Latest / Never Post / Reanimator: AI and Shared Histories

Transcript

Reanimator: AI and Shared Histories · Never Post

0:10
Mike Rugnetta: Friends, hello, and welcome to Never Post, a
0:13
podcast for and about the Internet. I'm your host, Mike
0:15
Rugnetta. This intro was written on Tuesday, 06/03/2025 at
0:20
08:27AM eastern, and we have a considered show for you this
0:26
week. In our second ever episode length segment, Hans confronts
0:31
the technological, moral, and philosophical impacts of using
0:36
artificial intelligence to animate still photos, including
0:41
those of both personal and historical significance. It's an
0:47
incredible, and as you've come to expect from Hans, incredibly
0:51
thoughtful piece, and I'm very excited for you to hear it.
0:54
And also pet sounds. But right now, we're gonna take a quick
0:59
break. You're gonna listen to some ads unless you're on the
1:01
member feed. And when we return, we're gonna talk about a few of
1:04
the things that have happened since the last time you heard
1:07
from us. Hold on to your butts.
1:10
I've got five stories for you this week. Academics left
1:15
Twitter for Blue Sky along with a lot of other people, but a new
1:20
study out of Zurich and The UK looks at academics specifically
1:24
asking why did those who switched make the switch. The
1:28
researchers call Blue Sky a, quote, rare natural experiment
1:32
for studying large scale migration from a long stable
1:36
platform, end quote, and find that, quote, the transition
1:39
behavior of accounts a user follows, followees, exerts
1:45
substantially stronger influence than the behavior of their
1:49
followers. In other words, academics and maybe people in
1:53
general, but who knows, follow information sources over access
1:58
to audiences when deciding to switch platforms. Researchers
2:03
also found that shock events, as they call them, account for huge
2:07
bumps in transition from one platform to another.
2:11
These are things like x instituting fees, the public
2:14
launch of Blue Sky when invite codes were no longer needed,
2:18
Brazil blocking x, and the biggest by far, the result of
2:22
the most recent US election. Speaking of x, CommunityNotes'
2:28
most prolific contributor is a cybersecurity startup. According
2:32
to Web three Antivirus founder Alex Dulube, the company
2:36
apparently uses a combination of machine learning and automation
2:40
tools to provide additional context in the community notes
2:44
section of x posts that forward crypto scams, which scam sniffer
2:49
estimates cost x users nearly $47,000,000 in March of this
2:55
year. Nancy Mace, the toilets obsessed Republican
3:00
congresswoman from South Carolina has apparently
3:03
commanded bot armies as well as her own staffers at the helm of
3:09
burners and sock puppets to counter criticism of her online.
3:13
This according to Wired.
3:15
We had to make multiple accounts, burner accounts, and
3:19
go and reply to comments saying that things weren't true even on
3:25
Reddit forums, a former staffer told Wired. We were
3:29
congressional staff and there were actual things we could be
3:32
doing to help the constituents. Mace, the staffers report,
3:36
describes herself as a self taught coder. RFK junior, the
3:42
brainworm having ist cabinet head secretary of the department
3:45
of health and human services, oversees the CDC, FDA, and so on
3:50
released his make our children healthy again report as part of
3:54
his so called make America healthy again commission, and it
3:58
contains citations for research which does not exist, likely
4:02
hallucinated by an LLM used to generate the report. According
4:07
to NODIS, at least seven of the studies cited in the report do
4:11
not exist, leading research scientists cited to clarify that
4:15
the work attributed to them is not real.
4:17
Other nonexistent studies are attributed to nonexistent
4:21
authors. And finally, Roblox grow a garden game hit 8,700,000
4:30
concurrent users at the end of last month. The game is a kind
4:33
of farming sim reminiscent of Animal Crossing in that the
4:37
landscape persists and changes while the player is away. Grow a
4:41
garden offers in game purchases for resources useful in the
4:45
tending of the farm, which Bloomberg reports have made
4:48
their way off platform, selling for inflated prices at various
4:52
digital goods marketplaces. Players of the game, which
4:56
accounts for nearly 30% of Roblox total activity at the
5:00
moment, spend millions of dollars a week buying secondhand
5:05
digital items according to marketplace site Eldorado.GG
5:09
founder, Vladis Jorkevicius.
5:12
According to Roblox terms of service, however, this is not
5:14
allowed. The company has yet to step in. That is the news I have
5:22
for you this week in this week's episode, Hans, on using
5:26
artificial intelligence to animate the past and the
5:30
feelings that inspires. But first, in our interstitials this
5:33
week, pet sounds.
5:36
Hans Buetow: Literally. Quick notice that this segment
6:29
contains discussions of war, execution, and murder.
