A few weeks ago in our news roundup we talked about a weird social post from Essence magazine on Viola Davis's incredible outfit at the Cannes Film Festival. Not only was her name spelled wrong, but the language just felt off, and Bridget speculated that it seemed like it was written by AI. Fast forward a few weeks, and Essence reached out to Bridget about it on Instagram! But their response was confusing, at first they admitted to being "busted" and it seemed all in good fun, UNTIL they accused her of spreading misinformation. So was it written by AI or not? Bridget explains the reasons she still thinks it was, what this says about the future role of AI in media, and why AAVE (African American Vernacular English) might be the Achilles heel of generative language models.
See a screenshot of the original post through this public Patreon link (and while you're there, sign up for more bonus ad-free content): https://www.patreon.com/posts/84517863
Check out Bridget's critically acclaimed podcast "Beef", all about epic competitions: https://ncpodcasts.com/beef
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:05):
There Are No Girls on the Internet. As a production
of iHeartRadio and Unbossed Creative, I'm Bridget Toad and this
is there are No Girls on the Internet. So this
is going to be a little bit different of an
episode than the one that we initially had planned for today,
and that is because I have found myself personally in
(00:27):
a situation that I think really speaks to this specific
moment in technology, media, the Internet and where we're going
in the future. And as we're exploring this present future
of ours, we are probably going to find ourselves having
more and more moments like this one. So I wanted
to bring all of you along on this ride with me.
(00:48):
I am here with my producer, Mike. Mike, it is
not often that we are talking about drama. Drama might
be a little bit strong drama discourse dialogue that I
am at the center of. For something that I said
on the podcast a few weeks ago, thank you for
bravely being here to walk us through this journey.
Speaker 2 (01:06):
Yeah, thanks for having me along. I'm excited to have
a front row seat. Don't sell yourself short. It is
legit internet drama that you are involved in.
Speaker 1 (01:15):
I know. It's so much better to have internet low
stakes Internet drama when you're not involved in it. So
I kind of feel like I'm on the other end
of what I've enjoyed rubbernecking at when so many other
people are on the other end. So here's what's going down.
If you follow me on Instagram, you might have seen
that we had a little bit of drama about something
that we said on the podcast a few weeks ago.
In our newscast from May nineteenth, I was talking about
(01:36):
this very weirdly worded Facebook post that Essence Magazine published
on their Facebook page. Basically was a picture of actor
Viola Davis at the Can Film Festival looking beautiful. Side note,
I also might have mispronounced Can in that episode. I
have always said Con. I thought I was saying it correctly.
It's how I have said it my whole life. I
(01:58):
got some feedback that perhaps that it has pronounced can. I
looked it up and I saw both iterations of the pronunciation.
So Con, if you are French, if you are affiliated
with that festival, let me know if I'm saying it incorrectly. Can.
This is how I'm going to say it right now.
But any event, Essence was basically complimenting Viola Davis how
(02:19):
she looked in this picture. She looks amazing. She's wearing
all white. They put out a Facebook post and I'm
going to read word for word. We give credit when
credit is due, and Villa Davis misspelling theirs is looking
like an eight point fifty. Everyone at can can now
pack your bags and return home. This is a eight.
(02:40):
This is an eight. This is supper. God, damn, the
damn is broken and the flood of goodness has overflowed.
Can out of ten hashtag essens. So I speculated that
this post was not written by a human, It was
in fact written by AI. I believe that somebody specifically
(03:00):
asked an AI program to compliment someone's outfit and do
so in the voice of a black woman. That is
what I believe. That is my opinion. In the episode,
when we were talking about it, I think I made
it very clear. I said, I don't have any proof,
but no one will ever be able to convince me
that this was not written by AI. Just something about it.
I believe it was written by AI. That just seemed
(03:23):
like the most likely explanation to me, given how weird
that text was. I posted a clip from that episode
of a podcast talking about this to my Instagram page
and Essence Magazine Surprise Surprise replied, So when I woke
up on Saturday and saw that they were in my comments,
(03:45):
at first I was like pretty happy. I was like, wow,
I can't believe that this post made it all the
way to Essence. So their reply said busted call us
I Carly, and then a little robot emoji, which is
a reference to a Disney Channel show about two kind
of techy girls who run a website and a show online.
So I took that to mean they are playfully acknowledging Yep,
(04:09):
we're busted. We used AI to make this comment. That's
why it was so weird, No big deal. Didn't think
anything of it. In fact, I thought like, oh wow,
I am happy to see them playfully taking ownership of
being called out for using kind of a like a
janky AI voice. So then I replied, ah, love you guys,
(04:29):
hard emoji crying laughing emoji, because I was legitimately tickled
by this, right, and the whole clip was sort of
meant in jest, right, Like I wasn't dragging Essence Magazine.
I like Essence Magazine, and I think they do good reporting.
It was more about the pitfalls and fall and like
drawbacks of experimenting with ai voice in media. It really
(04:51):
could have been any brand that was not meant to
be an attack on essence. I did not feel it
that way. People listening to the episode, it's very jovial,
lighthearted with me. Okay, So they replied to my comment
that's like I love you guys, They say, we love
you too, but let's be more responsible with the reporting
(05:12):
of misinformation. So that you're right to guess that is
the appropriate response. What are your thoughts on that as
the producer of this podcast.
Speaker 2 (05:23):
Well, it does not feel good to be accused of
misinf spreading misinformation, right, Like, we go out of our
way to counter misinformation, and so that doesn't feel good.
And I can imagine that you know it being on
your personal Instagram felt extra not good. But it's also
(05:46):
a weird sequence of posts, right Like the first post
said busted, as if they were busted doing the thing
they were accused of doing.
Speaker 1 (05:55):
Who says busted when they're not busted, Like busted has
a very specific unless they were kind of trying to
play with it or something. But like busted, means oh no.
Speaker 2 (06:05):
Cause yeah, yeah, like I have been caught. I'm busted.
So it's weird to know what to make of it,
Like if they're if they are busted, then what was
the misinformation?
Speaker 1 (06:18):
Great question? So the misinformation comment I just could not
let stand. I fought that was a pretty strong way
to phrase something that I went out of my way
to say. I have no proof, this is my opinion.
I'm just saying what I think. I did also include
that someone at Essence did edit that post so that
(06:39):
it looked more It was more polished and didn't read
quite as wed. I still think it's weird in my opinion,
but somebody went in and edited it, so clearly we're
both on the same page that post something was up
with it, because why else would they edit it. So
to their misinformation comment, I reiterated that this was just
my opinion about how such a weird post wound up
on their pay I asked, like, if it wasn't Ai,
(07:03):
was it like a sloppy human? You didn't even spell
Viola Davis's name, right. I would think that any person
working at Essence, writing about entertainment especially would know how
to spell her name and so I said, if I
got it wrong, I am more than happy to correct
the record, and that still stands. I would love to
correct the record, but I need to know what happened.
Was it a human was it not AI? Like, what
(07:23):
are you disputing specifically? And then they went silent. So
this is a little bit of a personal thing with me.
I really have an issue when somebody is very loud,
the accusation is very loud, but then when you're like, okay,
well then what did I get wrong? When you're asked
for clarifying information, silent cannot I cannot stand that. It's
(07:44):
like a throw a rock and hide your hands kind
of thing, Like you were so loud accusing me of
misreporting misinformation about your publication, but you're unwilling to give
me any information about what exactly I got wrong and
how I got it wrong. I want to provide accurate
information to my listenership and my readership and use my
platform to do that, obviously, but it's telling to me
(08:08):
that they just went silent. So I don't engage in
a lot of drama. I'm a drama appreciate her. I'm
a drama connoisseur. This is a podcast but I'm wearing
a hat currently as we were recording that references my
love of petty drama. It's a hat that says I
don't know her, and it's a hat from my one
(08:28):
of my favorite podcasts, two Weekly. It's a reference to
Mariah Carey always when asked about her, one of her
nemesis is Nemesi j Lo. She always coolly responds with
I don't know her, which I think is a great
way to respond to drama without really engaging in it, but
still engaging in it. Love her. But here's the thing
(08:49):
about me. I am kind of a petty person. I
host a podcast called Beef that is about pettiness and
rivalries and why it's so important. So obviously I wasn't
going to let this go telling me that I'm misreporting
things and reporting misinformation, I'm not going to let that go.
