Day 6

Automate This! How A.I. is taking over the entertainment industry, from pop hits to blockbuster films

The A.I. revolution has come to factory floors and operating rooms. Now, it's taking on the entertainment industry — with consequences that could reverberate far beyond Hollywood.
Listen12:13

Welcome to the third instalment of 'Automate This!,' a Day 6 series about the future of work in an artificially intelligent world.

You might trust an algorithm to build your Spotify playlist. But would you trust it to compose the songs?

If you answered 'no,' you might want to reconsider. Because musicians like Philippe Pasquier are already working to make it possible.

Philippe Pasquier demonstrates one of his creative computer programs at Simon Fraser University. (Anne Penman / CBC)
In his windowless computer lab, also known as the "Black Box," Pasquier tinkers daily with computer programs whose musical compositions can already compete with the likes of Chopin and Brahms.

A musician and artificial intelligence researcher based out of Simon Fraser University, Pasquier works in a field called "computational creativity."

In other words, he teaches machines how to perform creative tasks.

"The main thing people do with computers is organize pictures, prepare video montages and play video games," Pasquier says.

"So that's really what we try to do in the lab, is to empower users to create better; create faster; and eventually create different things than they would without those technologies."

Pasquier believes A.I. is a source of untapped potential for the creative sector.

"Provided I have my cell phone, there are a number of things I don't even bother memorizing anymore — so I've been offloading some of my cognition to this machine, [and] I see no reason why we wouldn't do the same thing for creative tasks."

But while Pasquier is optimistic about A.I.'s potential to enhance creativity, it may well throw the entertainment industry into disarray.

Live DJs are already feeling the effects of music industry automation. (Tavits Photography)

           

No industry left behind

Machine learning technology is already transforming workplaces around the globe — from the factory floor to the operating room.

Researchers at the University of Toronto's Mowat Centre say as many as 7.5 million Canadians could lose their jobs to automation in the next decade alone.

And according to Nick Bilton, the creative sector will eventually face a similar disruption.

"I think it's gonna touch everything ... the actors, the musicians, the singers, producers." - Nick Bilton, special correspondent for Vanity Fair

Bilton is a special correspondent at Vanity Fair. And he says the entertainment industry is totally unprepared for the changes that are coming.

"I think it's gonna touch everything — from the actors, the musicians, the singers, producers," he says. "When you think about a movie, or a musical number or something like that, the number of people involved can sometimes be in the hundreds."

"And I think in the future, the number of people involved could be in the double digits rather than the triple digits."

Creative sectors like film and music won't be immune to disruption by artificial intelligence. (Tristan Fewings / Getty Images)

      

Artificially intelligent pop?

In the world of classical music, computers are already writing songs that are indistinguishable from human compositions — at least to the untrained ear.

When Pasquier tested his computer programs' best imitations of famous composers against the originals, he found most people couldn't tell the difference at all.

See if you can tell which of these musical excerpts was written by a computer and which one was written by Chopin. 0:55

Pasquier isn't the only one building computer programs that can compete with Chopin. California-based composer David Cope has spent the past 30 years programming computers to write their own music.

Perhaps unsurprisingly, the technology has also caught the eye of the mainstream pop music industry.

Last year, Sony released its first A.I.-generated pop tune, a Beatles-inspired track called "Daddy's Car."

             

The track was composed by a machine, but it wasn't totally void of human influence. French composer Benoît Carré penned the lyrics and arranged the song.

That leaves room for skepticism from industry watchers like John Seabrook, author of "The Song Machine," who argues computer algorithms will ever be able to create a true hit.

"A.I. is definitely going to change the way we experience music, the way we discover music. It already is. [But] the one area which I think machines will not automate is the actual writing of hits," Seabrook says.

"That's kind of what being an artist is, really, ultimately — is taking off in some direction that is not the one that the data is necessarily telling you to take. That's what artists do, and I don't know if that is what an A.I. could ever do."

