks.

Adapting and extending the RAN Alliance classifications, the use cases can be classified into six broad categories:

Near term AI applications

According to a recent survey conducted by NVIDIA (State of AI in Telecommunications: 2024 Trends (nvidia.com)), operators expect AI integration in the following areas in the near term:

AI for the Customers

Enhancing customer support with AI has clear, immediate benefits. Many operators are already deploying AI-powered virtual assistants, chatbots, agent assists, and audio transcriptions to improve efficiency and effectiveness. Unlike traditional rule-based chatbots, modern AI systems offer more dynamic and responsive customer interactions.

Agent assists are also becoming popular with operators as they improve the efficiency of customer support, resulting in reduced churn. These assists are powered by various AI technologies such as Automatic Speech Recognition (ASR), Natural Language Processing (NLP), Machine Learning (ML) and Generative AI.

Automatic Speech Recognition (ASR) can transcribe calls to the call centres in real time, aiding subsequent processes that utilize NLP, ML, Neural Networks and Generative AI to:

• Route calls to the best resource

• Assess call sentiment

• Predict follow-up queries

• Gauge customer satisfaction to anticipate churn

• Summarize calls

In addition to resolving issues accurately and rapidly, these solutions can also help operators personalize the interaction with customers in terms of content, tone, and timing as well as  personalised plans and tailored promotions.

For some operators who use AI voice agents, Generative AI also enhances the ‘humanness’ of the conversation by using human-like sentences.

To further democratize these solutions, challenges such as phonetic ambiguity, audio quality, and domain vocabulary are currently being addressed.

AI for the Workforce

Mobile networks are complex and incorporate a vast number of processes. AI has a role to play in terms of supporting a wide spectrum of processes within the operators’ environment to increase workforce productivity.

Generative AI tools like MS office co-pilot and GitHub co-pilot can optimize everyday tasks and coding activities respectively. Large Language Model (LLM) based solutions also assist field engineers by synthesizing vast technical manuals for troubleshooting.

When it comes to operations and maintenance, one of the more nascent branches of AI, computer vision offers some unique solutions to the operator workforce. Computer vision can identify and analyse infrastructure issues such as cable damage or equipment malfunction. Operators can leverage computer vision to do real time monitoring of network infrastructure and detect anomalies, outages and damage to reduce downtime and overall network efficiency.

Gen AI also has the power to make the processes more conversational. An area where this is particularly useful in enhancing workforce productivity is network trouble shooting. ML and Gen AI based solutions can augment the role of the network engineer by assisting and accelerating trouble shooting and problem solving.  For example, they can automatically decipher network logs and convert into English to speed issue resolutions. 

AI for RAN

AI can support the radio access networks performance optimisation in a number of ways:

·        Spot anomalies in the radio network or detect radio interference

·        Predict network traffic, allowing the operators to meet the changing demands of the network

Mobile  networks consume 3% of the world’s electricity and hence one of the biggest challenges for the operators is to reduce the consumption of electricity. AI can help operators optimise their networks to improve energy efficiency and reduce overall carbon footprint.

Continuing the magic - Disappearing apps and other innovations!

Looking ahead, the 3GPP has already commenced work on 6G which will be AI native. In addition we will see many exciting new use cases in each of the 6 areas described above. AI on devices will be an interesting area with some operators and vendors already exploring app-free mobile phones. The device will learn what the user needs and in a similar way to a David Copperfield illusion, AI will make the apps disappear!