I'll admit, they're catching me more often lately on social media – newsreel-style videos, dancing Donalds, Michael Jordan saying stuff that Michael Jordan never said, influencers that never existed. I'd pride myself on being reasonably savvy, but the telltale signs of what's generated and what's real are starting to blur.

And as a reeling world struggles to get to grips with that, here comes the next salvo: real-time AI-generated video. Runway has teamed up with Nvidia to present an as-yet-unnamed video model that's under 100 milliseconds from prompt to first frame. The blink of an eye takes between 100-400 ms, for reference.

Demonstrated at Nvidia's GPU Technology Conference, or GTC, in San Jose yesterday, this is the first AI video generation tool that genuinely seems to start streaming instantly in high-definition. So what does that mean?

Well, on the good side, it's definitely a step towards the holodeck-style convergence I've been banging on about for years now, in which we'll be able to interact in real time with characters, situations and worlds of our own choosing. The ultimate video game experience in virtual reality, delivered frame by frame, à la carte, in response to your every whim. Indeed, Runway's one of many companies working on exactly this idea through playable world generation:

On the bad side ... Well, everything that's now taking you a second glance to spot as AI will at some point start coming at you in real time, tailored to convince or persuade you, generated using real images as a starting point, and potentially capable of reading and responding to your body language in real time. Such a strange world we're constructing for ourselves!

Either way, it'll be a minute yet. Runway's demo was running on Nvidia's Vera Rubin – an AI-focused supercomputer running 36 Vero CPUs, 72 Rubin GPUs, 54 terabytes of CPU memory, 20.7 terabytes of GPU memory, and more teraflops than I'd be comfortable poking a stick at. This thing can probably run Crysis.

So it's not in the hands of the spammers and scammers yet – but it's absolutely within reach of governments and corporations, and hardware limitations don't tend to remain limitations for long.