That's kind of what being an artist is, really, ultimately — is taking off in some direction that is not the one that the data is necessarily telling you to take.- John Seabrook, author of "The Song Machine"

But while "Daddy's Car" never made it to the Billboard Top 100, Bilton believes it's only a matter of time until we see pop hits that are entirely penned by machines.

"In many respects, pop hit-making today is [already] formed by machines in the fact that artists like Lana Del Rey … are manufactured," Bilton says.

"They're not these really creative people that come up with this form of music; they're a bunch of marketing people that come up with a concept and create someone like her. And that happens in the music industry every day."

"And I think that [if] you give them an opportunity to use computers and algorithms to help them do that, you can be sure they're going to."

Pop artists like Lana Del Rey are already manufactured, Bilton says. (Alberto E. Rodriguez / Getty Images for CBS Radio Inc)

Bilton sees myriad other possible uses for A.I. technology in the music industry.

"You know, you could imagine doing what we call 'A/B testing.' So you create different versions of a song, and you can imagine that you create 1,000 different versions of a song that you have written with some sort of artificial intelligence that's helped you do it, and you put it out there to see which one people like the most."

And he says it could eventually go even further.

"Maybe you don't necessarily need actual musicians. You could start a band with artificially intelligent versions of them."

           

Hollywood, interrupted

Movie sets could also look drastically different in the future, according to Bilton.

"Actors are not going to be needed in the same way that they are today," he says.

"There were two people in the last Star Wars [film] that were not real actors; they were computer-generated versions of that. So what if you could do that with everyone?"

Bilton says it's possible that in the future, movie stars like Jennifer Lawrence might just license out their 3D image to a production company rather than actually appear on set.

"I think that there will be another world where you are watching a movie with Jennifer Lawrence, or even people who have passed away — maybe Marilyn Monroe, or something like that — and it's all computer-generated and done by artificial intelligence."

Thanks to big data and A.I. technology, Bilton envisions a day when the films we watch at home could be custom-edited to suit our needs.

"Imagine that you come home and you've had a long day at work. And you say, 'I want to watch a comedy that's 45 minutes long because I have to be up early for a meeting. And I want it to be based in New York and have two female lead actors."

"Those are the things that you will be able to do."

Actor Jennifer Lawrence arrives on the red carpet for the film "Mother!" at the Toronto International Film Festival (TIFF), in Toronto, Canada, September 10, 2017. REUTERS/Mark Blinch - RC138545C630 ((Mark Blinch/Reuters))

                

From fake actors to fake news

Already, companies like Adobe are building advanced audio- and video-editing software that could eventually help reduce the need for live actors on set.

One such program, which they demonstrated at a live conference event last year, is a version of Photoshop for audio that allows users to manipulate audio recordings and change the words that are coming out of someone's mouth.

"You could take … 10 to 20 minutes of audio, you put it into this machine, and then you can type words and it will speak them in a way that you literally cannot tell that that person did not say those things," Bilton says.

In Canada, an A.I. startup called Lyrebird is developing a similar software program for audio editing that would allow users to create a digital version of their voice.

Video-editing software is making similar advances. Researchers at Stanford University have developed a program that allows a user to manipulate footage of public figures, changing their speech and facial expressions in real time.

The technologies are intended mainly for commercial purposes, such as film editing or advertising.

But Bilton worries that as these programs become more common and inexpensive, the consequences could be felt far beyond Hollywood.

"In the next 2018, 2020 elections in the U.S., you're gonna have not just fake articles, but you're gonna have fake videos of Vladimir Putin and Donald Trump talking that are gonna be created on computer-generated technologies — or fake audio from fake radio transcripts that never happened," Bilton says.

"Once you control that, you can control anything."

      


Click here for Part OneTwo and Four of Automate This!, our special series about the future of work in an artificially intelligent world.

For the full interview with Nick Bilton, download our podcast or click the 'Listen' button at the top of this page.