The Dishonesty of AI
Having just returned from arguably one of the biggest tech and creative conferences in the U.S., South by Southwest in Austin, I can confidently say the prevailing IRL sentiment regarding the creative dishonesty of AI matches much of the uproar online. Not always the case. So let’s look at this branding of AI as blasphemous with some perspective.
When Albert Cheng, Head of AI Studios, Amazon MGM Studios, was asked by an Axios reporter during his presentation if Amazon Studios would start flagging content as AI, he said no. This was100% the right answer. And not just for narrative features but many documentaries, too. Before you start lighting the torches and sharpening the pitch forks, hold on.
What you see on a screen, is always shaped by those who make it. Whether it’s narrative-based or a documentary, there is always a point of view that shapes the intent and the “truth” of the images. Am I suggesting journalists should start fabricating footage to produce disinformation? No, not at all. AND, footage is already being used in ways that is representational. Think stock footage, re-enactments, or even how footage is cut together now.
Using AI in any type of film is no different than traditional post: its use is 100% in the hands of filmmakers and organizations who must decide what their ethical boundaries are, based on the audiences expectations. Like always, filmmakers need to consider these frameworks: the appetite for deception during a presidential speech is miles away from a show at The Magic Castle. This is not unique to AI.
How “truthful” documentaries are, for example, has been debated since the beginning of the form. Google 1942’s Nanook of the North by Robert J. Flaherty for starters. How a film is made matters less than the why. AI is merely a tool to enable more stories to be told, in a more dramatic fashion, if so desired.
When Errol Morris used recreations in The Thin Blue Line, many thought it was profane. So much so it was snubbed by the Oscars & many others despite the rave reviews even then. Reenactments are so commonplace now we don’t even give this technique a second thought.
AI, on its own, doesn’t suddenly twist the moral center of the work because the pixels you see were output by the latest video generator. It just means the director, who may be scraping every penny they can find, out of the couch cushions of anyone who will let them, doesn’t have to spend that money shooting B-roll or licensing expensive footage. The final output is still dependent on human storytellers to script, direct, cut, analyze and deliver a compelling and truthful narrative—based on the cinematic form.
Tagging AI content in narrative films is even more counterproductive. AI is just like CGI, miniatures, matte paintings, animation and all the other tricks of cinema. No one was expecting a disclaimer when James Gunn digitally recreated David Corenswet’s hair in Superman to make the flying scenes more “believable.” Nor did audiences take up pitchforks because Gary Sinise hadn’t really lost his lower legs in Forest Gump. I’ve also never seen a Pixar film with a “truth in cinema” footnote.
AI is just a faster, less expensive and less cumbersome “how” than traditional production or post processes. The upside is AI can provide a huge incentive for ad agencies, clients, networks and streamers to take chances on “riskier” content. That is, IF the script and the director have the potential to make a compelling show. We’ll discuss what that could look like next.