Drew MacFarlane, Head of GSMA Advance, explains how telcos can build the expertise they need to harness the full potential of artificial intelligence
Now evolving rapidly, artificial intelligence (AI) is set to bring immense benefits to the telecoms industry and the people it serves. As telecoms operators acquire the necessary skills and expertise, AI will deliver new services, improve connectivity and enrich the customer experience.
If AI is applied smartly and responsibly, the economic benefits could be enormous: AI could contribute US$15.7 trillion to the global economy by 2030, according to PWC. But those benefits may not accrue to all: the uneven distribution of expertise and resources means there is a risk that AI will be adopted far quicker and more effectively in certain geographies and businesses than others.
Telcos have been using AI for more than a decade, primarily for cost reduction. But the advent of generative AI, which can be used to automate the creation of text, images, video, software code and other content, ups the ante. Telcos need to build in-house AI expertise to keep pace with the speed of innovation, according to McKinsey, which notes that one telco built and deployed four generative AI solutions in the time it took another telco to draft requirements for outsourcing gen AI use-case development.
Large language models, large numbers of use cases
Investors have high expectations. At its capital markets day in November 2023, KPN said it had identified more than 60 use cases for generative AI ranging from the development of autonomous networks to intelligent service interactions and hyper-personalized customer experiences.
“We now use AI in some 400 projects, Deutsche Telekom-wide,” Deutsche Telekom CEO Tim Höttges told a shareholder meeting in April 2024. “It helps us optimise quality. It brings us closer to our customers. It enhances our productivity by up to 50 percent for routine tasks.” For example, Deutsche Telekom is using AI to reduce the labour required for the structural planning of its fibre-optic build-out by 75%. As it analyses digital images of relevant roads, the system recognises different surfaces, such as asphalt or cobblestones, checks whether trees or streetlights are in the way, and makes suggestions for the optimum fibre-optic route.
Deutsche Telekom has teamed up with SK Telecom, e&, Singtel and SoftBank Corp. to form a joint venture, the Global Telco AI Alliance (GTAA), to co-develop and launch multilingual large language models (Telco LLM) specifically tailored to the needs of operators. The Telco LLM will be multilingual supporting Korean, English, German, Arabic, and Bahasa among other languages.
Of the 100 mobile operators surveyed by GSMA Intelligence in August 2023, 18% had commercially deployed generative AI solutions and a further 56% were testing such solutions.
The AI gap: uneven coverage and adoption
Yet adoption of generative AI varies widely varies across operators and geographies, partly because most of the underlying models have been trained in one of only seven languages. With most training models run by a small number of large technology firms, there is a risk that smaller markets are sidelined from the rapid advances in AI.
It is critical that AI is democratised to ensure that the entire telecoms industry and its customers, wherever they are in the world, benefit. Giving operators access to AI tools, knowledge and expertise is key to achieving this.
To address this emerging challenge, the GSMA has forged a partnership with the Barcelona Supercomputer Centre, which is developing AI resources in minority or underrepresented languages. VEON and its subsidiary Beeline Kazakhstan have joined the initiative, to support the development of AI models and systems for the local languages of the countries where VEON operates.
Helping business leaders get ahead of AI
At the same time, GSMA Advance is working with IBM to run strategic workshops designed to help telecoms executives assess the potential value of both generative AI and conventional AI, and then build business cases for deploying these technologies across various use cases. After attending these meetings, participants will be well placed to accelerate and shape the future of generative AI in their organisations. Importantly, the workshops will also explore the ethical, legal, and regulatory considerations surrounding the adoption of generative AI.
At the GSMA, we are emphasising the need for AI solutions to be designed, developed and deployed in a responsible and ethical way that is human-centric and rights-oriented. As an increasingly essential element of the infrastructure on which our society is built, AI needs to be fair, open, transparent and explainable. To fully protect customers and employees, it is vital that mobile operators ensure there is no entrenched inequality or bias in their AI systems and that they function reliably and fairly for all stakeholders. To that end, the global mobile industry has published the AI Ethics Playbook and a related self-assessment questionnaire as practical tools to help bridge the gap between ethical principles and ethical practice.
Following these principles will help the telecom sector realise both the short-term and the long-term opportunities presented by the extraordinary advances in AI and ensure that all parts of society can benefit.
About the Author
Dr. Drew MacFarlane is head of GSMA Advance, responsible for the learning, training, and accreditation pathways at GSMA for their members and wider sector. An expert in higher education, he trained in psychology and linguistics, lecturing at the University of York before taking up a leadership role in Shanghai to run an education consultancy. After returning from China, he led the QS University Rankings, the world’s largest ranking of higher education institutions, before joining the GSMA in 2024. He has presented at education conferences globally and has been invited to speak at numerous universities across the world.