6:43
City Council Member: Alright. Good evening and welcome to the
6:46
City of Falcon Heights City Council meeting on Wednesday,
6:50
03/26/2025. We'll call this meeting to order at 07:03PM.
6:57
Hans Buetow: Just a couple of months ago, my wife and I
6:59
attended our first ever city council meeting for our small
7:02
town. We were there along with about probably 80 other citizens
7:06
because on the agenda that night was the debate on public versus
7:12
paid street parking for the Minnesota State Fair. And unlike
7:17
most of the food at the fair, it was a pretty spicy night, which
7:21
started with the head of our local public access TV, Jeff
7:25
Ongstad, doing an annual report.
7:27
Jeff Ongstad: I had no idea there was this much interest in
7:29
community television. In 2025 and beyond, we anticipate
7:36
another 25% increase in all community coverage long form and
7:40
short form.
7:41
Hans Buetow: I love community services like this. So I was
7:44
pretty stoked to hear about covering local high school
7:47
sports and doing video for city council meetings.
7:50
Jeff Ongstad: We're also having a larger presence at community
7:52
events. So if you have events where we can have a table or a
7:57
tent or something, we'd like to be there, show some of our
7:59
stuff. We're buying a TV that you can play outside. So we're
8:02
excited about that.
8:04
Hans Buetow: And then towards the end of his presentation,
8:07
things suddenly changed.
8:09
Jeff Ongstad: And I'd like to show you just a little example.
8:10
We're embracing AI assisted production to enhance
8:13
capabilities we have along with speeding up our workflow.
8:15
Hans Buetow: Let's just pause here real quick. I want to
8:17
explain what I'm looking at in this moment. On the screen
8:21
behind Jeff Ongstad is a photograph that he's showing of
8:26
a group of three adult men and a dozen boys who look probably, I
8:30
don't know, 10 years old. They're all in a parking lot.
8:33
You can see the tail fin of a nineteen fifties car off to the
8:36
right.
8:37
The boys are looking in every direction, mostly at each other.
8:40
A few of them are looking at the camera, and, of course, one kid
8:43
in the back is blowing a raspberry.
8:44
Jeff Ongstad: Just one area that we're using is is like in old
8:47
photographs and stuff to animate those.
8:50
Hans Buetow: As he says this
8:51
Jeff Ongstad: So we can take old photos like this and not only
8:54
clean them up, but then bring them to life.
8:56
Hans Buetow: Everyone in the picture starts to move.
8:59
Jeff Ongstad: And that's just from an old photograph.
9:01
Hans Buetow: Like, literally, they all just start looking at
9:03
each other. Most start smiling. A few of them, like, mouth words
9:07
to each other. One kid bends down. The other kids all look at
9:10
him while he does that.
9:12
Even the trees in the background start to blow lightly in the
9:16
breezes. My first thought, this is cool. Followed by a strong
9:23
desire to do this with photos of my family, I am my generation's
9:28
keeper of the family history and object. So I have boxes, photos
9:33
going back several generations, which I've cataloged, I've
9:36
digitized, and I could put into AI a photo of my great
9:41
grandfather who I'm named after, but who died before I was born.
9:45
And this man who I never met would smile at me.
9:58
And I immediately think of our friend contributing producer
10:03
Tori Dominguez Peek, who last year filed a story for us about
10:07
companies who are trying to get her to make AI representations
10:10
of her dead mom for Mother's Day. And Tori's piece focused on
10:14
what feels like to reanimate your dead family members with
10:18
AI. And even though she's confident that she doesn't want
10:22
to do that, she doesn't want to build an AI version of her mom,
10:25
I'm not so sure.
10:34
Errol Morris: What I like about photographs is that they're
10:37
wormholes into history.
10:40
Hans Buetow: This is the director Errol Morris, and he's
10:42
doing an interview about his incredible book, believing is
10:45
seeing, which is all about the idea of truth and photography.
10:49
Errol Morris: What if I could walk into the picture itself,
10:52
look around, and ask myself, what am I really looking at?
10:56
What is really there? What is the reality that this photograph
11:01
is in part recording? It's the mystery of what we're looking
11:05
at. The mystery of what is it.
11:09
Hans Buetow: His is a very human inclination, one that has fueled
11:14
archaeology, genealogy, history, all kinds of science, like the
11:17
team who scanned the larynx of a 3,000 year old mummy so that
11:21
they could synthesize its voice, and the team that hooked up an
11:25
air compressor to the splayed open respiratory system of a
11:28
dead lion so they could get it to roar again. Yes. This desire
11:39
gives us this sense of connection, especially to those
11:43
we can never meet.