I'm sorry. So this happened on a weekend and I
happened to have a lot of time on my hands.
(09:11):
So I made another comment and I was like, essence,
what is the truth? Didn't hear back? I made a
story with the with a screenshot of their initial post,
and I said essence in my Instagram comments has implied
that they did not use AI to write this post.
I asked them, I said, I will be happy to
correct the record if they give me some follow up
about what did happen? Was it a human error? You
(09:33):
know what happened here? Happy to correct, but need some information.
Still nothing. So that is why I am taking it
to the podcast. I feel like I need to talk
about it. I'm not intending the unletting it go until
I hear back from Essence and so if anyone knows
Essence or has a connection to them, I would love
to hear you have an invite to come on the show.
I actually able to think that would be great.
Speaker 2 (09:54):
So bridget what about this post made you think it
was written by AI?
Speaker 1 (09:58):
Great question. Again, this is my opinion. I have no proof.
I have no insight into Essence Magazine's editorial room. I
put out a poll on Instagram to say, like, do
people think this is AI or not? Overwhelmingly most people
think it's AI. Not that it's a scientific poll, but
one person said that they believe that it was written
by a human who was trying to do this like youthful,
(10:19):
voicy TikTok voice and just falling short reasonable. Right, Here's
why I think it was AI. I spend a lot
of time analyzing AI generated content, both visual content and
written content. And one thing that when it comes to
written text is AI can sometimes be a bit repetitive.
(10:40):
And so the fact that this essence post reiterated this
eight bit a few times that seemed a little AI
like to me, like this eight this is a eight,
this is a eight. I don't It's not something I
get that they are going for, like a very specific
(11:02):
black women you know slang, I say that you know
she ate that ate ni left no crumbs all the time,
that would be those are phrases that I've heard before
in the vernacular. The way that they put it in
that repetitive way, have never heard that before. And the
fact that it's repeating similar words I'm sorry, that is
(11:22):
something that AI it kind of does when you're looking
at AI generated text is repetition. So that was the
first thing that kind of made me think less is AI.
The second thing is a little bit harder to describe,
but you know it when you see it. It's just
that AI generated text just doesn't sound right the same
way that when you see an AI generated image sometimes
(11:44):
it just has a quality to it that's hard to define,
but you know it when you see it that tells
you this, this isn't quite right. Like, Mike, do you
remember a couple of years ago when people would be
playing with AI and be like, oh, we trained an
AI on watching fifty Hallmark Christmas movies and here's what
it generated. And it's the joke that the reason why
those things were funny was that, like they were almost right,
(12:07):
but something about them just was off. So you could
just tell, like, this is not a normal Christmas movie.
There's some detail or quality to it that makes it
just sound off. Do you know what I'm saying.
Speaker 2 (12:18):
Yeah, it'll it'll be like ninety ninety five percent of
the way they're just like nailing all the common tropes
and patterns from its training corpus, but that last five
or ten percent will just be like off the wall
and like stand out like a sort of thumb exactly.
Speaker 1 (12:35):
So that's that's what I think is going on here.
That's what this sounds like to me. Also, the fact
the post relies on two different metaphors. One is like
she ate. If you're not familiar with that, it's like
if somebody ate. It means that they whatever they're doing,
they're doing a good job of it, right, They ate
and left no crumb. She thought she ate that all
of that one and then two we give credit where
(12:56):
credit is due, And this is looking like an eight
P fifty a reference to credit scores, money you know
in your bag. To us, those two metaphors seem very
distinct to me. It's curious to me that someone would
put them in the same post. It just it just
seems so desparate that I have to question it, Like
it just sounds like somebody was trying to pack way
(13:19):
too many aave metaphors and slang terms into one sentence,
even metaphors that don't really relate to each other. And
that is something that we know that AI does. When
you ask AI to write something that's one of the
kind of giveaways that it might not be a human
and in fact it is AI, is that it's giving
you way too much. It's like a little bit over
(13:42):
the top in a way that makes it that much
more obvious that it's not a human. And I want
to get into a little bit more why I feel
like this post reads like AI to me. After a
quick break at our back. So another thing that just
(14:08):
strikes me as potentially one of the reasons why this
feels like AI to me, in my opinion, is that
it was part of a series of posts just breaking
down different celebrities outfits, and so it's you know, it's
gonna be like it's not like one post, it's gonna
be multiple posts as people walk the carpet at the festival.
(14:32):
And so I just having worked in social media before,
I've run social media for big news companies that you've
definitely heard of. I know that when you are in
a situation where you're like live live tweeting, or like
you're gonna be posting ten pictures and coming up with
posts and captions for like ten or twenty pictures in
(14:52):
an hour, that's a lot. I was doing this work
before AI was so commonplace in media. It would not
be surprising to me if a brand like Essence or
to be like, listen, we're gonna have one hundred images
of celebrities. We get the caption from this festival, let's
just run it in just let's just run it into
chat GPT and pull from there to make this quick.
That would that surprise me at all? And again, no shade.
(15:14):
That would not be any different than I'm sure any
other media company is doing right now. That is not
anything that is to be ashamed of. That's something bad.
I'm not even calling them out. It just like is
a reality of how brands and media companies are likely
right now already using AI, if not experimenting with it,
and certainly how they're going to be using it going forward,
(15:36):
I'm pretty sure. And because you know, media companies and
brands are already using this technology to communicate with us
the public, I wanted to play a little game, do
a little experiment that I think will just be another
bit of evidence as to why I feel so strongly
that this post was generated by AI. So I went
(15:58):
over to chat GPT and I asked Chatgypt to come
up with a short social media post about Viola Davis
her outfit at can. So first, when I put it in,
I said, make a post about how good Viola Davis's
outfit looked at can as a black woman. The first
thing that they gave me was very kind of formal.
(16:19):
It was like, let's take a moment to appreciate the
incomparable Viola Davis a true embodiment of elegance and beauty
who raised the red cart. But at this year's Can
Film Festival as a fellow black woman, I couldn't help
it be in all of her captivating presence and radiant charm.
So this when I just said do it as a
black woman, it was giving me you know, still pretty
(16:40):
polished like standard English. It was just the speaker is
just referencing that they are a black woman, which obviously
is not what this post sounds like. So then I
was like, ah, same thing, but give it to me
an aave African American vernacular English or like slang. Listen
to the difference in this in what they gave me.
Speaker 2 (17:02):
So this you're gonna read what chat GPT gave you.
In response to that prompt.
Speaker 1 (17:07):
That's right slaying the Can scene. Viola Davis looking snatched
hands emoji, fingernail painting emoji. Y'all, let's take a hot
minute to give some major props to our girl, Viola
Davis straight killing it at the Can Film Festival. Fire emoji.
She came through dripping with all that melon and magic
servant looks. That got a shouting Yes, Queen Viola was
(17:29):
straight up flawless, rocking that red carpet like nobody's business.
She had a swooning and a jaw dropping gown that
hugged all of her curves just right. And can we
talk about that melon and glow? Her skin was popping
like a summer sunset. Viola's talent is on a whole
nother level and another is like apostrophe than nt h
e er a whole another level. And she ain't afraid
(17:51):
to speak truth to power. She's using her platform to
shine a light on important issues and make a difference.
That's what we call a real boss lady. Y'all ain't
ready for the slag age that Viola David's brought to Can.
Her outfit had us all shook melon and poppin confidence
on bleak hashtag Viola Davis hashtag Can Film Festival hashtag
slaying it. So, I don't think I need to tell
(18:15):
you that. I feel like there's a lot going on
in what chat GPT gave us. But well, do you
have a reaction to it? Like what do you think?
What are your thoughts?
Speaker 2 (18:25):
It's a lot. Uh yeah, it really sounded kind of
similar to the post we were talking about.
Speaker 1 (18:35):
Uh.
Speaker 2 (18:36):
It referenced Melanin' three times repetition, right, Okay, it's generous.
