But he’s definitely following the… ahem… career of Tilly
Norwood, the attractive young woman pictured above. She’s featured in a newly released two-minute
film titled “AI Commissioner.” She also, of course, doesn’t exist in any
unmediated, three-dimensional, sense.
“She” (“it”?) is completely AI generated, and the folks at Particle6 seem
pretty damned proud of their creation.
There has been a huge uproar from actors,
SAG/AFTRA,
and other predictable sources, and of course an equally predictable defense by Particle6
founder Eline Van der Velden. In fact, about every media outlet you can
name has covered the story, and several have published opinion pieces. The British newspaper The Guardian has
already published five different articles about Tilly and the attendant
ramifications.
Curmie finds himself agreeing with Stuart Heritage (why isn’t this guy a historian of the Jacobean period?) that “Sure, it should also be pointed out that her existence alone is enough to fill the pit of your stomach with a sense of untameable dread for the entire future of humanity, but that’s Hollywood for you,” but also that “even if it means that the market will soon be flooded by absolute slop, the betting is that she’s here to stay.” Heritage also points out that other new ideas that were expected to revolutionize the industry—3D, for instance—turned out to be duds, so perhaps the degree of weeping and wailing is unwarranted.
The Guardian article that interested Curmie the most,
though, was a collection of comments by readers of the first couple of articles to appear.
Virtually all the commenters have a point: no, the threat, such as it
is, isn’t immediate; yes, one can presume that “ostlers and bridlemakers were
furious with Gottlieb Daimler and Henry Ford.”
But Curmie wants to highlight the commentary of AshMordant:
It’s too late to be scared.
Hollywood is not about making art, it’s about making money.
Give us one good reason why studios should pay for cameramen, makeup artists, set designers, lighting, catering and of course actors when AI can do the job and make money.
Films made with real people – actors as well as all the other innumerable people listed in the end credits – will soon be something like ballet or opera: enjoyed by a few cinéastes who are willing to pay all the money for this art form.
But why would a fan of the, say, Fast and Furious franchise or the Marvel universe or whatever it is called do that? All they care about are visual and aural stimuli, and AI can deliver that perfectly.
Well, as Curmie used to say a lot in his Asian Theatre
class, yes and no. Certainly there is
little interest in Hollywood in making art; mediocrity sells, after all. Whether films featuring human actors will go
the way of ballet and opera is speculative.
But the idea that movies that rely on “visual and aural stimuli” would
not be seriously affected by replacing a couple of humans with AI-generated
versions is, well, simply a statement of fact.
Regular readers here know that Curmie and Beloved Spouse are
fans of TV whodunnits. Not infrequently,
the stars of such shows clearly got their jobs more for being conventionally
attractive than for their acting ability.
Almost always, they’re fine most of the time: no one will confuse them with
Ian McKellen or Emma Thompson, but they can handle the vast majority of what
they’re required to do. Inevitably,
however, they’re called upon to show a spontaneous reaction or display a real
emotion… and Curmie finds himself shouting at the TV, “don’t make [insert
gender-appropriate pronoun here] act!”
AI is all about processing the past, which makes it a paean to
conventionality and mediocrity. But if what
it replaces is already, well, conventional and mediocre, then the loss is
negligible.
Well, sort of. A
number of ultimately excellent actors weren’t necessarily brilliant from the
get-go. Whether they were learning their
trade in full view of millions of people, or whether they were relegated to eye-candy
roles that didn’t allow for actual acting doesn’t matter. Talent can’t be taught; skill can. And AI poses a threat in that it stands to
short-circuit the career of those with the former but not yet the latter.
Moreover, the actors who will be affected by this aren’t the ones making eight figures for every film or seven figures for every TV episode. Rather, it will be the ones whose names are unfamiliar to everyone but their family and friends. As of two years ago, only about one in seven members of SAG/AFTRA earned even the $26,470 in film and TV work required to receive insurance through the union. It may be easy to scoff at the not-really-so-hard lives of the millionaires and billionaires, but the overwhelming majority of actors in the trenches are struggling to make ends meet while living in the extremely expensive area around Los Angeles. These, not the stars, are the folks whose jobs Tilly and her manipulators want to put out of work.