11:44
Errol Morris: As if somehow we can actually reach out and touch
11:48
the past. Right. It seems so close, particularly when it's a
11:55
photograph of someone that we have known in life or that is
12:00
deeply connected to us, as if somehow time really has been
12:04
defeated, which is one of the great dreams. Look, we can
12:11
defeat time.
12:13
Hans Buetow: This desire makes deep sense to me and feels
12:15
different from what's happening with the animations that I'm
12:19
seeing when I'm sitting in the city council meeting, which
12:21
those feel more risky. But why? Why is that? In Tori's story,
12:33
making a new photo of her dead mom herself, that feels like one
12:37
thing. Lowercase h history.
12:40
But in this room, these photos are of and for our town, and
12:46
they feel like they belong to all of us. They're shared
12:49
uppercase h history. If we start messing with our uppercase h
12:54
history, I think that feels very different, especially when we're
12:57
using the hand of AI to do it. After the city council meeting,
13:07
I was talking about this with my wife saying, I should probably
13:10
try out animating some photos to see what it's like. She
13:15
countered with a strong no.
13:18
There are just certain things that should not exist, she said.
13:21
This isn't a practical issue about could we. This is a moral
13:25
issue about should we. I started thinking then about photos that
13:32
I don't want AI to try to bring to life. Like what would the
13:36
computer imagine comes after Robert Capa's nineteen thirty
13:39
six photo, the one of a Spanish soldier whose mid collapse after
13:43
being shot in the head?
13:44
Or the 1930 photo by Lawrence Beitler of the lynching of
13:48
Thomas Ship and Abram Smith with their killers looking on
13:51
smiling. I feel like I know what AI would do with those grins,
13:56
those bodies. Photo after photo started springing to mind of
14:01
things that are shared, important images that document a
14:05
piece of our collective history. Over the next couple of weeks, I
14:11
started thinking about this technology and the photo that
14:14
has burrowed its way most into my memory. This photo is
14:19
gruesome or it's about to be gruesome.
14:21
Don Carlton: Well, this photograph, which has many
14:23
names, by the way, over the years, know, but it's been
14:27
generally called the Saigon Execution.
14:30
Hans Buetow: This is doctor Don Carlton. Don helped acquire the
14:34
estate of Eddie Adams who died in 02/2004. Eddie was the
14:38
Associated Press photographer who took the photo, the Saigon
14:41
execution, and it's one you've probably seen. Two men in a
14:45
street, one with a gun to the head of the other. So Don
14:51
specializes in the history of news media and the use of
14:53
historical photographs, and I asked him to tell me the story
14:57
of this photo.
14:59
And in case you hadn't gathered with a title like the Saigon
15:02
Execution, this is gonna be about an execution. So if that's
15:05
not what you wanna hear about, just skip ahead about three and
15:09
a half minutes. Okay. Here we go.
15:13
Don Carlton: He took the photograph during the Tet
15:15
Offensive, and the Vietnam War, which was in February of
15:19
nineteen sixty eight.
15:21
Hans Buetow: Eddie Adams was out scouting a story of a nearby
15:24
skirmish between police, soldiers, and Viet Cong.
15:27
Eddie Adams: Xolon is a Chinese section of Saigon.
15:30
Hans Buetow: This is Eddie in an interview with the AP prior to
15:33
his death.
15:33
Eddie Adams: And we got within a couple of blocks of the area. It
15:38
was very quiet. There wasn't any there's no movement. When
15:41
there's no movement, no people, you know that something is up.
15:45
Don Carlton: He saw when he was going to his car, he happened to
15:47
look over and see members of the national police there, the South
15:51
Vietnamese national police, leading this accused terrorist
15:55
just down the middle of the street. Gunfire was going all
15:58
over the place, so there wasn't much traffic. Okay?
16:01
Eddie Adams: And they just kept walking up maybe about a hundred
16:05
yards to the corner.
16:07
Don Carlton: This man, who was unidentified at the time, had
16:11
his hands tied behind his back.
16:13
Eddie Adams: And they stopped for a minute. I was about five
16:16
feet away from the prisoner.
16:19
Don Carlton: And the police told him that the man in their
16:21
custody was a Viet Cong. He had killed a police officer, one of
16:27
their fellow, you know, one of their colleagues in the police
16:30
department. So there was a lot of anger.
16:32
Eddie Adams: To my left came this guy, I had no idea.
16:36
Don Carlton: General Lone, he was their boss, walked up to the
16:39
group and he reached for his pistol.
16:42
Eddie Adams: He went over and I seen him go for his pistol.
16:44
Don Carlton: Eddie was standing there with his camera.
16:47
Eddie Adams: It had a 35 millimeter lens and a single
16:49
frame camera.