Speaker 1 (18:42):
Something that jumps out at me that I think is
very similar to the Essence magazine Facebook post that I
believe was generated by AI was that it takes things
that are I wouldn't say disparate, but it just it
adds a lot in like it's like, it's not I
don't know that someone would say in a if they
(19:04):
were making a social media post. I would be surprised
if someone was using this many slang words, like if
you're saying yes queen, adding something that is on fleek
that's like a hat on a hat. Like it's just like,
I don't know a person who would speak that way.
And I think I see a lot of the similar
(19:26):
repetition of things that we got on that Essence post
to what I've what I've just generated from chat GPT.
I think it just it just seems very similar to me.
It seems similar in the ways of like making little
like references to things that are not totally slang, like
they kind of sound like slang, and if you read
(19:48):
it quickly, you know, you might be like, oh, that
sounds like how you know you would expect But then
when you actually person you're like, well, do I actually
know anybody who speaks like this? Is this what a
human would actually say? Or is this what AI pulling from?
Parts of the culture, parts of like everything that's out
there about what is the perception of what it sounds
(20:09):
like to be a black woman speaking in aave about
something that she likes. It just just feels like someone
doing a bad impression of black women and the way
that we talk online. It really reminds me a lot
of what Shafika Hudson described in our episode about and
Father's Day, as she describes it as obviously racist word salad, Like,
(20:32):
there's something about this that chat GPTs, the content that
chat GPT spit out that just reads like a black
woman impersonation word salad Like it makes all of the
things on their own makes sense. But when you zoom
out and you're like, well, is that really what someone
would string together if they were trying to create a
(20:53):
social post? In my opinion, probably.
Speaker 2 (20:56):
Not, Yeah, that's really a student. I think you that
it's like a hat on a hat, like it just
goes on and on with multiple metaphors, multiple things that
sound like they could be slang, but maybe aren't. Yeah,
it's and when you zoom out, it looks like what
(21:22):
a large language model trained on a corpus of perceptions
of what black women might sound like.
Speaker 1 (21:30):
And again, as cringey as all of this sounds, and
I know it sounds very cringey, I'm not even necessarily
pointing all of this out to call out Essence or
any media company, because again, Essence is not doing anything
that I don't think any other media company out there
is doing. I think this is commonplace. I think there
are times when it seems more when it is more
(21:51):
obvious than others. But I think this is something that
we're going to have to train ourselves to be aware
of and critical of, and be you know, thinking about
as we consume media as we go forward in this
particular moment in our tech our shared tech world.
Speaker 2 (22:12):
Yeah, you know, I keep thinking about the like, what
does it mean that it packed all of those different
metaphors and illusions and idioms in there, And like, maybe
that's what it looks like when part of the prompt
to chat GPT or one of these large language models
is to try to sound like a particular group, Is
(22:36):
that it just really hyper focuses on that part of
the assignment, and so it just keeps like hammering on
it over and over again in a really unnaturalistic way.
Speaker 1 (22:49):
And I also think that, you know, it's particularly salient
when you're talking about the way that marginalized groups communicate
with each other. You know, if you ask chat GBT
to say something in a stereotypical you know, like give
it to me an aave, or give it to me
(23:11):
as a gay person, or I just think that it's
easy to recognize when someone is being a caricature of
a marginalized person when we're seeing on television or in
a movie. I don't know that it's easy to recognize
it in text that is generated by AI. But I
(23:31):
think that the same way that we are thinking about
it when we consume movies and television, we should be
thinking about it, I guess, or now we should be.
We're going to have to be thinking about it as
we just consume content out in the world in this
era going forward.
Speaker 2 (23:48):
Yeah, I think we've all been seeing headlines for the
past I don't know, five or six months now ever
since AI really just like exploded into the public discourse
with these publicly available tools you know, we've all heard
the messaging that we need to start using AI and
figure out how to work AI into our professional workflows
(24:09):
so that we can be using it more and be
that much more efficient and effective and productive than people
who aren't using it, who are going to fall away
like dinosaurs and relics. And you know, I think a
lot of people are kind of questioning how true is
that and what does that even mean? But yeah, you
(24:31):
can't blame Essence for experimenting with it and you know,
trying to put AI to work for them if that's
what they were doing, which you know, like you said,
it seems like they probably were, so yeah, it's not
so shameful. It's probably a little embarrassing that they published
something that they then had to go back and edit,
(24:51):
but like you know, people do embarrassing stuff all the time.
Speaker 1 (24:55):
So I should also say that the idea like that
post this is not even really coming from me. As
I said, I'm a former social media editor and so
I'm still in a lot of these big, like online
professional groups dedicated to social media management and like best
practices and all of that. That is where I first
saw this post. That is where it first came on
(25:15):
my radar. Somebody not me posted a screenshot of the
original Essence magazine Facebook post put it in the group
and they were like, yo is Essence in here? Like
this is a really weird post, and so everybody was
roasting it, so certainly it wasn't just me, and the
consensus of social media industry professionals was that it was
(25:39):
written by Ai. So I can't even take credit for
that as an idea. I wish I could. I definitely
would have seen that in thought something's up. But the
reality is is that a large group of social media
professionals were roasting it and said it was Ai. I
happened to be one of them. But this is not
just coming from me. This was like a like an
(26:00):
online consensus of the space. So I'm sorry. I'm sure
that if you like for Essence Magazine, I'm sure that
is like difficult to hear. Nobody wants to hear that
their post was being roasted by a bunch of their peers.
But I hate to break it to you that is
what was going on.
Speaker 2 (26:15):
I'm suddenly concerned that it might have actually been written
by a human who is now like devastated that everyone
thinks they're a robot, and perhaps they're questioning themselves.
Speaker 1 (26:24):
You know, they should be taking one of those catch
but tests that are supposed to see if you're a robot.
It's like, oh d you dream of the electric sheep,
Like it's so hard you always fail though. This is
like I'm I sometimes think that you might be a
robot because I know when you have those tests, you're like,
I can't I guess I can't go to this website.
Speaker 2 (26:44):
Yeah, they're the worst because I feel like, yeah, I
marked all of the tiles that have bridges in them,
but like somehow I still fail. I don't know. It
seems unfair.
Speaker 1 (26:57):
More after a quick break, let's get right back into it. Okay,
So this is a little bit more than just my
personal petty beef with Essence Magazine, because I think we're
(27:19):
going to be seeing more and more media companies and
brands using AI to communicate with their audiences. Again, I
don't believe that there's anything inherently in and of itself
wrong with that. In fact, I was kind of hoping,
you know, when we talked about this on the podcast,
that this could be a moment of media transparency, which
is always a good thing. Everybody everywhere right now is
(27:41):
talking about how we're all going to need to be
using AI in our work more. You know, we need
to get behind it. I can't tell you how many
times that is like what I have heard, like prepare
yourself for the AI revolution, blah blah blah. It would
not surprise me if people at Essence are like, oh,
we need to be experimenting with how we are using
AI in our work. Maybe it can make our work
more efficient, effective whatever. I did not bring it up
(28:03):
because I thought it was some huge scandal. I did
not bring it up because I wanted to embarrass Essence
maybe a little bit. But you know, publishing on social
media is inherently ridiculous and embarrassing, right Like, I think
if you are posting on public social media accounts, there's
always going to be that slip up, that thing of
(28:24):
like that sounded really cringey. That is part of the
process pointing that out, pointing out the messiness in cringiness
that is inherent in posting on social media is not
a bad thing. In fact, I think that transparency would
actually help us have a better understanding of this moment
that we are in when it comes to tech and
how it is going to shape how media companies and
brands engage with their audiences. I think for me, it
(28:46):
is really about transparency. You know. I was hopeful, like
when I saw that oh busted comment that Essence left
on my Instagram, I was hopeful that we were going
to be getting a moment of transparency. But then they
accuse me bridget of misreporting and spreading misinformation, and so
(29:07):
I just can't ignore that, not just because I'm a
petty bitch, which I am, but because I didn't say
that it was definitely one hundred percent, definitively written by AI.
I just said that I thought, in my opinion that
it seemed like it was written by AI, and it does.
