“AI is gaining attention: there are all sorts of concerns regarding its application across different sectors, how it processes data, and how it can be aligned with other data processing regulations. It's a bit unruly at the moment, and we're acting on trust in most cases.”

Though this statement feels like it could have been pulled from even the most mainstream news site these past few crazy post-ChatGPT debut months, in all modesty it was something Redmill Communications said…nearly three years ago.

In fact, this observation was made long before all the current brouhaha about all things Artificial Intelligence in the course of the virtual (remember all those?) Telco AI World Summit Europe 2020 Event in ‘London,’ November 2020.

In the same presentation, this warning was also made:

“Another reason for taking leadership [in AI and telecom] now is to protect against crazy conspiracy theories. If you think the public was angry about COVID 5G transmitters, wait until they start to learn and explore the issues of AI-driven processes in the network and what that means for the machines that are processing their data.”

And as we also said back then, we still need to face the twin dangers that the standards community may well fragment and the public is increasingly paranoid about certain new innovations like AI and (still) 5G.

Given the madness of the last few months and the predictions coming at us left, right and centre about the rise of Skynet, the defence rests.

But of course, we weren’t trying to be scaremongers nor Nostradamuses three years ago—and this post isn’t a smug attempt to show off how clever we are!

It’s more about how recent events have only underlined the veracity of the point our contribution to the debate via the conference was trying to make (and which echoed Redmill’s contribution to the IEEE imprint’s important book on all this, Shaping Future 6G Networks: Needs, Impacts, and Technologies).

Which was—and very much remains, this: the fact that AI and automation needs to be brought fully into the standardisation process--and why 6G provides the driver to do so. 

The Wild West of AI needs to start looking more like how we built 5G

Right now, we have an open economy building AI-based systems and deploying them at pace and arguably a little irresponsibly; AI’s not standardised and it’s very much unregulated. But it’s a technology that is of course fundamental to automation—and so, also fundamental to 6G. Indeed, it’s a foundational pillar of 6G: deeply integral to the processes and operational principles being worked on today in Network 2030.

We already know 6G services will depend on AI and the network itself will be built on AI-based processes--specific AI-enabled functions, and chains of these will be brought together dynamically by agile orchestration functions. (To take just one example among many, one of these is bound to include AI in what are known as a Network AI/MF Empowerment Method, or an AIMEM.) These will be orchestrated, as mentioned, and chained to deliver functionality as required by different applications and processes.

Importantly for this discussion, you can't do the things 6G is expected to deliver without AI. Unfortunately, the techniques for delivering AI and 6G don't exist yet. AI must evolve to deliver on the requirements that that body’s trying to formulate.

So, the argument is:

• 6G needs AI to work
• and AI can only really ‘work’ when it is standardised—‘standardised’ here being the best way to make AI trustworthy, which we’ll get to in second.

That being said, of course, clearly 6G doesn't exist yet; it won’t be here until the end of the decade at the earliest. But if we are sensible as an industry and get there the way we (in some cases, eventually) arrived at 3, 4 and most recently 5G, we could be in very good shape and head off a lot of the scandals and fears we we’re now seeing around AI.

That happened so well, if you recall, because there was an IMT—that is, an international mobile telecommunications recommendation by the ITU. Various standards bodies then created a realisation of this. There were multiple approaches to get to the next G in previous generations (all of which met the requirements of the relevant IMT), but 5G was the first to follow a single path and to result in a single set of standards and we'll soon have a new IMT, for 2030.

This will result in standards, but not for a while--but the process is underway, and the industry is already working out the problems to be solved. That’s one pathway to standards safety; the other is AI itself. There are all sorts of concerns regarding its application across different sectors, how it processes data, and how it can be aligned with other data processing regulations.

However, this has been recognised and AI is now subject to an increasing level of standardisation efforts--from the ISO as well as IEEE. Much more importantly,  governments are taking a keen interest; AI has become strategic in the US and China, and the EU wishes to get in on the game, too.

The key problem around this is trustworthiness: How will the data an Artificially Intelligent system be processed, protected and used? Well, we had similar discussions regarding GDPR, EPA and other such standards, and they got solved. There is ample basis for thinking the same will happen with AI—and thus, the 6G we want to see it making possible.

AI might be sketchy today and raise a host of issues, in other words—but it’s a key component of the Future Network, and so will fall within the purview of regulators sooner rather than later. Therefore, standardisation for AI is inevitable—and since it is crucial for 6G, standardisation within this context is equally inevitable.

Let’s ensure we make AI and 6G safe, effective—and profitable

It may not be that smooth. In all likelihood, 6G will probably not follow a single standard path and we think the IMT will stimulate a series of responses—perhaps even a revival of 3GPP2, among other groups.

But the same people who see a role for 6G are also those eyeing regulation for artificial intelligence—giving us all reassurance that the one process will support the other to a mutually (and societally) beneficial conclusion.

To protect innovation and our industry the sector must drive this process--both to ensure its interests are safeguarded, and to prepare for 6G.

We also need to make sure there is concerted effort from across the industry involving all relevant stakeholders to make sure that we standardise to suit our needs before standards are imposed from elsewhere.

So--now’s the time to seize the high ground and look to help sort out the AI that both we need, and the world itself wants to us safely.

Seems like an important job to us.

In the next part of this three-part AI and Future of Telco series, we’ll look at more depth at what’s happening around AI at the future networks level