Machine Learning Invades the Metaverse

After its brief flowering in the early nineties, immersive technology or extended reality (XR) returned with a vengeance in the late 2010s. Its resurgence was down in part to the fact that, second time round, it looked like it could revolutionise not just video games but applications as diverse as product design, architecture, museums and medical diagnosis and treatment. Lockdown gave it an extra shot in the arm as the attractive novelty of virtual meetings and concerts suddenly became a necessity.

Unfortunately, talk of the “Metaverse” turned out to be somewhat overblown. Mark Zuckerberg repurposed (and rebranded) his company around the prospect and invested a lot of money in the sector, most notably acquiring Oculus, but has yet to make that money back. Much of the challenge of scaling up XR adoption is down to cost and practicality – the hardware is still expensive and unwieldy and, as a result, its uptake has — outside the gaming world – been limited.

XR then found itself further eclipsed in about 2022 when the full potential of AI suddenly became generally recognised, and it has been top of the headlines ever since. However, this is not necessarily bad news for XR – in fact, it may give it the boost it needs because AI has the potential to enhance existing R and VR applications and accelerate the development and introduction of new ones. 

For instance, thanks to generative AI capabilities, users will soon be able to generate their own 360 degree immersive content by delivering prompts to the AI tool via their wearable device rather than relying on developer-generated content. AI could also make immersive interaction and personalization much more immediate and versatile. Virtual worlds will be able to adapt immediately to the environment they are interfacing with or to the preferences, movements, instructions – nor even expression or pulse rate – of the user. The potential for XR-driven training, education, research and product design to be exponentially enhanced is obvious.

In many ways, these applications have already arrived. For example, ChatGPT’s Video Mode enables AI to understand video in real time and to screen share with users. One can easily see this trend ending with fully realised, AI-generated XR worlds that are as limitless as the real one and indistinguishable from it and then the Metaverse may have arrived after all.

Of course, this pairing of two revolutionary technologies gives rise to some ethical and legal challenges, not just the already recognised ones for AI and XR but some new ones as well.

Firstly, the increased complexity, versatility and interactivity which AI will give virtual worlds means that the existing legal challenges regarding security, privacy and the scope for physical injury or psychological damage caused by physical injury are magnified. Furthermore, if AI-generated content infringes third party intellectual property or is illegal or damaging in any other way, who will be liable, the licensor or the user? All this will need thrashing out in revised terms of use and privacy policies. 

Sticking with intellectual property, existing XR product licence terms make it clear who, as between the licensor and the user, owns the copyright in the immersive content. However, if one throws AI into the mix, this issue is complicated by the fact that in some jurisdictions, such as the UK, copyright can apply to computer-generated content but in many others it does not.

Finally, any given XR tool will most likely be made available to users in multiple jurisdictions, which means that the licensor will have to navigate different regulatory regimes in different countries, ranging from the regulation-heavy approach in the EU to the lighter touch in Japan and the still evolving frameworks in the US and UK.

So this blending of two of the digital transformation field’s most revolutionary technologies may offer great prospects for tech companies and their customers, but is an added challenge to lawmakers who are already struggling to fit the round hole of regulation into the square peg of ever-accelerating digital innovation.