It still does to me. That's still my opinion. But
they accused me of something that I take pretty seriously,
(29:28):
and like I was saying before, when we're talking about
marginalized communities especially, I do think that there is a
level of trust that you have with your community that
you're speaking to that is just important to honor. You know,
no one wants to feel like AI is being used
to mimic them, or to mock them, or to create
a kind of cruel, hollow stereotype about how the world
(29:52):
sees them. You know that a piece of technology is
just spitting out what they think the world thinks about
how we sound. You know that's not an authentic version
of who I am. That is just a fun house
mirror image of how the world sees me. And we
know our society is so full of biases. I don't
(30:13):
want to be kind of tricked into connecting with a hollow,
false depiction that is simply recreating what our bias society
already thinks about how I sound as a black woman.
For the sake of argument, to be clear, I am
not saying that this is what happened, But let's just say,
for the sake of argument, that they were hiring white
(30:35):
male social media editors who were then using AI to
mimic how they thought black women smoke and black women's
speech patterns. I probably would not be the only person who, like,
wasn't super comfortable with that dynamic, right. I don't like
the idea of, you know, being a black woman navigating
the Internet and then engaging with or identify with a
(31:01):
speaker who I assume is also a black woman like me,
but is actually AI right like that, that's not something
that sits well with me. That's not something that makes
me feel seen or comfortable. So I think, especially when
it's brands where their whole thing is about making a
marginalized community feel seen and feel supported and heard and amplified,
(31:25):
I do think there needs to be a little bit
of transparency around the use of AI there, if that
is what's going on. And so all of that is
to say, essence, you tell me, was that weird post
written by AI or was it published by a human editor?
And that's how I got on Facebook. I just want
to know you accused me of misreporting something. I would
(31:49):
be so pleased to come on here and say, looks
like I got it wrong. This is the human who
wrote that they were having a weird night, they misspelled
her name whatever. I've been there. Part of the reason
why I want to know is because I've been there,
like I. If you like, if you want to hear
some stories about like epic work fuck ups on the
(32:10):
social media account, talk to my old boss at MSNBC.
She will have some stories for you. Because I've been there,
I've done it. So if that's what's going on, totally
willing to correct the record, and even that is like
a transparency thing, like I'll never forget. My second favorite
social media mess up was a tweet from I think
(32:30):
it was the Boston Globe where they were using They
wanted to say their authorities are investigating, but for some
reason they've tweeted investigating and they had to come back
on and be like, we're not deleting the tweet. Like
we saw the tweet it was. It was incorrect, but
we're not deleting it because it's already been retweeted so
(32:52):
many times and it would be pointless. So yeah, we're
just leaving it up. And then that's it up. And
I respected it because a little transparency. Oh do you
need them just to be like, oh, yeah, we got
that's wrong, here's what happened, our bad and move forward.
That is the point of social media, not to get
so defensive and then so cagey when things don't come
(33:13):
off right. Social media is one big test kitchen of
trying things and seeing it they land, and they don't
always land, so just just acknowledge that.
Speaker 2 (33:21):
So bridget. What was your first favorite social media fuck up?
Speaker 1 (33:25):
Oh my god, it was the greatest day in social
media history. It was when Yahoo News was reporting about
the Trump administration trying to have a bigger navy, but
instead of the B in the word bigger, they used
an N and they tweeted it. And when I tell
(33:47):
you that, like even here here we go. When I
google it, one of the first things that comes up
is an Essence magazine article. So Essence, you're out there
doing the good reporting on social media. Like, yup, we're
still laughing at black Twitter's jokes following Yahoo finances navy tweet, right,
that was when I tell you, like, that was like
one of those Twitter moments where you just had to
(34:08):
be there. You really just had to like be there
in twenty seventeen when that happened, I'll tell you this,
I got no work done. Is too busy reading those
tweets because they were too funny.
Speaker 2 (34:19):
Uh, that was a pretty special moment.
Speaker 1 (34:21):
I think that was a pretty special moment. And you know, again,
I like Essence. I think they do good work. Sometimes
I've heard some rumors that sometimes they don't always pay
for that work, But that's not me saying that. That's
the interest saying that, look it up yourself. That's what
the streets say not me. But ultimately I think that
in this moment in the intersection of media and technology,
(34:43):
audiences deserve transparency, and audiences are smart enough to know
when something's up. All of those social media professionals, nobody
told them that this was AI. We were smart enough
as savvy audiences to say, this doesn't sound right. Your audiences,
you've got to trust them, and you've got to trust
that they are critical, savvy people. And so I think
(35:04):
in this moment where AI and all of this weird
new stuff is happening with technology, it is imperative that
media companies really trust their audiences and respect their audiences
and give us a little bit of transparency. And I
don't think it bodes well for how we move forward.
If a little Dog and a Pony show, that is
this podcast gently ribs Essence Magazine about maybe using AI
(35:28):
and like, oh, it didn't sound good. If how they
respond is to get super defensive, accuse me of spreading misinformation,
which I did not do, and then just like go silent.
I want this to be a dialogue. If Essence Magazine
can totally come on the show, but I have to
say until I get some information, some clarifying information from them.
(35:52):
I will go to my grave believing this is written
by AI. I am sorry, like it is my opinion.
I don't know. As I said, I have no proof,
but that I think that we would both be in
agreement that that post didn't sound right they edited it. Yeah,
So I guess that's really the bottom line. Like, this
is not just about my petty beef with them. It
is about what we could expect from brands, what transparency
(36:16):
we could expect, And I think early on we should
be established in some norms about about that transparency, right,
Like I do, I personally do think that media companies
should come up with some kind of an understanding of
how they will communicate when something has been generated by
AI to their audiences. The same thing that we were
talking about on the newscast last week about whether or
(36:37):
not platforms should include a watermark when an image is
AI generated.
Speaker 2 (36:43):
You believe that they should, Yeah, I absolutely do. I
was thinking about that. You know, It's like when you
have an AI generated image like there was that one,
uh we talked about last week of DeSantis, the Dasantus
campaign posting those images of Trump like hugging and kissing Fauci.
There should be a watermark indicating that, like that they're
(37:04):
fake images, right, And it shouldn't just fall on the
end user in the public to figure out what is
real and what is fake. That's but unfortunately, I guess
that's kind of where we're at, right. That's like one
of the key skills of literacy in twenty twenty three
is distinguishing what's real, what's fake, what was written by
(37:27):
a robot, what wasn't.
Speaker 1 (37:29):
I would argue that it shouldn't be Media literacy is important,
but it shouldn't be up to your average person to
be so critically engaged with the media that they mindlessly consume.
That's not how media consumption always works. You'll always have
that moment when you're on the toilet or just in
the checkout line, or don't haven't had your coffee where
you see something quickly and you're like, wait what. It
should not be up to individuals to have to parse that.
(37:52):
And I think that we need to start establishing some
guardrails and some guidelines for how brands and platforms are
going to ethically and responsibly use AI going forward, and
we got to do it soon. I think that's what
this whole kerfuffle with me an essence really comes down to,
(38:12):
you know, if I have to be the poster person
for those guardrails, so be it. I'll do it.
Speaker 2 (38:20):
Well, good luck to you. We need it. It's hard.
It's hard to know we need some guardrails, and so
Todd versus Essence. Here we go, the Ultimate Beef.
Speaker 1 (38:32):
The Ultimate Beef. Tune into season two of Beef. It'llills
be a like a slow Burn style breakdown between me
and my beef with Essence magazine. But listen, No but
in all series says, I want to know what folks think.
I'll put a copy of the original post in the Patreon.
Take a look. I'll do a poll. You do not
have to subscribe or be a paying member to see this.
(38:54):
It'll be a public post. That's how much I want
to know what y'all think. You can find the link
to the Patreon to vote, and through the post and
the description for this episode. Do you think it was
AI generated or do you not? I cannot be the
only one who sinks this, But if you really think
it was just a sloppy human, I want to know
that too. Let's get to the bottom of this together.
An essence, call me, got a story about an interesting
(39:19):
thing in tech, or just want to say hi. You
can reach us at Hello at tangody dot com. You
can also find transcripts for today's episode at tengody dot com.