“AI Commissioner” is remarkably unfunny for what is
advertised as a comedy. Tilly is not
required to do much, and what she does do requires neither significant talent
nor skill. But, just as other variations on the theme of AI have improved significantly over even the last few months, we can reasonably
expect that future Tilly clones will be qualitatively better than what we’ve
seen so far. And we can reasonably
suspect that some future project might actually hire some decent writers.
It strikes Curmie that there are three areas of contestation
here: the technological, the ethical, and the pragmatic. Let’s tease those out a little.
We start with the technological. Clearly, Tilly isn’t ready to play a major
role yet, and probably won’t be by the time the current SAG/AFTRA contract
expires roughly nine months from now.
But progress will be made, and everyone concerned had better be ready
for the negotiations. Actors, the real
ones, still have enough clout that the threat of another strike will mean
something. But the studios and producers
aren’t without power, either.
One of the principal disputes that led to the SAG/AFTRA
strike two years ago was what Curmie called an “obscene” proposal by producers that background actors be paid a
single day’s wage for the rights to use their image in perpetuity without
consent or remuneration. Obviously, variations
on the theme of CGI were intended to replace background actors, whose images
could be manipulated to fill in for a presumably slightly different group of
actual humans.
We know, of course, that CGI has been employed in manifold
ways both before and after the latest SAG/AFTRA contract was finalized. But, and here’s the transition to ethical
consideration, CGI is different to the extent that a CGI-generated is generally
employed to make a character do something a human actor can’t do (think
superhero movies, for example) or to create a character that isn’t, in fact,
human (e.g., Gollum, or Thing in the Addams Family franchise). Curmie isn’t sure how the negotiations ended
up on replacing background actors, so there may be an exception there, but some
of the commentary he’s reading suggests that there is at least some protection
for human background actors.
The unions’ claim that AI is “trained on the work of
countless professional performers — without permission or compensation” is both
true and ethically ambivalent. Where is
the line between studying and de facto plagiarism? When Curmie was 17, he did a show with an
experienced comic actor who consistently got laughs by separating line and
gesture. Curmie adopted a similar
strategy in future shows, and indeed taught it to younger actors when he began
his teaching and directing career. Was
that stealing? Or just learning?
Still, there is a line there that should not be crossed,
especially when we’re talking about an AI “actor” doing literally the same
thing as the source rather than simply employing a general concept. In scholarly writing, even a two- or
three-word description (something like “farcical tragedy”) sometimes shouldn’t
be repeated without attribution, but at other times a much longer passage, even
word-for-word (significant person X was born on this date in this place to this
couple, whose occupations were… etc.), won’t cause much concern. (Curmie, being Curmie, still wanted students
to cite sources, even for that generic stuff.)
One supposes that some rules could be put in place to allow some uses of
AI but nor others, but it’s difficult to imagine how.
So now we turn to the pragmatic. This is a genie that isn’t going back into
the bottle, at least until everyone is satisfied that there neither are nor
will be further advances in the technology (unlikely), or studios start
thinking of rewarding actual artists instead of corporate suits (even more
unlikely). Hollywood is the very crystallization
of late-stage capitalism, uninterested in anything but making more money for
those in power.
Tilly Norwood is being touted as the next Scarlett Johansson. That’s not gonna happen, at least as long as the
real one draws movie-goers in large numbers.
Tilly won’t do that, except perhaps very briefly as a variation on the
theme of a freak show. But she is a
helluva lot cheaper, even after you figure in the salaries of the geeks who
manipulate her, and we can expect studio heads to start circling around Tilly
and her successors like sharks around a wounded seal if given half a chance.
Curmie isn’t seeing much of an upside here. Perhaps, though, the citizenry is more
interested in good work than in profits for overpaid capitalist hacks. Or they’ll just get bored with the gimmick. ‘Tis a consummation devoutly to be
wished. On the other hand, theatre,
fiction, and film have all warned us of the dangers of allowing any form of
artificial intelligence too much scope: think R.U.R. (which is over a
century old!), 2001: A Space Odyssey, or the first Star Trek movie, for example. That doesn’t mean the prospects are
necessarily dire; they are just extremely unlikely to be positive unless,
Gentle Reader, you own stock in Netflix or Amazon or Disney or whoever.
We shall see.
