Artificial intelligence has ascended to the top of minds in the telecoms sector over the past year due to its potential to transform the industry.
Much has been touted about the endless possibilities of the technology. At the recent AI Safety Summit hosted by UK Prime Minster Rishi Sunak, controversial technology entrepreneur, Space X and Tesla CEO Elon Musk brazenly claimed in an interview that through the evolution of AI “there will come a point where no job is needed, you can have a job if you want to have a job for personal satisfaction, but AI will be able to do everything.”
Such a lofty (or frightening) future is a long way off before we can see even an inkling of the future Musk painted. Work with AI has already begun in the telecoms sector, as operators are looking to streamline surface-level operations such as marketing and customer service. The long-term goal is to have it reduce expensive operational expenses, and deliver new and profitable use cases.
As of now, operators are getting to grips with generative AI from the overall AI toolset, the technology behind Open AI’s ChatGPT. In following ChatGPT’s lead, operators have moved to upgrade their own chat bots with generative AI to deliver a more ‘hyper-personalised’ experience.
But is it the right time for AI to come to developing markets? A key difference between operators in developing and developed markets is ARPU. Economic prosperity of course delivers more spending power to everyday consumers, meaning those lucky enough to live in the UK for example are able to upgrade to Apple’s latest and greatest iPhone annually - and fill operator purses.
No option
At Futurenet Asia in Singapore last month, this challenge was posed to Axiata Group Head of Analytics Ahmed Saady Yaamin, who argued operators from developing markets with low ARPU have “no option” but to rely on innovative new technologies such as AI to improve operational efficiencies and increase profitability, despite the high cost of deploying AI.
“In a high ARPU market, by design, the market is safe, as the profitability is there. But in a low ARPU market it's a volume game which is basically an efficiency game. In those markets there is a necessity for invention as we can’t expand the ARPU. We are left with no option but to rely on innovation to find a way to improve profitability,” said Yaamin.
Speaking at Digital Transformation World in September 2023 in Copenhagen, MTN Chief Information Officer Nikos Angelopoulos said although AI can deliver operational cost savings, its main objective should be finding new use cases that will deliver growth. He pointed to how MTN had invested heavily into Big Data which enabled the operator to “extract hundreds of millions of dollars from customer value management”, and he sees AI as having the potential to have a similar effect.
Angelopoulos said MTN tested AI to offer personalised products to subscribers and equipped chatbots with language detection, but he noted the costs for AI infrastructure are expensive, presenting a barrier for operators in developing markets.
“In one market, to assure performance for the volume of customers, we're talking about easily a seven-digit figure in US dollars for the infrastructure and that has to produce three to five times the return in order to justify the investment. We are still learning, I think it's going to take longer than people like to believe. But I also think AI is exciting because of the creative use cases that will form, cases we cannot even imagine yet,” said Angelopoulos. “AI is an opportunity to enable use cases that can help us grow the industry in ways we couldn't otherwise do. I think understanding product development will be key and by doing that, we can really transform and revolutionise the industry using AI.”
Challenges
Speaking to Developing Telecoms, ABI Research senior analyst Reece Hayden said generative AI can be used effectively in marketing, billing and other internal operations. However, expectations should be reined in due to the looming challenge of costs.
“Generative AI can do some very impressive things. If deployed effectively it is useful. From our perspective, it's not really going to have as many use cases as telcos expect. Instead what we expect is in the next year, telcos will get very excited about generative AI and they'll deploy it in a couple of use cases, and quickly realise how expensive it is to deploy and run. Then they'll move towards a more cohesive system, which combines different AI frameworks to effectively build out both the cost proposition and solve use cases effectively,” said Hayden.
Hayden noted the true value of AI lies in infrastructure equipment vendors creating applications that “target telco transformation”, and they are doing so as AI has that “potential return on investment”.
“Ecosystem players are coming out with their generative AI solution where you can blend the data coming in. Operators will be able to transform in-house data and leverage that to deploy AI within their back office or even somewhere within network operations. But if an ecosystem doesn't build up around AI, we could have another situation where it is similar to the metaverse or VR/AR, where it won't really take off aggressively after a lot of hype,” said Hayden.
AI is indeed being embraced by the telecoms sector more than the metaverse ever was, Hayden pointed out, due to its lower barrier of entry. This has enabled OpenAI to progress from ChatGPT 2 to ChatGPT-4 Turbo in five years, and Meta’s Llama AI platform to evolve to Llama 2 in half a year.
“That speed of innovation is much faster in generative AI and AI in general, than it is in equivalent ‘hype’ technologies of the past. That's simply down to the accessibility and the democratisation of AI which is much clearer to us.”
Hayden warned if operators from emerging markets don’t eventually tap into AI then “what will start developing will be a bigger wealth divide between developed and developing countries, especially in the operator realm.” He advised emerging market operators, in preparation, to first begin looking at sorting out their data to effectively train and leverage AI software, as well as upgrading back office systems.
AI is indeed a technology to get excited about and is spurring operators in emerging markets to throw their weight behind it. But more development is needed to see if AI will have more of an impact than other buzzword technologies that faltered, like the Metaverse. However the initial impressions and activities of operators in AI compared to the Metaverse and AR/VR, there seems to be much more momentum from the technology that Stephen Hawking once predicted would be the best or worst thing for mankind.
