Down The AI Rabbit Hole
I'm not obsessed, you're obsessed.
One of my favorite video games lately is Midjourney. People often refer to it as an AI art generator, but to me it’s the most elaborate and full-featured character creator in existence. It also builds wonderful landscapes—castles, bogs, cyberpunk metropolises. You name it. Ethical concerns aside1, it’s really astonishing. Unlike ChatGPT—which is also quite neat—the images you can create can be beautiful and evocative. ChatGPT’s writing is almost always soulless and bland, but if you’re any good at all with Midjourney, you can make art that has emotional resonance and depth. Or perhaps we should say “art” here.
In any case, I’ve gone even further down the AI rabbit hole getting into some AI animation and voice-work. Here are a couple of the videos I’ve created.
The Hitman
I made this guy in Midjourney and then plugged his face into D-ID, an AI animation software that turns static images into speaking faces. I recorded a video using the software and one of their AI voices, then stripped that voice out and recorded my own, dubbing that into the video along with some royalty free music. (I’ll have to try AI music generators next).
I didn’t use an AI voice for this video because I just figured it would be easier and better using my own. Check it out:
The Wizard and the Halfling
Next, I took this wizard and typed out some lines for him in D-ID, recording it with one of their AI voices. Then, I created a sample of an elderly British man’s voice in ElevenLabs and used that software to record the same lines. I edited these together in Adobe Premiere Elements and then decided to add the halfling. I didn’t like the AI voice from D-ID and Eleven is really bad at British accents (I lucked out with the old guy) so I just recorded my own voice for the halfling lines and then sped it up to make it higher-pitched. I put all this together and added some music and an intro shot of a castle. Behold:
I’m curious where all of this leads.
In the news, AI is making waves with this new Drake song, Heart On My Sleeve, featuring AI versions of Drake and The Weeknd. It’s an instant hit, and people are still having a hard time believing it’s AI (and there are valid theories that this could be a stunt by Drake and UMG):
Others are already making new knock-offs of this song:
Pretty wild.
Here’s my theory about the future: Someday, all of this will be so quick and easy that people will just sit around and create their own movies and games and animation and whatever they want and most of it will be absolute garbage, but some will be really good. You’ll have one-person teams coming out with full feature films and TV shows, AAA video games, the whole shebang. I don’t know how to feel about that.
On the one hand, it’s kind of exciting. Those little videos above were fun to make. Time-consuming to put it all together, but not as time-consuming as drawing and animating it all by hand—something I’d never be able to do, anyways.
On the other hand, it’s hard to know if this really democratizes art and creation or if we just get new corporations making money off user-generated content at the expense of real artists and creators. I suppose it’ll be a bit of both, in the end, but there’s no turning back time. This genie has sprung. Be careful what you wish for.
You can check out the software I used in the above videos at the below links:
Midjourney / ElevenLabs / D-ID
Yes, I agree that this software is engaged in copying (or stealing?) other peoples’ work and then remixing it, and yes that has ethical implications that should be explored and dealt with somehow. Of course, I think we all steal and call it inspiration. This is a bit more like DJing with images rather than writing your own songs.