Could an A.I.-Generated Cary Grant Win an Oscar?

Could an A.I.-Generated Cary Grant Win an Oscar?

One of the great ironies of Mary Wollstonecraft Shelly’s 1818 novel Frankenstein is that the title has often been used, erroneously, to refer to the ‘monster,’ or creature as it’s called in the story. The title actually refers to Victor Frankenstein, the story’s protagonist. In the tale, he’s an arrogant scientist who reanimates a corpse in his quest to play God. And if there’s any monster in the tale, it’s him. He’s not only an egomaniac, but he abandons the creature once he realizes he’s been ‘born’ hideous-looking and with the functionality of a four-year-old. Rejected and alone in the world, the poor creature eventually becomes self-sufficient but also vengeful, wreaking havoc on the society that shunned him.

It’s not just a horror story, it’s a tragedy.

I could not help but think of Frankenstein and his creation when Open A.I. CEO Sam Altman appeared before a Senate committee in May to cajole the body to pass legislation which would give A.I. (artificial intelligence) a wide swath of freedom as the rapidly evolving technology becomes more and more ubiquitous. Despite Altman’s enthusiasm, many senators on the committee expressed concern that such technology could too easily be used for nefarious purposes. One even went so far as to compare it to a danger as significant as that of a nuclear weapon.

Such fears are becoming more and more prevalent when it comes to contemplating what A.I. will be used for. Those in favor of A.I. make the argument that you can’t hold back technology, and they’re right. Of course, but A.I. isn’t going to stop at helping shopping websites understand a customer’s preferences or tracking fraud prevention for the government.

Its implications in the arts field are already quite worrying as A.I. can vamp visual and writing styles, musical composition, and even human voices and faces. The technology has already been used to duplicate a voice perfectly from little more than existing sound recordings. It can put a different person’s head on someone else’s body almost flawlessly in any video. And it can write sales copy based on previous ad campaign prose, potentially eliminating all kinds of scribes. With A.I., the artist of whatever stripe is in jeopardy. Indeed, there is talk of A.I. being able to write entire stories or scripts, let alone ad copy, given the proper input and supervision.

Sounds rather catastrophic, no?

The more that senate panel probed in May, the more Altman was obliged to agree with their suspicions, particularly when it came to using the tech to replace artists, create fake news stories, challenge copyrighted original content, and interfere with all kinds of lives and livelihoods. Toward the end of Altman’s testimony, he was forced to admit that the genie was already out of the bottle. Altman even went so far as to ask the senators to be extremely diligent about monitoring and controlling his industry.

Mr. Altman, meet Doctor Frankenstein. You two have a lot in common.

The potential for A.I. to upend the world has been a worry for decades already. Computer technology started replacing humanity in all kinds of ways back in the 1950s, and it’s always come tinged with danger. In fact, the computer has become such a threat to mankind that it’s been the go-to villain for screenwriters for decades. Be it 2001: A Space Odyssey, Westworld, or The Terminator, to name three anti-technology films, A.I. is rarely portrayed as a force for good on the big or small screen.

And today, it’s being painted as especially vile since it seems to be threatening to become a content creator for studios worried about their shows and movies becoming stagnant during the writer’s strike.

Will more and more execs turn to ChatGPT, an app that can do everything from debug computer programs to compose music, write essays and teleplays, and strategically plan business models? The app isn’t exactly getting laudatory press these days; indeed, it seems to be the poster child for A.I. excess. Nonetheless, A.I. is the fastest growing consumer software application in history and has gained over 100 million users in just a year. Such wild success has allowed OpenAI to grow to $29 billion in valuation.

Two of the biggest questions swirling around A.I. software couldn’t be more basic: what could A.I. be doing, and what is it doing already? For years now, A.I. has been present in all kinds of lanes, but are most people even aware of it? Probably not. They likely first heard about A.I. in movies like Ex Machina or A.I.: Artificial Intelligence where the technology was mostly interpreted as lifelike robots.

But the fact is, A.I. has been a growing part of our digital world, gaining more momentum with each passing year, from apps to games to shopping. And lately, we’ve been seeing what A.I. can do artistically as we’ve been asked to partake in it to create various interpretations of our selfies up and down social media.  

We now all know that A.I. can create a portrait of us “painted” in the style of the Renaissance, but should it? For that matter, is it legal or ethical for A.I. to imitate the style of any existing artists, photographers, and musicians? And what about the aesthetics?

