Twitter user ‘Progen’ just released a short video proudly showing AI acting like a digital design assistant for some basic everyday tasks. The A.I. in question is not one entity, but three tools – two of which are open source – working together to deliver a scarily competent service to their human.

The demo is probably the first time we’ve seen voice commands used to create digital art, and it is being run on a humble PC – not a supercomputer in sight. The whole thing is put together from several popular AI tools of the moment. OpenAI’s GPT-3 system ‘understands’ and translates the incoming voice commands, OpenAI Whisper does the speech to text work and finally Stable Diffusion converts the resulting prompt into a piece of art on the screen. Oh and an Unreal Engine game making tool is used to create the AI avatar.

 

AI Takes Over The World

The result is astonishing! With a few simple commands, the user gets the system to create an apple, a rather cute cat in a bottle and a Norwegian house in the winter. This is direct ‘imagination to computer’ magic, with no keyboard inputs or any form of action except a voice command. Sure it’s s.l.o.w., but the latency is something that will obviously disappear as the software and hardware are optimised over time.

Of course the big deal goes way beyond this crude demo. It doesn’t take a genius to see where this is heading. If a desktop computer now has the power to convert commands into useful, if laggy, action, imagine how this is going to go once they iron out the tech and start to refine speed and flexibility.. It’s going to make Alexa look like a Victorian steam engine. Using your voice to create music and videos, manage spreadsheets, write a novel, produce a movie…the list goes on.

We’ve been playing around with Stable Diffusion and a bunch of AI tools over the past few months, and they’re still miles away from creating amazing and reliable results – without a colossal amount of human input to correct mistakes. But boy, this stuff has so much potential. It’s exactly like using the first clumsy Symbian smartphones, and recognizing what the future could hold with more user friendly phones and cool apps.  All A.I. needs is a new Apple to transform the tech into something useful. And it can’t be far away.

Meanwhile, you’ll find us over here, just kicking back and waiting with the popcorn.