There Are No Girls on the Internet was created by
me bridget Toad. It's a production of iHeartRadio, an unbossed creative.
Jonathan Strickland is our executive producer. Tari Harrison is our
producer and sound engineer. Michael Amado is our contributing producer.
I'm your host, bridget Tod. If you want to help
(39:41):
us grow, rate and review us on Apple Podcasts. For
more podcasts from iHeartRadio, check out the iHeartRadio app, Apple podcast,
or wherever you get your podcasts.
There Are No Girls on the Internet. As a production
of iHeartRadio and Unbossed Creative, I'm Bridget Toad and this
is there are No Girls on the Internet. So this
is going to be a little bit different of an
episode than the one that we initially had planned for today,
and that is because I have found myself personally in
(00:27):
a situation that I think really speaks to this specific
moment in technology, media, the Internet and where we're going
in the future. And as we're exploring this present future
of ours, we are probably going to find ourselves having
more and more moments like this one. So I wanted
to bring all of you along on this ride with me.
(00:48):
I am here with my producer, Mike. Mike, it is
not often that we are talking about drama. Drama might
be a little bit strong drama discourse dialogue that I
am at the center of. For something that I said
on the podcast a few weeks ago, thank you for
bravely being here to walk us through this journey.
Speaker 2 (01:06):
Yeah, thanks for having me along. I'm excited to have
a front row seat. Don't sell yourself short. It is
legit internet drama that you are involved in.
Speaker 1 (01:15):
I know. It's so much better to have internet low
stakes Internet drama when you're not involved in it. So
I kind of feel like I'm on the other end
of what I've enjoyed rubbernecking at when so many other
people are on the other end. So here's what's going down.
If you follow me on Instagram, you might have seen
that we had a little bit of drama about something
that we said on the podcast a few weeks ago.
In our newscast from May nineteenth, I was talking about
(01:36):
this very weirdly worded Facebook post that Essence Magazine published
on their Facebook page. Basically was a picture of actor
Viola Davis at the Can Film Festival looking beautiful. Side note,
I also might have mispronounced Can in that episode. I
have always said Con. I thought I was saying it correctly.
It's how I have said it my whole life. I
(01:58):
got some feedback that perhaps that it has pronounced can. I
looked it up and I saw both iterations of the pronunciation.
So Con, if you are French, if you are affiliated
with that festival, let me know if I'm saying it incorrectly. Can.
This is how I'm going to say it right now.
But any event, Essence was basically complimenting Viola Davis how
(02:19):
she looked in this picture. She looks amazing. She's wearing
all white. They put out a Facebook post and I'm
going to read word for word. We give credit when
credit is due, and Villa Davis misspelling theirs is looking
like an eight point fifty. Everyone at can can now
pack your bags and return home. This is a eight.
(02:40):
This is an eight. This is supper. God, damn, the
damn is broken and the flood of goodness has overflowed.
Can out of ten hashtag essens. So I speculated that
this post was not written by a human, It was
in fact written by AI. I believe that somebody specifically
(03:00):
asked an AI program to compliment someone's outfit and do
so in the voice of a black woman. That is
what I believe. That is my opinion. In the episode,
when we were talking about it, I think I made
it very clear. I said, I don't have any proof,
but no one will ever be able to convince me
that this was not written by AI. Just something about it.
I believe it was written by AI. That just seemed
(03:23):
like the most likely explanation to me, given how weird
that text was. I posted a clip from that episode
of a podcast talking about this to my Instagram page
and Essence Magazine Surprise Surprise replied, So when I woke
up on Saturday and saw that they were in my comments,
(03:45):
at first I was like pretty happy. I was like, wow,
I can't believe that this post made it all the
way to Essence. So their reply said busted call us
I Carly, and then a little robot emoji, which is
a reference to a Disney Channel show about two kind
of techy girls who run a website and a show online.
So I took that to mean they are playfully acknowledging Yep,
(04:09):
we're busted. We used AI to make this comment. That's
why it was so weird, No big deal. Didn't think
anything of it. In fact, I thought like, oh wow,
I am happy to see them playfully taking ownership of
being called out for using kind of a like a
janky AI voice. So then I replied, ah, love you guys,
(04:29):
hard emoji crying laughing emoji, because I was legitimately tickled
by this, right, and the whole clip was sort of
meant in jest, right, Like I wasn't dragging Essence Magazine.
I like Essence Magazine, and I think they do good reporting.
It was more about the pitfalls and fall and like
drawbacks of experimenting with ai voice in media. It really
(04:51):
could have been any brand that was not meant to
be an attack on essence. I did not feel it
that way. People listening to the episode, it's very jovial,
lighthearted with me. Okay, So they replied to my comment
that's like I love you guys, They say, we love
you too, but let's be more responsible with the reporting
(05:12):
of misinformation. So that you're right to guess that is
the appropriate response. What are your thoughts on that as
the producer of this podcast.
Speaker 2 (05:23):
Well, it does not feel good to be accused of
misinf spreading misinformation, right, Like, we go out of our
way to counter misinformation, and so that doesn't feel good.
And I can imagine that you know it being on
your personal Instagram felt extra not good. But it's also
(05:46):
a weird sequence of posts, right Like the first post
said busted, as if they were busted doing the thing
they were accused of doing.
Speaker 1 (05:55):
Who says busted when they're not busted, Like busted has
a very specific unless they were kind of trying to
play with it or something. But like busted, means oh no.
Speaker 2 (06:05):
Cause yeah, yeah, like I have been caught. I'm busted.
So it's weird to know what to make of it,
Like if they're if they are busted, then what was
the misinformation?
Speaker 1 (06:18):
Great question? So the misinformation comment I just could not
let stand. I fought that was a pretty strong way
to phrase something that I went out of my way
to say. I have no proof, this is my opinion.
I'm just saying what I think. I did also include
that someone at Essence did edit that post so that
(06:39):
it looked more It was more polished and didn't read
quite as wed. I still think it's weird in my opinion,
but somebody went in and edited it, so clearly we're
both on the same page that post something was up
with it, because why else would they edit it. So
to their misinformation comment, I reiterated that this was just
my opinion about how such a weird post wound up
on their pay I asked, like, if it wasn't Ai,
(07:03):
was it like a sloppy human? You didn't even spell
Viola Davis's name, right. I would think that any person
working at Essence, writing about entertainment especially would know how
to spell her name and so I said, if I
got it wrong, I am more than happy to correct
the record, and that still stands. I would love to
correct the record, but I need to know what happened.
Was it a human was it not AI? Like, what
(07:23):
are you disputing specifically? And then they went silent. So
this is a little bit of a personal thing with me.
I really have an issue when somebody is very loud,
the accusation is very loud, but then when you're like, okay,
well then what did I get wrong? When you're asked
for clarifying information, silent cannot I cannot stand that. It's
(07:44):
like a throw a rock and hide your hands kind
of thing, Like you were so loud accusing me of
misreporting misinformation about your publication, but you're unwilling to give
me any information about what exactly I got wrong and
how I got it wrong. I want to provide accurate
information to my listenership and my readership and use my
platform to do that, obviously, but it's telling to me
(08:08):
that they just went silent. So I don't engage in
a lot of drama. I'm a drama appreciate her. I'm
a drama connoisseur. This is a podcast but I'm wearing
a hat currently as we were recording that references my
love of petty drama. It's a hat that says I
don't know her, and it's a hat from my one
(08:28):
of my favorite podcasts, two Weekly. It's a reference to
Mariah Carey always when asked about her, one of her
nemesis is Nemesi j Lo. She always coolly responds with
I don't know her, which I think is a great
way to respond to drama without really engaging in it, but
still engaging in it. Love her. But here's the thing
(08:49):
about me. I am kind of a petty person. I
host a podcast called Beef that is about pettiness and
rivalries and why it's so important. So obviously I wasn't
going to let this go telling me that I'm misreporting
things and reporting misinformation, I'm not going to let that go.
I'm sorry. So this happened on a weekend and I
happened to have a lot of time on my hands.
(09:11):
So I made another comment and I was like, essence,
what is the truth? Didn't hear back? I made a
story with the with a screenshot of their initial post,
and I said essence in my Instagram comments has implied
that they did not use AI to write this post.
