- read

Headless Metaverse Simulations, AI and safety

Steve Jones 58

The biggest use of the “Metaverse” isn ‘t consumers with big headsets on looking around a 3D space. That is the marketing side of it, and it’s the “cool” part that will be beloved of TV programs. But it won’t be the biggest thing, the biggest thing in the Metaverse will actually be a ‘headless’ metaverse. That is a pure digital and contained metaverse with no required visualisation.

Think for a moment about a computer game, one with loads of non-player characters (NPCs) but properly ‘smart’ NPCs, NPCs who are based on models of how people would really behave. Metahumans which look like humans, and through AI can act like humans within proscribed environments. Now think of the city in which they live, the buildings, the roads, the power grid. All of this can be digitized, creating a digital mirror, the meta-mirror, for that city.

Using this you can now run millions of simulations of how people can move around the city, based on the individual AI for each meta-human making decisions learned from actual behaviours. How are they flowing on public transport, if you have an event how does it alter the patterns. If a retail store changes its signs does it effect how people view the store? Depending on how sophisticated the metahumans can become the ability to run simulations of reality would become a new way of forecasting business outcomes. This is not tomorrow, but it is the future of using AI to simulate cities and business strategies. Metaverse not as a visualisation, but as a way to test theories against a simulated reality.

This means being able to test the AIs that you build in a perfectly safe environment, one that is enclosed and where you can simulate billions, indeed trillions of conditions as you train and test your AI. One where you can pose the trolley problem in a billion different ways and see how the AI reacts. One where you can simulate bias in the metaverse and see how your AI reacts. What this means is that the goggles and VR headsets are just a part, and I’d argue a small part, of what the metaverse will mean to businesses and consumers in future. There is much to do, and many problems to solve, not least the power consumption challenge.

To the philosophical question “Are we living in a computer simulation?” there is an underpinning assumption: that someone has created that simulation. We are a stratospherically large distance away from “The Matrix” but there are already areas where a headless metaverse could deliver significant value.