When this screencap of a hypothetical AI-generated sitcom hit social media, critics focused, perhaps not unfairly, on the phantom hand holding the floating guitar. Notice also that it mashes Penny’s kitchen together with Monica Geller’s living room, or that some pieces of furniture are vanishing into other pieces, Beetlejuice-style. Or that both women are clearly Alyson Hannigan circa 2010. Visually, it’s a mess.
The source video was yanked shortly after posting, so heaven only knows whether this “sitcom” is funny. But how many of them are? Based on this orphaned screencap, it looks like an unholy collaboration between Chuck Lorre and Hieronymus Bosch. Worse, it looks like an inbred bastard child of every White sitcom of the last thirty years. Which explains why the poster, a generative AI enthusiast, thinks it’s great.
It also explains why, if programmers could successfully remove the psychedelic artifacts, networks will likely buy it. Since Friends debuted in 1995, most American sitcoms have featured some variation on its beat sheet. How I Met Your Mother, The Big Bang Theory, Everybody Loves Raymond, and Community have been largely reskinned variations on Friends. But even Friends itself simply scrubbed the Blackness from Living Single and resold it.
Generative AI consumes the largest possible sample of existing media, whether images or text, and regurgitates something passably similar. The law of averages foretells that, the larger our core sample, the more the product will drift toward the middle. Thus, an AI enthusiast will regard the blandest, most derivative product as a triumph of the art form. We know it works because it resembles everything we’ve seen before.
Unfortunately, the network commissioners who buy content will probably also love it. I’ve written before, primarily about music, how the corporate conglomerates that control our media have become risk-averse. They literally have analytic software and focus groups, modeled on the latest behavioral economics, to reassure shareholders that all new content sufficiently resembles everything that’s been successful before.
We could argue that this happens because corporate overlords see their product not as art, but as content. It also reflects chokepoint capitalism, since all music, to succeed, must pass Spotify’s filters, and all TV ultimately heads for Hulu, Netflix, or Disney+. But it also reflects changes in consumption. The streaming services that control our viewing and listening require massive audiences to keep subscription prices low, and 🌟art🌟 can’t do that.
Gone are the days of subcultures and audience segmentation. Streaming services generate audiences, not by creating something innovative or uplifting, but by remaining minimally offensive. Admittedly, this isn’t new in mass media, as network TV’s largely interchangeable Westerns of the 1960s segued into spy thrillers in the 1970s. I’ve already commented on how Friends’ cinematic, character-driven style displaced Norman Lear-style three-camera sitcoms anchored to bankable stars.
Generative AI will reduce this trend to silliness. The nameless spec clip memorialized above shows how the system reduces storytelling so thoroughly, it pinches the same actress twice. If the existing model reduces BIPOC actors to contract players holding supporting parts, generative AI will eliminate them completely. One needn’t ponder hard how it will treat disabled actors, or even capable but slightly funny-looking actors. Like me.
Children and the aged are also underrepresented on existing TV, and often reduced to stereotyped, moralistic roles. Aging actors already struggle to get parts, especially if they don’t have established star power—and that goes double for women. If existing sample TV shows and movies have few children or older adults, the middle-of-the-road produce will inevitably have even fewer, a trend likely to only compound over time.
I don’t want to romanticize the past, especially the recent past. Disney currently shepherds talent like Ariana Grande or Olivia Rodrigo from childhood, molding them according to data and metrics, creating a media landscape already struggling with anodyne blandness. Constant remakes of lucrative IPs like Dune, Harry Potter, Gossip Girl, and Stephen King, reveals corporate leadership paralyzed with fear at even the most rudimentary innovation.
AI is thus not a declension, but an extension of already existing trends toward the mean. It accelerates corporate media’s desire to produce the blandest, most studiously inoffensive oatmeal. In the constant struggle between art and commerce, it supports the bean counters. By controlling the kinds of stories and faces available to viewers, it promises to continue starving the public imagination at a pace once unimaginable.
In short, it helps nobody but the shareholders. One wonders what will help them, when we shut off the TV and read a book.
No comments:
Post a Comment