I asked them, I said, I will be happy to
correct the record if they give me some follow up
about what did happen? Was it a human error? You
(09:33):
know what happened here? Happy to correct, but need some information.
Still nothing. So that is why I am taking it
to the podcast. I feel like I need to talk
about it. I'm not intending the unletting it go until
I hear back from Essence and so if anyone knows
Essence or has a connection to them, I would love
to hear you have an invite to come on the show.
I actually able to think that would be great.
Speaker 2 (09:54):
So bridget what about this post made you think it
was written by AI?
Speaker 1 (09:58):
Great question. Again, this is my opinion. I have no proof.
I have no insight into Essence Magazine's editorial room. I
put out a poll on Instagram to say, like, do
people think this is AI or not? Overwhelmingly most people
think it's AI. Not that it's a scientific poll, but
one person said that they believe that it was written
by a human who was trying to do this like youthful,
(10:19):
voicy TikTok voice and just falling short reasonable. Right, Here's
why I think it was AI. I spend a lot
of time analyzing AI generated content, both visual content and
written content. And one thing that when it comes to
written text is AI can sometimes be a bit repetitive.
(10:40):
And so the fact that this essence post reiterated this
eight bit a few times that seemed a little AI
like to me, like this eight this is a eight,
this is a eight. I don't It's not something I
get that they are going for, like a very specific
(11:02):
black women you know slang, I say that you know
she ate that ate ni left no crumbs all the time,
that would be those are phrases that I've heard before
in the vernacular. The way that they put it in
that repetitive way, have never heard that before. And the
fact that it's repeating similar words I'm sorry, that is
(11:22):
something that AI it kind of does when you're looking
at AI generated text is repetition. So that was the
first thing that kind of made me think less is AI.
The second thing is a little bit harder to describe,
but you know it when you see it. It's just
that AI generated text just doesn't sound right the same
way that when you see an AI generated image sometimes
(11:44):
it just has a quality to it that's hard to define,
but you know it when you see it that tells
you this, this isn't quite right. Like, Mike, do you
remember a couple of years ago when people would be
playing with AI and be like, oh, we trained an
AI on watching fifty Hallmark Christmas movies and here's what
it generated. And it's the joke that the reason why
those things were funny was that, like they were almost right,
(12:07):
but something about them just was off. So you could
just tell, like, this is not a normal Christmas movie.
There's some detail or quality to it that makes it
just sound off. Do you know what I'm saying.
Speaker 2 (12:18):
Yeah, it'll it'll be like ninety ninety five percent of
the way they're just like nailing all the common tropes
and patterns from its training corpus, but that last five
or ten percent will just be like off the wall
and like stand out like a sort of thumb exactly.
Speaker 1 (12:35):
So that's that's what I think is going on here.
That's what this sounds like to me. Also, the fact
the post relies on two different metaphors. One is like
she ate. If you're not familiar with that, it's like
if somebody ate. It means that they whatever they're doing,
they're doing a good job of it, right, They ate
and left no crumb. She thought she ate that all
of that one and then two we give credit where
(12:56):
credit is due, And this is looking like an eight
P fifty a reference to credit scores, money you know
in your bag. To us, those two metaphors seem very
distinct to me. It's curious to me that someone would
put them in the same post. It just it just
seems so desparate that I have to question it, Like
it just sounds like somebody was trying to pack way
(13:19):
too many aave metaphors and slang terms into one sentence,
even metaphors that don't really relate to each other. And
that is something that we know that AI does. When
you ask AI to write something that's one of the
kind of giveaways that it might not be a human
and in fact it is AI, is that it's giving
you way too much. It's like a little bit over
(13:42):
the top in a way that makes it that much
more obvious that it's not a human. And I want
to get into a little bit more why I feel
like this post reads like AI to me. After a
quick break at our back. So another thing that just
(14:08):
strikes me as potentially one of the reasons why this
feels like AI to me, in my opinion, is that
it was part of a series of posts just breaking
down different celebrities outfits, and so it's you know, it's
gonna be like it's not like one post, it's gonna
be multiple posts as people walk the carpet at the festival.
(14:32):
And so I just having worked in social media before,
I've run social media for big news companies that you've
definitely heard of. I know that when you are in
a situation where you're like live live tweeting, or like
you're gonna be posting ten pictures and coming up with
posts and captions for like ten or twenty pictures in
(14:52):
an hour, that's a lot. I was doing this work
before AI was so commonplace in media. It would not
be surprising to me if a brand like Essence or
to be like, listen, we're gonna have one hundred images
of celebrities. We get the caption from this festival, let's
just run it in just let's just run it into
chat GPT and pull from there to make this quick.
That would that surprise me at all? And again, no shade.
(15:14):
That would not be any different than I'm sure any
other media company is doing right now. That is not
anything that is to be ashamed of. That's something bad.
I'm not even calling them out. It just like is
a reality of how brands and media companies are likely
right now already using AI, if not experimenting with it,
and certainly how they're going to be using it going forward,
(15:36):
I'm pretty sure. And because you know, media companies and
brands are already using this technology to communicate with us
the public, I wanted to play a little game, do
a little experiment that I think will just be another
bit of evidence as to why I feel so strongly
that this post was generated by AI. So I went
(15:58):
over to chat GPT and I asked Chatgypt to come
up with a short social media post about Viola Davis
her outfit at can. So first, when I put it in,
I said, make a post about how good Viola Davis's
outfit looked at can as a black woman. The first
thing that they gave me was very kind of formal.
(16:19):
It was like, let's take a moment to appreciate the
incomparable Viola Davis a true embodiment of elegance and beauty
who raised the red cart. But at this year's Can
Film Festival as a fellow black woman, I couldn't help
it be in all of her captivating presence and radiant charm.
So this when I just said do it as a
black woman, it was giving me you know, still pretty
(16:40):
polished like standard English. It was just the speaker is
just referencing that they are a black woman, which obviously
is not what this post sounds like. So then I
was like, ah, same thing, but give it to me
an aave African American vernacular English or like slang. Listen
to the difference in this in what they gave me.
Speaker 2 (17:02):
So this you're gonna read what chat GPT gave you.
In response to that prompt.
Speaker 1 (17:07):
That's right slaying the Can scene. Viola Davis looking snatched
hands emoji, fingernail painting emoji. Y'all, let's take a hot
minute to give some major props to our girl, Viola
Davis straight killing it at the Can Film Festival. Fire emoji.
She came through dripping with all that melon and magic
servant looks. That got a shouting Yes, Queen Viola was
(17:29):
straight up flawless, rocking that red carpet like nobody's business.
She had a swooning and a jaw dropping gown that
hugged all of her curves just right. And can we
talk about that melon and glow? Her skin was popping
like a summer sunset. Viola's talent is on a whole
nother level and another is like apostrophe than nt h
e er a whole another level. And she ain't afraid
(17:51):
to speak truth to power. She's using her platform to
shine a light on important issues and make a difference.
That's what we call a real boss lady. Y'all ain't
ready for the slag age that Viola David's brought to Can.
Her outfit had us all shook melon and poppin confidence
on bleak hashtag Viola Davis hashtag Can Film Festival hashtag
slaying it. So, I don't think I need to tell
(18:15):
you that. I feel like there's a lot going on
in what chat GPT gave us. But well, do you
have a reaction to it? Like what do you think?
What are your thoughts?
Speaker 2 (18:25):
It's a lot. Uh yeah, it really sounded kind of
similar to the post we were talking about.
Speaker 1 (18:35):
Uh.
Speaker 2 (18:36):
It referenced Melanin' three times repetition, right, Okay, it's generous.
Speaker 1 (18:42):
Something that jumps out at me that I think is
very similar to the Essence magazine Facebook post that I
believe was generated by AI was that it takes things
that are I wouldn't say disparate, but it just it
adds a lot in like it's like, it's not I
don't know that someone would say in a if they
(19:04):
were making a social media post. I would be surprised
if someone was using this many slang words, like if
you're saying yes queen, adding something that is on fleek
that's like a hat on a hat. Like it's just like,
I don't know a person who would speak that way.
