How carriers are preparing for the onslaught of generative AI | Technology


The technology behind ChatGPT, Google Gemini and so many other similar applications is expected to give rise to a market of 1.3 trillion dollars in 2032. It is the so-called generative artificial intelligence, which in 2022 moved a volume of 40,000 million dollars, and will grow at an overwhelming rate of 41% annually, according to Bloomberg Intelligence.

In this explosion to come, the companies best positioned to obtain economic benefit are AI service providers. OpenAI, Google or Microsoft will bring their products to end consumers. But they will do it through the networks of the operators, who do not want to be left out of the distribution of the pie. Other times, with the explosion of video on demand, social networks or mobile applications, the entire volume of business has moved from end to end of the network, without giving up anything along the way. The question is whether the same will happen with artificial intelligence.

The increase in the use of generative AI could also bring an increase in network traffic. In a study on the sector of telecos, the IDC consulting firm It expects the amount of traffic to skyrocket and speaks of a significant impact on infrastructure capacity and latency. The most basic case: a ChatGPT query generates more congestion than a Google search

For now, the operators have not noticed anything out of the ordinary. “The growth of data traffic on our networks has been a constant for more than ten years; it grows 30% a year,” says Bernard Despres, vice president of core network, automation, security and E2E services at Orange. “I would say that it is not difficult to manage the increase in volume, it is more complicated when there is more diversity in the type of data,” he adds, referring to the variety of traffic generated by different services and applications.

Julia Velasco, Network Director at Vodafone Spain and director of the group’s Data Network for Europe, agrees that traffic was already growing a lot without generative artificial intelligence. Although, she points out: “When you look at what types of use cases could be developed faster with generative AI, there is a very important part of image processing and generation. Is this necessarily going to lead to an increase in traffic on the networks? I understand so,” she points out in reference to specific use cases, such as identifying threats in security camera images, which would lead to localized traffic spikes.

Benefit from AI services that pass through networks

It is already a traditional demand of operators that the large platforms that provide services through their networks, such as Netflix, Google, Spotify or WhatsApp (the so-called OTT or companies over-the-top), share income. Among the arguments they put forward is their need to make costly investments in infrastructure, to which they are asked to contribute.

In order to obtain greater economic benefit from their networks, operators around the world have launched the initiative Open Gateway. This system standardizes programming tools, APIs, so that an application can access specific network capabilities. A formula designed for services with more experience, such as video on demand, but that could also be applied to generative AI tools.

“An application could optimize its service or add value based on capabilities that the operator could dynamically activate or deactivate on the network. This could be exposed through APIs,” explains Velasco. “If your application, as a developer of a generative artificial intelligence solution, improves with that additional capacity, I think that now we operators will be in a much better position to monetize our services.”

There are still no APIs intended for generative AI services, but Orange also believes that the Open Gateway framework would allow developers of this type of applications to obtain income. “What we have today is an OTT model, but thanks above all to 5G we have very advanced capabilities, such as very low latency, high speed or great reliability. We believe that we will evolve this OTT model, because the applications will need individualized quality. It is about giving application developers the possibility of having access to advanced capabilities of the 5G network. We are doing this through the Open Gateway initiative,” Despres details.

The Orange manager illustrates what these advanced capabilities could be used for: “On Netflix, for example, you can have signal delay. No problem. Instead, the application needs to deliver high-quality video. For its part, DAZN or anyone who broadcasts a live match needs a real-time service, because the user does not want to see the goal two minutes after their neighbor.” In this way, each application could use the network APIs that best suit its specific use case.

“As teleco, I can offer a product based on generative AI or I can look for how to become part of the value chain linked to this technology,” says Velasco. “And here come concepts that would be supported by Open Gateway to expose those possibilities. I expose developers of generative AI solutions to the APIs that allow them to optimize their systems through what the network offers. I believe that initiatives like Open Gateway will allow us to not be a mere support for this technological evolution.”

Most large operators are immersed in Open Gateway. Also Telefónica, although this company only points out that there are still no specific APIs for generative AI applications. On the other hand, they do take advantage of this technology to improve their networks. Carolina García Sánchez, head of AI at Telefónica Global CTIO, says: “We are trying to build chatbots or assistants to help technicians know at all times what is happening on the network and how they should solve it.”

In network operation, generative AI can be used for management and even anticipation of incidents. “Another thing we are exploring is the possibility of using generative AI as a natural language interface to directly ask the network what is happening,” says García Sánchez. Here traditional artificial intelligence monitors the connections, while a generative model is used to make the query.

Both Vodafone and Orange also have programs that explore this use of generative AI to resolve incidents on their networks. After all, operators will not only be technology intermediaries. They will also be recipients. The McKinsey consulting firm estimates the potential impact of generative AI in the global telecommunications industry in a range between 140,000 and 180,000 million dollars.

You can follow The USA Print in Facebook and x or sign up here to receive our weekly newsletter.

Subscribe to continue reading

Read without limits


Leave a Reply