Having made his living as a theatre scholar/teacher/practitioner for a few decades, Curmie is well aware that what appears to be real on stage, isn’t. That isn’t really a medieval Danish prince up there; it’s an actor pretending to be one. But there is informed consent all around: the actor voluntarily assumes the role, and the spectator agrees to pretend to believe in the impersonation. No one in the audience believes that’s actually Hamlet.
Curmie also knows that headshots for actors should show
actors in the millisecond in which they look their absolute best. There’s a caveat, however: they have to look
like their subject. If you’re
auditioning for strangers, you can be brilliant, but if the casting director
doesn’t recognize you from your headshot, your chances of getting that callback
go down. And if you’re called back
because of what your headshot looks like, you’re no longer in the running if
you don’t look like that photo. Inauthenticity
is not merely unethical; it’s counter-productive.
Why these introductions? Well, Stanford University has admitted to using AI to doctor a photograph of students celebrating the Lunar New Year in 2024. The trio are holding plates of noodles outside an eatery called Star Ginger, apparently a franchisee in one of the campus dining halls. That’s the image you see on the left in the photo above. It was used in a marketing campaign almost immediately. One presumes the university photographer got the appropriate permissions from the pictured students. So far, so good, right?
Sure. But it’s a more
recent version of that shot, the one on the right, that’s been getting a lot of
attention of late. As you can see,
that’s both the same photo and not at all the same photo. The Hispanic man on the right wearing a GAP
sweatshirt has been magically transformed into a black woman in a Stanford
shirt. The face of the Asian woman on
our left has been conspicuously narrowed.
They played around with the face of the other man, the one in the middle,
too, although not as obviously. Curmie
assumes that whatever that is that blocks the area above the guy’s head and
below his chest is a function of the photo’s use on a “welcome” banner
sponsored (if that’s the right word) by Stanford Residential
& Dining Enterprises.
The guy on the right of the original photo, the one who was
literally disappeared, is Billy Ramirez.
He was a freshman when that picture was taken; he’s a senior now. He says he was driving to campus for the fall
semester this year when a friend sent him the two images together. He told The Stanford Review that he was initially “baffled,” and thought it was “hilarious.” Ultimately, though, he was upset: “I don’t
agree with Stanford making those choices about how we were represented. Seeing
my identity changed and being left out of the picture made me feel, in a way,
silenced and erased from a representation that was supposed to include me.”
That latter reaction may be a bit much, but in general terms
it’s understandable. And it’s easy to
see why, after all the local publicity, Ramirez doesn’t want his “legacy at
Stanford to be ‘the AI guy.’” It is
certainly more than a little ironic that the change was made for a welcome
banner. The apparent subtext: “Everyone
is welcome here at Stanford. Well,
except for people who look like that guy. At least the other two are OK if we touch
them up a little.”
Anyway, the image seems to have been posted on Fizz, a social
media platform for Stanford students only, on the 20th of September. That Review story was published on the
21st. It was in the New York Times the next day. One day later, the
self-described “Other Two” released a “Letter to the Stanford Community” on Substack. Finally, on the 24th, the Review
ran an article about the university’s admission that yes, that photo had been
altered by AI.
Curmie finds it interesting that the statement from Stanford
officials seems rather unconcerned with the misrepresentation per se,
but they’re pretty steamed up about the use of AI to do it. They’ve got their skivvies in a twist about a
violation of university rules. They do
concede that “Not only the fact that the image was altered, but the ways in which it
was altered, reflect a serious error in judgment. We have apologized to the
students in the photo whose images were altered or erased…. Quite simply, this
was wrong and should never have happened.”
There are several ways in which the doctored image goes
against university policy as defined in a document titled “AI guidelines for marketing and communications.” []. Curmie quotes a couple sections: “Use AI in
ways that are transparent, accurate, and do not mislead.” Obtain written
permission before using an individual’s name, image, voice, or likeness
(including realistic or recognizable AI-generated depictions or imitations) in
published content.” “Do not create or publish AI-generated content that falsely
depicts Stanford people, events, research, facilities, or achievements.” You get the gist, Gentle Reader.
Of course, as that New York Times article points out,
universities have been doing making similar misrepresentations for years, using
Photoshop or a variation on the theme. Usually,
although not always, it’s to pretend the school has more minority students than
it actually does. That makes it even
more ironic that it’s a Hispanic man—a representative of a demographic that is,
according to students, under-represented in the Stanford undergrad population—whose
image was removed altogether.
It’s certainly understandable that a university would want
to exercise some control over content used in their own forums. You don’t want someone showing up in a photo
wearing a shirt from a competitor, for example.
Thus, when Curmie took students to Ireland, one of his rules was that no
clothing could represent an American university other than our own, nor could
it overtly advertise a product. (This latter
part might justify changing Ramirez’s GAP sweatshirt to something less commercial,
but of course no more than that.)
That said, were Curmie of a cynical disposition (perish the
thought!), he might suggest that the university’s emphasis on the AI element
might be an intentional distraction from the larger issue of intentional
misrepresentation and the motives for the… erm… over-enthusiastic editing. All the students directly involved—the three
people in the original photograph and the editor of the Review—are
certain that the impulse to change the image was to present a false image of
student diversity. Curmie suspects the
black woman would agree if she… you know… actually existed.
This doesn’t mean the students are correct in their
assumptions, but they are at the very least plausible. And here we see a variation on the Streisand
Effect. Seriously, Stanford: you
couldn’t even stage a photo with an actual black female student? At best, that’s transcendently lazy. At worst, it calls attention to what many
people regard as the situation your clumsy attempt at manipulation sought to
cover up. Even if that wasn’t the
intention, it sure looks that way.
Yes, it could be relevant that threats from the Trump
administration about anything that might possibly be regarded as a DEI
initiative put Stanford in a difficult situation. But “dishonest,” “sloppy,” and “stupid” are
adjectives that no university, let alone one of Stanford’s reputation, ought to
want attached to their name, and you can’t blame Trump for the fact that those
descriptions sure seem accurate.
More to the point, the most relevant section of those AI
guidelines is this: “You must take personal responsibility for any work you
produce using AI. This means reviewing outputs for accuracy, ensuring alignment
with institutional values, and confirming compliance with these guidelines and
university policy. This responsibility cannot be delegated.”
Curmie understands that Stanford promises to prevent
anything like this ever happening again.
He believes that they will at least make every effort to do so… or,
rather, he will when they announce the names (plural) of the people they fired
over this incident.
No comments:
Post a Comment