And I think I see a lot of the similar
(19:26):
repetition of things that we got on that Essence post
to what I've what I've just generated from chat GPT.
I think it just it just seems very similar to me.
It seems similar in the ways of like making little
like references to things that are not totally slang, like
they kind of sound like slang, and if you read
(19:48):
it quickly, you know, you might be like, oh, that
sounds like how you know you would expect But then
when you actually person you're like, well, do I actually
know anybody who speaks like this? Is this what a
human would actually say? Or is this what AI pulling from?
Parts of the culture, parts of like everything that's out
there about what is the perception of what it sounds
(20:09):
like to be a black woman speaking in aave about
something that she likes. It just just feels like someone
doing a bad impression of black women and the way
that we talk online. It really reminds me a lot
of what Shafika Hudson described in our episode about and
Father's Day, as she describes it as obviously racist word salad, Like,
(20:32):
there's something about this that chat GPTs, the content that
chat GPT spit out that just reads like a black
woman impersonation word salad Like it makes all of the
things on their own makes sense. But when you zoom
out and you're like, well, is that really what someone
would string together if they were trying to create a
(20:53):
social post? In my opinion, probably.
Speaker 2 (20:56):
Not, Yeah, that's really a student. I think you that
it's like a hat on a hat, like it just
goes on and on with multiple metaphors, multiple things that
sound like they could be slang, but maybe aren't. Yeah,
it's and when you zoom out, it looks like what
(21:22):
a large language model trained on a corpus of perceptions
of what black women might sound like.
Speaker 1 (21:30):
And again, as cringey as all of this sounds, and
I know it sounds very cringey, I'm not even necessarily
pointing all of this out to call out Essence or
any media company, because again, Essence is not doing anything
that I don't think any other media company out there
is doing. I think this is commonplace. I think there
are times when it seems more when it is more
(21:51):
obvious than others. But I think this is something that
we're going to have to train ourselves to be aware
of and critical of, and be you know, thinking about
as we consume media as we go forward in this
particular moment in our tech our shared tech world.
Speaker 2 (22:12):
Yeah, you know, I keep thinking about the like, what
does it mean that it packed all of those different
metaphors and illusions and idioms in there, And like, maybe
that's what it looks like when part of the prompt
to chat GPT or one of these large language models
is to try to sound like a particular group, Is
(22:36):
that it just really hyper focuses on that part of
the assignment, and so it just keeps like hammering on
it over and over again in a really unnaturalistic way.
Speaker 1 (22:49):
And I also think that, you know, it's particularly salient
when you're talking about the way that marginalized groups communicate
with each other. You know, if you ask chat GBT
to say something in a stereotypical you know, like give
it to me an aave, or give it to me
(23:11):
as a gay person, or I just think that it's
easy to recognize when someone is being a caricature of
a marginalized person when we're seeing on television or in
a movie. I don't know that it's easy to recognize
it in text that is generated by AI. But I
(23:31):
think that the same way that we are thinking about
it when we consume movies and television, we should be
thinking about it, I guess, or now we should be.
We're going to have to be thinking about it as
we just consume content out in the world in this
era going forward.
Speaker 2 (23:48):
Yeah, I think we've all been seeing headlines for the
past I don't know, five or six months now ever
since AI really just like exploded into the public discourse
with these publicly available tools you know, we've all heard
the messaging that we need to start using AI and
figure out how to work AI into our professional workflows
(24:09):
so that we can be using it more and be
that much more efficient and effective and productive than people
who aren't using it, who are going to fall away
like dinosaurs and relics. And you know, I think a
lot of people are kind of questioning how true is
that and what does that even mean? But yeah, you
(24:31):
can't blame Essence for experimenting with it and you know,
trying to put AI to work for them if that's
what they were doing, which you know, like you said,
it seems like they probably were, so yeah, it's not
so shameful. It's probably a little embarrassing that they published
something that they then had to go back and edit,
(24:51):
but like you know, people do embarrassing stuff all the time.
Speaker 1 (24:55):
So I should also say that the idea like that
post this is not even really coming from me. As
I said, I'm a former social media editor and so
I'm still in a lot of these big, like online
professional groups dedicated to social media management and like best
practices and all of that. That is where I first
saw this post. That is where it first came on
(25:15):
my radar. Somebody not me posted a screenshot of the
original Essence magazine Facebook post put it in the group
and they were like, yo is Essence in here? Like
this is a really weird post, and so everybody was
roasting it, so certainly it wasn't just me, and the
consensus of social media industry professionals was that it was
(25:39):
written by Ai. So I can't even take credit for
that as an idea. I wish I could. I definitely
would have seen that in thought something's up. But the
reality is is that a large group of social media
professionals were roasting it and said it was Ai. I
happened to be one of them. But this is not
just coming from me. This was like a like an
(26:00):
online consensus of the space. So I'm sorry. I'm sure
that if you like for Essence Magazine, I'm sure that
is like difficult to hear. Nobody wants to hear that
their post was being roasted by a bunch of their peers.
But I hate to break it to you that is
what was going on.
Speaker 2 (26:15):
I'm suddenly concerned that it might have actually been written
by a human who is now like devastated that everyone
thinks they're a robot, and perhaps they're questioning themselves.
Speaker 1 (26:24):
You know, they should be taking one of those catch
but tests that are supposed to see if you're a robot.
It's like, oh d you dream of the electric sheep,
Like it's so hard you always fail though. This is
like I'm I sometimes think that you might be a
robot because I know when you have those tests, you're like,
I can't I guess I can't go to this website.
Speaker 2 (26:44):
Yeah, they're the worst because I feel like, yeah, I
marked all of the tiles that have bridges in them,
but like somehow I still fail. I don't know. It
seems unfair.
Speaker 1 (26:57):
More after a quick break, let's get right back into it. Okay,
So this is a little bit more than just my
personal petty beef with Essence Magazine, because I think we're
(27:19):
going to be seeing more and more media companies and
brands using AI to communicate with their audiences. Again, I
don't believe that there's anything inherently in and of itself
wrong with that. In fact, I was kind of hoping,
you know, when we talked about this on the podcast,
that this could be a moment of media transparency, which
is always a good thing. Everybody everywhere right now is
(27:41):
talking about how we're all going to need to be
using AI in our work more. You know, we need
to get behind it. I can't tell you how many
times that is like what I have heard, like prepare
yourself for the AI revolution, blah blah blah. It would
not surprise me if people at Essence are like, oh,
we need to be experimenting with how we are using
AI in our work. Maybe it can make our work
more efficient, effective whatever. I did not bring it up
(28:03):
because I thought it was some huge scandal. I did
not bring it up because I wanted to embarrass Essence
maybe a little bit. But you know, publishing on social
media is inherently ridiculous and embarrassing, right Like, I think
if you are posting on public social media accounts, there's
always going to be that slip up, that thing of
(28:24):
like that sounded really cringey. That is part of the
process pointing that out, pointing out the messiness in cringiness
that is inherent in posting on social media is not
a bad thing. In fact, I think that transparency would
actually help us have a better understanding of this moment
that we are in when it comes to tech and
how it is going to shape how media companies and
brands engage with their audiences. I think for me, it
(28:46):
is really about transparency. You know. I was hopeful, like
when I saw that oh busted comment that Essence left
on my Instagram, I was hopeful that we were going
to be getting a moment of transparency. But then they
accuse me bridget of misreporting and spreading misinformation, and so
(29:07):
I just can't ignore that, not just because I'm a
petty bitch, which I am, but because I didn't say
that it was definitely one hundred percent, definitively written by AI.
I just said that I thought, in my opinion that
it seemed like it was written by AI, and it does.
It still does to me. That's still my opinion. But
they accused me of something that I take pretty seriously,
(29:28):
and like I was saying before, when we're talking about
marginalized communities especially, I do think that there is a
level of trust that you have with your community that
you're speaking to that is just important to honor. You know,
no one wants to feel like AI is being used
to mimic them, or to mock them, or to create
a kind of cruel, hollow stereotype about how the world
(29:52):
sees them. You know that a piece of technology is
just spitting out what they think the world thinks about
how we sound. You know that's not an authentic version
of who I am. That is just a fun house
mirror image of how the world sees me. And we
know our society is so full of biases. I don't
(30:13):
want to be kind of tricked into connecting with a hollow,
false depiction that is simply recreating what our bias society
already thinks about how I sound as a black woman.
