Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Open Channel: Can Artificial Stupidity Save Us From AI?

Will AI will ruin music production? Craig Anderton suggests that the only way to beat artificial intelligence is with artificial stupidity.

Craig Anderton
Craig Anderton

There’s a herd of elephants in the room these days, but let’s talk about the one that many people won’t even admit exists. The one that makes people say, “Oh, that could never happen.” The one that, so far, has been sitting quietly in a corner as it grows from being a baby elephant into an adult.

Yes, the elephant that can put an end to our industry as we know it. The one that will make you look nervously over your shoulder as you revise your resume. The one that smiles and says it’s here to serve us, and believes that it is. The one we’ve named artificial intelligence.

Oh, but we don’t have to worry about that…we’re creative types! We base what we do on experiences that no machine could have. Machines don’t laugh, cry, fall in love, take mind-expanding drugs, or fall into the depths of despair—then turn their lives around, rise like phoenixes and go on Oprah to talk about it.

But thinking that machines can’t replace us ignores the fact that they already are. Need a narrator? Advanced speech synthesis sounds are not only indistinguishable from a human voice, but a perfect human voice. With perfect pacing, pronunciation, timing, the ability to recognize context, and pitch variations that add expressiveness. And the perfect narrator never mouth clicks, inhales, uses poor mic technique, p-pops, has a sore throat, burps or wants to unionize. But surely, this AI narrator couldn’t do James Earl Jones? Of course it can. Just not quite yet.

You might think that, thankfully, you still need a human to write the words the narrator will say…right? But now algorithms can write draft ad copy. Feed in key words and word count, and out comes generic copy— which isn’t that different from human-generated generic copy.

And even though we have samplers with incredibly accurate emulations, someone still has to write the music, yes? For now. Maybe. But not in a few years, when you tell the machine “action movie chase, 1:32 long, peak at 1:06:45, end with big sound, symphony orchestra.” If you’ve ever played with Sony Creative Software’s Cinescore program (which mysteriously disappeared shortly after its introduction), you know that algorithmically generated scores, albeit primitive, are already possible.

Then again, there’s always the final comforting thought: Okay, AI can generate stuff with the look and feel of real words or music. But humans know when something’s fake, and, ultimately, they’ll prefer reality.

Dream on.

Open Channel: Warning—Stability Ahead

In Korea, virtual influencers are taking over. These AI-created “people” post on Instagram, interact with followers, travel to exotic locales, model fashions and, of course, do product placements. Many of them are gorgeous virtual women with dewy eyes, perfect skin and come-hither expressions. Companies love these influencers because you can create them in a day or two, then do commercials in a few days that don’t require makeup, hair styling, lunch, lighting, or any of those other messy things humans need. The virtual humans are perfect—they don’t talk back or ask for residuals. They release perfect singles, sung perfectly. And they not only “want to be forever young”…they are forever young.

For a generation raised online, whose window to the world is a smartphone display, it makes perfect sense. When that window to the world becomes a high-res VR headset, it will make even more sense.

As to what this means for our industry, let’s be honest: We sowed the seeds of our own destruction. We aimed for “perfect” recordings that prioritized timing and focus group reactions over lightning in a bottle. We acclimated people to thinking that composite, phoned-in parts by groups that never existed are music.

One-hundred-percent synthesized music soundtracks support CGI-based movies (sometimes with re-animated dead actors). We’ve accepted software that analyzes mixes and makes edits based on a database of hit music. We have mastering services that master via algorithm, and singers who chase unattainable standards of beauty through botox and body shaping. We sent the message that perfection is the goal. And now, machines can deliver that perfection. So, revise your resume, because you’ll never be as perfect as what AI can produce.

Except…we have one big advantage over AI because of something it can’t do: Fail.

Open Channel: The Subscription Kerfuffle

Yes, fail. Failure can spur pushing the envelope, innovating and questioning the status quo. Starting with perfection eliminates the crucial process of pursuing perfection. That process includes failures, which lead to learning. Strings of failures can put in motion successes that would never have happened otherwise. Failures of old models give birth to new models.

Sure, AI can refine conclusions it makes that aren’t perfect and “learn” from them. But it can’t fail in the way humans can, or understand the ramifications of the resulting unpredictable consequences.

So, the only way to beat artificial intelligence is with artificial stupidity. Do dumb human things just to see what happens. Make mistakes, embrace your flaws and fall down. So what if you fail? Only then does the interesting part happen: You climb your way back up.

Over time, the sum total of those mistakes, flaws, overreaching attempts that became epic fails, and ultimate redemption, will coalesce into something uniquely fascinating…and human.