Sure, A.I. can duplicate faces, replicate paint strokes, interpret prose, and rejigger musical phrases, but often the output leaves a lot to be desired. A.I. may be innovative from a technological standpoint, but from an aesthetic one, it lacks in genuine artistry. Worse yet, some of its interpretations are downright wonky. I’ve seen numerous interpretations of actors illustrated in recast superhero roles showcased in videos on TikTok and the resemblances to the stars are absolutely dreadful.

A.I.’s artistry may be getting better and better, but it still needs a whole lot of finesse.

When it comes to dreadfully executed A.I., Altman thinks people will eventually be savvy enough to separate the wheat from the chaff, like certain trained eyes can weed out bad Photoshop. But if millions of Americans can’t tell fact from fiction regarding the last presidential election, or the various fantastically made-up news stories that clog social media, why would we expect them to be able to decipher the difference between fake videos from the real McCoy? I doubt few could determine the difference between the words that an ad agency copywriter wrote versus ChatGPT doing the same, working off of previously existing copy.

Lines are blurring, but I don’t see people’s vision sharpening.

And how much do we allow or want A.I. to do anyway, let alone how would we ensure it doesn’t overstep its boundaries? If a company’s CEO wants to use an A.I. program to create presentations or newsletters, is such low-hanging fruit acceptable? What if a local politician runs an ad where his opponent’s voice is recreated to say something he or she never said? That’s a bridge too far, surely, but what are the repercussions of such A.I. usage? How fast can any A.I. overreach be pulled out of communication channels, disallowed, or even punished? Will the monitoring of A.I. require some sort of Internet Interpol?

And even if there are certain legal restrictions and laws put into place, what about the matter of ethical standards? If A.I. can save companies money and deliver content faster, replacing those folks who used to create such goods, why wouldn’t they?

Look at what’s already been acceptable for years in the world of show biz. CGI has replaced physical production design. Deceased actors have been rendered for commercials and movies. Voices can be created to say anything and everything. Does it matter to producers, artists, the public? If they are onboard, who’ll determine what lines are being crossed?

In “Joan is Awful,” the first episode of the newest season of Netflix’s "Black Mirror," a corporate executive named Joan discovers that she signed the rights to her own life away by signing up for a streaming service, and her daily existence is now being played out by a CGI-created world. It stars Salma Hayek in this meta sci-fi fantasy and all the world is watching a la The Truman Show.

Here, both the computer and Hollywood execs are being skewered, but what happens when such likenesses aren’t coerced, but rather, become a deliberate choice, a way to put a star into a production without the actor ever having to show up? Think that’s a dystopic fantasy? We’re almost there as the new Indiana Jones film proves that a mostly A.I.-created version of Harrison Ford can carry the opening 20 minutes.

And what happens when a Hollywood executive gripes, “You know, they don’t make stars like Cary Grant anymore,” and he or she chooses to have A.I. create an entire Cary Grant performance via the computer for a new movie? Why wouldn’t a studio chief make such a bargain if it meant not having to pay George Clooney $20 million?

For that matter, writers should ask themselves if they would follow suit, given the right opportunity or price. Imagine not being limited by time, history, or death—who would you cast in your dream project if there were virtually no limits? If you could pen a romance for A.I.-created versions of past stars like the long-gone Audrey Hepburn and Clark Gable to star in, would you? Could a new star even be created whole cloth by A.I.? That premise informed the Al Pacino sci-fi comedy S1mOne in 2002, regarded at the time as too inconceivable to be anything other than a fantasy.

Today? Not so much.

Perhaps we need to start contemplating more and more of such things that were once inconceivable and figure out if we’re going to take the bait.

Could an A.I.-version of Cary Grant win an Oscar some day? Will the Turing Test become the new version of a test screening?

More than likely, such questions will start to appear in the debate of what constitutes art and creativity in our brave, new world we live in today. The debate won’t focus on what we allow A.I. to do, but rather, how much we allow it to do.

If audiences keep buying, expect the goal line to keep moving.

If people want Cary Grant to return to cinemas, and A.I. can make it happen, then ethics will be less and less of anyone’s concern. As long as people are consuming, A.I. will thrive.

And if that’s the case, technology won’t be regarded as a monster, but rather, a monstrous success.

*Feature image by Jeff York

Jeff York is an optioned screenwriter, film critic, illustrator, and ad man. He’s also a member of the Chicago Indie Critics, SAG-AFTRA, and a cat lover.
More posts by Jeffrey York.
Newer
Older
Share
Twitter icon Twitter Facebook icon Facebook Pinterest icon Pinterest Reddit icon Reddit
Click here for our recommended reading list.

An Invitation

To a global community of creatives.

All Pipeline Artists members are eligible for monthly giveaways, exclusive invites to virtual events, and early access to featured articles.

Pipeline Artists
Thanks for Subscribing