![](https://s3.eu-north-1.amazonaws.com/cdn-site.mediaplanet.com/app/uploads/sites/45/2024/06/07162148/Thierry-Klein-Photo-EEC1-May-2023-.png)
Thierry E. Klein
President, Bell Labs Solutions Research, Nokia
Artificial intelligence (AI) has exploded into the collective consciousness with the advent of content creation tools like ChatGPT. However, AI is broader than generative AI and has many other applications.
Artificial intelligence is rapidly transforming how to manage, deploy and develop communications networks — and this is only the beginning of the journey. AI can support networks — and vice versa — throughout their entire life cycle, from design to day-to-day management. This makes them smarter, more efficient, more resilient and less costly to run.
Designing mobile networks to be fully AI-native is already underway in research. However, achieving this will require comprehensive standardisation efforts, likely coinciding with the adoption and rollout of 6G, expected to start in 2030.
AI potential in networks and industries
“Right now, AI in the popular domain has been focused on content creation. We’ve not fully moved it into the industrial domain, which is where we can harness the biggest potential for business metrics, such as productivity, efficiency, safety and sustainability,” says Thierry E. Klein, President of Bell Labs Solutions Research, Nokia’s industrial research arm.
“Networks will become much more optimised and automated, and that’s where we will see performance benefits — as well as reduced energy consumption of the networks and massive improvements on the user experience side.”
In the shorter term, AI tools are already being utilised for energy usage optimisation within networks. “As traffic patterns change, we can analyse the data using AI to change the configuration of the network to reduce energy consumption by up to 30% and still maintain the service and expectations of the customer, and that’s happening today,” says Klein.
Networks will become much more optimised and automated, and that’s where we will see performance benefits.
Why AI power depends on data
Any AI tool is only as powerful as the data on which it’s trained, insists Klein. As impressive as ChatGPT is, it has several limitations. That’s why Nokia has created a large language model (LLM), trained on 60,000 internal documents, which is much smaller but highly tailored to its products and users.
“These LLMs are not going to write you a great poem or cooking recipe, but that’s not their intended purpose,” says Klein. “I don’t think we’ll end up with one large-language model that everybody uses. I expect a collection of models more attuned to specific business contexts, which is far more equitable and democratised.”
There is a lot of potential in collaborating with customers and other technology partners to progress this technology. “By applying information communication technologies to other sectors — including healthcare, logistics, manufacturing and transportation — we see a huge potential benefit when it comes to digitising those industries,” he adds.
AI trust and governance
The future of AI relies heavily on trust, says Klein and, in many of its public deployments thus far, that trust has not come naturally. Like all other new technologies that have come before it, AI has been greeted with a healthy degree of scepticism from the public.
Some have already called for the creation of a regulatory body governing AI. However, the technology is moving incredibly fast and the need for oversight must be carefully balanced with harnessing the transformational benefits of AI and encouraging innovation.
Building responsible AI
Realising this need early on, Nokia Bell Labs defined six pillars of responsible AI: fairness, reliability, privacy, transparency, sustainability and accountability. “We don’t believe in building AI tools and figuring out the responsibility aspect later,” says Klein. “If you leave it until you’ve built the system, you’re very likely to have to undo and redo things, which is time-consuming and costly.”
For this reason, the company has kept its responsible AI principles front of mind from the beginning to guide AI research and development and encourage industry collaboration. “On the technology side, we see a lot of potential in this technology, and we want to make sure we — and everyone — can harness and benefit from that potential in a responsible way,” says Klein.