A quick mental exercise: Imagine electricity were just discovered, and during the residential rollout, there were no rules. No fire safety measures, no voltage standards, no common size for outlets.
I imagine there’s one image that comes to mind. FIRE! Everywhere.
That’s kinda what we’re hearing from some quarters about artificial intelligence right now. We dare not hinder development of AI with regulations and questions, lest we fall behind … in some sci-fi-inspired race to save humanity by destroying it, I suppose.
The analogy is imprecise, but the sentiment is real. AI-branded tech is being shoved into every kind of work right now, and as is often the case with Big Tech, valuations count much more than consequences.
Of course it often helps to explain the macro by storytelling the micro, so I leapt at the chance earlier this year to begin work on a podcast series for Duke University about the impact of AI on the music business. The final episode just dropped.
We have a great messenger for the topic — Tift Merritt, a Grammy-nominated folk singer who is also a music scholar and a very shrewd businesswoman. A credit to her industry, she’s taken a keen interest in the future of … gosh, I hate to type this phrase … human-created music. Particularly folk music, which tends to serve as the conscience of whatever culture it belongs to.
“I think very much of myself in terms of a lineage of people who came before me and a lineage of people who will take the stage long after me, and how I can do my best to care for the folk musician tradition in the moment that I inhabit. And that is carrying the news of the day,” she tells me in the fourth and final episode. “That is carrying the emotional message of being a woman in this moment. That is reacting politically to what’s happening about me, and hopefully igniting in other people the song that is within them. So I’m not sure that technology that’s training on backwards-facing material is going to be able to have the reaction that is really deeply embedded in the forward-facing moment and the human connection that we all hunger for.”
In other words, AI is not going to write great protest songs. Neither will anyone else, if AI-generated music puts actual musicians out of business.
This sounds like hyperbole; it’s not. Big Tech has already figured out how to siphon off big profits from the music industry, leaving precious few crumbs for today’s emerging musicians. Crumbs, as in 0.003 cents per stream. But that was only step 1. Those partial pennies will sound generous to bands who get nothing as AI-generated music has begun crowding their art out of listeners’ playlists.
By now, I hope you’ve realized, this is the same basic threat facing nearly all labor.
Fortunately, there is nothing inevitable about this dark future. Tift Merritt is working with Duke Professor David Hoffman at Duke’s new Trustworthy Technology Lab.
“Luckily, we have people studying this, and we know that there are solutions…(We are) taking a look at different public policy interventions that we would be able to make that would preserve some of the things that we really value out of music, and thinking about that as a canary in the coal mine for what we see in lots of other sectors of the economy,” David says in the final episode. “We actually know how to go and protect labor, and to protect the listener, and to protect society here.”
The series is wrapped around Tift Merritt’s new album, called Sugar — it’s her first studio work in a decade, which makes her an excellent witness to “things that have changed.” In a nod to her album, the series is called “Sugar High” — the kind of fake energy that bad food gives you, or bad music.
I hope you’ll give the whole series a listen. Tift is an engaging interview and her music is great. Below I’ll paste links to all four episodes and transcripts. But I’d like to leave you with an important nugget from my conversation with Prof. Hoffman, who is also executive producer of the series. If you care about AI or music, this is for you:
David Hoffman: Bob, maybe the one other thing I would add is I think we would be very interested in continuing to collaborate with people who have a skeptical approach to this call for no regulation whatsoever for artificial intelligence.
We’ve seen with some of the recent releases the potential for significant impact on critical infrastructure and on people in general. We’ve never, as far as I can tell, said that for a new area of technological innovation that’s going to have that kind of impact that we shouldn’t regulate at all.
Coming back to the food analogy, can you imagine if we were to say for either food or for the pharmaceutical area, “No, we don’t wanna chill innovation, therefore, we’re not gonna regulate at all the food that you eat or the drugs that you have to take”? That would obviously be insane. I think it’s actually equally insane that we’re taking the card that current approach with artificial intelligence.
Bob Sullivan: It was just striking me. What if we rolled out electricity without any rules? How would that go?
David Hoffman: Absolutely. Sam Altman recently came out and said that he actually believes that the future for these models is that they would become a utility, and basically serving up knowledge and thinking to individuals like they receive power.
I wonder how much he’d actually thought through that then the natural result of that is we regulate utilities, and we regulate the cost of utilities on individuals and the safety of them. I’m not saying that I necessarily agree that’s the foreseeable future, that, number one, we will go or that we should go, but I think it does show the lack of clear thinking right now about how we’re supposed to have governance models for this technology and to hold the powerful people who are advancing them accountable.
Home page for the entire series | Ep 1 | Ep 2 | Ep 3 | Ep 4
Be the first to comment