For the sake of argument, to be clear, I am
not saying that this is what happened, But let's just say,
for the sake of argument, that they were hiring white
(30:35):
male social media editors who were then using AI to
mimic how they thought black women smoke and black women's
speech patterns. I probably would not be the only person who, like,
wasn't super comfortable with that dynamic, right. I don't like
the idea of, you know, being a black woman navigating
the Internet and then engaging with or identify with a
(31:01):
speaker who I assume is also a black woman like me,
but is actually AI right like that, that's not something
that sits well with me. That's not something that makes
me feel seen or comfortable. So I think, especially when
it's brands where their whole thing is about making a
marginalized community feel seen and feel supported and heard and amplified,
(31:25):
I do think there needs to be a little bit
of transparency around the use of AI there, if that
is what's going on. And so all of that is
to say, essence, you tell me, was that weird post
written by AI or was it published by a human editor?
And that's how I got on Facebook. I just want
to know you accused me of misreporting something. I would
(31:49):
be so pleased to come on here and say, looks
like I got it wrong. This is the human who
wrote that they were having a weird night, they misspelled
her name whatever. I've been there. Part of the reason
why I want to know is because I've been there,
like I. If you like, if you want to hear
some stories about like epic work fuck ups on the
(32:10):
social media account, talk to my old boss at MSNBC.
She will have some stories for you. Because I've been there,
I've done it. So if that's what's going on, totally
willing to correct the record, and even that is like
a transparency thing, like I'll never forget. My second favorite
social media mess up was a tweet from I think
(32:30):
it was the Boston Globe where they were using They
wanted to say their authorities are investigating, but for some
reason they've tweeted investigating and they had to come back
on and be like, we're not deleting the tweet. Like
we saw the tweet it was. It was incorrect, but
we're not deleting it because it's already been retweeted so
(32:52):
many times and it would be pointless. So yeah, we're
just leaving it up. And then that's it up. And
I respected it because a little transparency. Oh do you
need them just to be like, oh, yeah, we got
that's wrong, here's what happened, our bad and move forward.
That is the point of social media, not to get
so defensive and then so cagey when things don't come
(33:13):
off right. Social media is one big test kitchen of
trying things and seeing it they land, and they don't
always land, so just just acknowledge that.
Speaker 2 (33:21):
So bridget. What was your first favorite social media fuck up?
Speaker 1 (33:25):
Oh my god, it was the greatest day in social
media history. It was when Yahoo News was reporting about
the Trump administration trying to have a bigger navy, but
instead of the B in the word bigger, they used
an N and they tweeted it. And when I tell
(33:47):
you that, like even here here we go. When I
google it, one of the first things that comes up
is an Essence magazine article. So Essence, you're out there
doing the good reporting on social media. Like, yup, we're
still laughing at black Twitter's jokes following Yahoo finances navy tweet, right,
that was when I tell you, like, that was like
one of those Twitter moments where you just had to
(34:08):
be there. You really just had to like be there
in twenty seventeen when that happened, I'll tell you this,
I got no work done. Is too busy reading those
tweets because they were too funny.
Speaker 2 (34:19):
Uh, that was a pretty special moment.
Speaker 1 (34:21):
I think that was a pretty special moment. And you know, again,
I like Essence. I think they do good work. Sometimes
I've heard some rumors that sometimes they don't always pay
for that work, But that's not me saying that. That's
the interest saying that, look it up yourself. That's what
the streets say not me. But ultimately I think that
in this moment in the intersection of media and technology,
(34:43):
audiences deserve transparency, and audiences are smart enough to know
when something's up. All of those social media professionals, nobody
told them that this was AI. We were smart enough
as savvy audiences to say, this doesn't sound right. Your audiences,
you've got to trust them, and you've got to trust
that they are critical, savvy people. And so I think
(35:04):
in this moment where AI and all of this weird
new stuff is happening with technology, it is imperative that
media companies really trust their audiences and respect their audiences
and give us a little bit of transparency. And I
don't think it bodes well for how we move forward.
If a little Dog and a Pony show, that is
this podcast gently ribs Essence Magazine about maybe using AI
(35:28):
and like, oh, it didn't sound good. If how they
respond is to get super defensive, accuse me of spreading misinformation,
which I did not do, and then just like go silent.
I want this to be a dialogue. If Essence Magazine
can totally come on the show, but I have to
say until I get some information, some clarifying information from them.
(35:52):
I will go to my grave believing this is written
by AI. I am sorry, like it is my opinion.
I don't know. As I said, I have no proof,
but that I think that we would both be in
agreement that that post didn't sound right they edited it. Yeah,
So I guess that's really the bottom line. Like, this
is not just about my petty beef with them. It
is about what we could expect from brands, what transparency
(36:16):
we could expect, And I think early on we should
be established in some norms about about that transparency, right,
Like I do, I personally do think that media companies
should come up with some kind of an understanding of
how they will communicate when something has been generated by
AI to their audiences. The same thing that we were
talking about on the newscast last week about whether or
(36:37):
not platforms should include a watermark when an image is
AI generated.
Speaker 2 (36:43):
You believe that they should, Yeah, I absolutely do. I
was thinking about that. You know, It's like when you
have an AI generated image like there was that one,
uh we talked about last week of DeSantis, the Dasantus
campaign posting those images of Trump like hugging and kissing Fauci.
There should be a watermark indicating that, like that they're
(37:04):
fake images, right, And it shouldn't just fall on the
end user in the public to figure out what is
real and what is fake. That's but unfortunately, I guess
that's kind of where we're at, right. That's like one
of the key skills of literacy in twenty twenty three
is distinguishing what's real, what's fake, what was written by
(37:27):
a robot, what wasn't.
Speaker 1 (37:29):
I would argue that it shouldn't be Media literacy is important,
but it shouldn't be up to your average person to
be so critically engaged with the media that they mindlessly consume.
That's not how media consumption always works. You'll always have
that moment when you're on the toilet or just in
the checkout line, or don't haven't had your coffee where
you see something quickly and you're like, wait what. It
should not be up to individuals to have to parse that.
(37:52):
And I think that we need to start establishing some
guardrails and some guidelines for how brands and platforms are
going to ethically and responsibly use AI going forward, and
we got to do it soon. I think that's what
this whole kerfuffle with me an essence really comes down to,
(38:12):
you know, if I have to be the poster person
for those guardrails, so be it. I'll do it.
Speaker 2 (38:20):
Well, good luck to you. We need it. It's hard.
It's hard to know we need some guardrails, and so
Todd versus Essence. Here we go, the Ultimate Beef.
Speaker 1 (38:32):
The Ultimate Beef. Tune into season two of Beef. It'llills
be a like a slow Burn style breakdown between me
and my beef with Essence magazine. But listen, No but
in all series says, I want to know what folks think.
I'll put a copy of the original post in the Patreon.
Take a look. I'll do a poll. You do not
have to subscribe or be a paying member to see this.
(38:54):
It'll be a public post. That's how much I want
to know what y'all think. You can find the link
to the Patreon to vote, and through the post and
the description for this episode. Do you think it was
AI generated or do you not? I cannot be the
only one who sinks this, But if you really think
it was just a sloppy human, I want to know
that too. Let's get to the bottom of this together.
An essence, call me, got a story about an interesting
(39:19):
thing in tech, or just want to say hi. You
can reach us at Hello at tangody dot com. You
can also find transcripts for today's episode at tengody dot com.
There Are No Girls on the Internet was created by
me bridget Toad. It's a production of iHeartRadio, an unbossed creative.
Jonathan Strickland is our executive producer. Tari Harrison is our
producer and sound engineer. Michael Amado is our contributing producer.
I'm your host, bridget Tod. If you want to help
(39:41):
us grow, rate and review us on Apple Podcasts. For
more podcasts from iHeartRadio, check out the iHeartRadio app, Apple podcast,
or wherever you get your podcasts.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.