How Generative AI is Revolutionising the Telco Industry — First Steps
This article is part of a series of practical articles of our journey with generative AI in one of the world’s biggest telecommunication groups — Virgin Media O2 and Liberty Global/Telefonica.
For as long as I can remember, innovation has been at the heart of my work as an engineer, data scientist and in technology management. Working in Data Science and Machine Learning has often led me to the frontlines of technological breakthroughs, compelling me to confront and grapple with the cutting-edge — and a lot of crazy hype that comes with the expectations of the cutting-edge. I want to share some of the thinking and strategy that kicked off earlier this year in Q1, that’s coming out of the Advanced Analytics and Data Science team here at Virgin Media O2.
Let’s start at the beginning — I mean we could start a lot of places — a lot of data scientists have been using Large Language Models (LLM) for over a decade, I remember using BERT back in Polecat and in Just Eat for trying to understand menus — but realistically this particular story starts sometime around October 2022.
At that time, my manager Alberto Rey Villaverde and I had our doubts — after living through the hype mess of “Chatbots V1” I might have had a lot more doubts than he did if anyone is counting — will this technology live up to the hype? Is it going to be a disruptor we’re all hoping for?
An AI-powered Revolution in the Telecom Sector
Our first understanding was that generative AI, with its capacity to generate new content, designs, and ideas, was going to touch every aspect of our operation. Whether it was customer service, network maintenance, or marketing, we realised that every employee would interact with generative AI in some way. The big question was how profound this impact was going to be.
For some roles, the impact was relatively subtle. For instance, in areas like network operations, generative AI offered opportunities for generating additional failure data for cases where we have limited data.
On the other hand, roles that dealt with massive amounts of data, like mine, or that required quick and accurate content generation, were about to experience a tectonic shift.
We generally agree that we don’t see this as a threat to employment; instead, we view it as an invaluable tool that bolsters productivity and augments our capacity for forward-thinking decision-making.
Taking Generative AI to the Next Level: Four Initiatives That Are Reshaping Our Business
As part of our ongoing commitment to stay at the cutting edge of technology, we’ve initiated several new strategies focusing on generative AI. Each one, while unique in its approach, has a shared goal — to harness the transformative power of generative AI and create unprecedented value for our organisation. Let me walk you through our four key initiatives:
1. Accelerate Engineering
Our first initiative, aptly named “Accelerate Engineering”, aims to empower our tech workers to harness generative AI. Let’s be honest; this move was partly self-serving, as the tech teams we manage wouldn’t let us off the hook if we hadn’t! We’re constantly swamped by demand in this business, as we try to shift Telco to Techco. By giving our software engineers access to this powerful tool, we’re helping them create innovative solutions, boost productivity, and foster a culture of continual learning and improvement.
2. Unlock and Exploit
Next up is “Unlock and Exploit”, our goal here is to ensure that every employee in our organisation has a grasp of generative AI and its capabilities. The initiative is all about demystifying AI and making it accessible and usable across all levels of the company. This move is not just about job security, but rather creating a tech-savvy workforce that can leverage AI to enhance their roles and ultimately improve our overall performance.
Our first move in this space was to invest in a workspace where generative AI can be safely used. This might also be a bit selfish, neither of us particularly want to be on the front page of the Times. We worked with our partners in Data Governance (big thank you to Naureen Hussain and team), to build UIs backed by logging and security scanning which would let us understand and block dangerous content. I understand these tools may not achieve perfection, yet it’s imperative that we strive to optimise safety to the best of our abilities. After all, inaction and complacency could pose even greater risks.
To reiterate, we don’t perceive this tech as a menace to job security; rather, we see it as an asset that enhances efficiency. But to make that happen, we need to get it into everyone’s hands, and not have those hands burned in the process.
3. Top Use Case — Customer Service and Care
Our third initiative, “Top Use Case”, is all about focus. Instead of spreading our efforts thin, we chose a single high-value use case to demonstrate the profound impact generative AI can have on our business. By dedicating our resources and energy to a single problem, we are showcasing the true power of generative AI in creating efficiencies, driving cost savings, and boosting revenues.
We chose customer service and care, and immediately started investing and evaluating the technology for this use case. At the time of writing, we are already doing live trials with GenAI interacting with customers (under the watchful eye of data science and also customer service management) in order to understand the capabilities. More on that in a future article.
4. Evaluating Large Language Models
Lastly, the breakneck pace of AI development led us to start an initiative aimed at developing robust evaluation methods for Large Language Models and their outputs. With Alberto’s insight, we foresaw that every consultancy would soon be selling us on the benefits of their own models. To avoid getting swept up in the hype, we’re developing our own standards and protocols for evaluating these models. This ensures we maintain our competitive edge and make the most informed decisions about which tools and technologies to adopt.
These initiatives represent our ongoing commitment to embracing generative AI and the powerful potential it holds.
Wrapping Up: Generative AI, A Force of Transformation
The future of generative AI in the telco industry is bright, and we are just scratching the surface of its potential. As I continue my journey, I am excited and inspired to see how this transformative technology will continue to reshape the landscape, create new opportunities, and drive our growth.
From subtle enhancements to profound transformations, generative AI has not only lived up to the hype but exceeded our expectations. Stay tuned for more practical articles about how the above initiatives are applying generative AI in Virgin Media O2.
Written by Gilles Comeau, with thanks to Alberto for his tireless motivation, ideas and foresight, and to Kay Schwabedal for his insight and support, to Fortune Banyard and Naureen Hussein for helping keep us safe, and to the entire Advanced Analytics and Data Science team who really are doing all the building here.