16:50
Don Carlton: He had just zero idea about what Long was going
16:54
to do.
16:55
Eddie Adams: Well, when somebody goes for their pistol, they
16:58
normally threaten. But I've taken pictures like that.
17:01
Somebody threatening somebody, you know, do this or I'm gonna
17:02
shoot you, and nothing ever happens. So I seen him go for
17:05
his pistol. As soon as he raised his pistol, I took one frame.
17:08
Don Carlton: At the moment, literally, that Lone fired the
17:12
shot
17:13
Eddie Adams: That was the incident that he shot him.
17:15
Don Carlton: Into the alleged terrorist's right temple, almost
17:19
point blank.
17:20
Eddie Adams: And I didn't I had no idea that he was gonna do
17:23
this.
17:25
Hans Buetow: The photo shows the two men with general loan in
17:28
military fatigues casually looking down his outstretched
17:30
arm to a revolver that's in his hand, and that gun is maybe a
17:34
few inches away from the head of Windvom Lem, who's a captain for
17:38
the Vietcong. Lem is dressed in a plaid shirt. He's got his arms
17:42
behind his back, and he's leaning a little bit away from
17:45
the general with his body, And he has his eyes closed, and he's
17:49
just forming an expression of surprise.
17:51
Eddie Adams: According to the US army, they said the bullet was
17:54
still in his head when the picture was taken.
17:59
Hans Buetow: I think for me, part of the power that has
18:01
always been in that and the reason I think about that photo
18:04
is because it's the moment right before you know what's gonna
18:09
happen and you get to live in this world where it doesn't
18:11
happen even though you know it has to happen.
18:15
Don Carlton: It's also the millisecond when that man is
18:17
still alive probably. If the bullet is in his head, he is
18:22
probably still alive. That's his last millisecond.
18:26
Hans Buetow: I think there's so much power in that.
18:29
Don Carlton: Well, it's a it is an incredibly powerful
18:31
photograph.
18:39
Eddie Adams: This
18:39
Hans Buetow: photograph makes me feel more things than almost any
18:44
other photograph in the world. I find it challenging actually to
18:47
look at. It makes my heart hurt for both of these men, for the
18:51
world that they had to survive in, and for that moment that
18:55
brought them together. When I look at this, I stare at Lon's
18:59
outstretched arm that's holding the gun and it feels to me like
19:04
there's this ray of energy that's between him and his
19:08
victim that's connecting them forever. It feels immediate and
19:13
eternal.
19:15
But it's not beautiful.
19:31
Eddie Adams: There's not a great work of art in terms of
19:33
photography. Number one, it's a wrong time of day. The light
19:37
wasn't right. The composition, it was terrible. But on the
19:41
other hand, that was a moment that was, I guess, very
19:43
important.
19:44
And I still don't believe, you know, I I still don't understand
19:48
to this day why it was so important because I've heard so
19:52
many different versions of what this picture did, like it helped
19:56
end the war in Vietnam.
19:58
Hans Buetow: The day after it was taken, the photograph
19:59
appeared on front pages of newspapers all around the world.
20:03
Don Carlton: I remember vividly when I first saw that
20:06
photograph. Houston Post and Houston Chronicle, both of them
20:10
featured prominently. And so when I saw it, I was shocked.
20:16
God, here is confirmation of the atrocities that are being waged
20:22
in this country. It was a repulsion.
20:26
Hans Buetow: This repulsion was broadly shared. The Saigon
20:30
execution is considered a powerful force in changing
20:33
public opinion about the Vietnam War. Eddie Adams won the
20:36
Pulitzer Prize, and the act and the people captured in it
20:39
changed the world. They changed history. I asked Don about the
20:47
AI animating technology, and that was really the reason that
20:50
I wanted to talk to him in the first place.
20:52
I wanted to know his views on it because of how complicated mine
20:56
were. I went looking for some services that do this, and I
21:04
found there's there's a few of them, that are out there. And I
21:07
found one that has different settings that you can put on it.
21:11
And one of the settings is you can, have two people in the
21:17
photograph hug. And I thought, what if I put Eddie Adams' photo
21:25
in to that algorithm, and instead of shooting him in the
21:31
head, the two men hug.
21:34
So if you can imagine that, right? Like I feed it in AI,
21:38
like how would that change the meaning of the photograph?
21:42
Don Carlton: Well, totally, mean, it's not reality any
21:45
longer. It's not what the photograph, the original
21:48
photograph depicted. So it's a totally new, it's a whole new
21:52
image that lacks any kind of realism to it because it's been
21:57
altered pretty profoundly when you do something like that. And
22:02
it's just AI in the way that you're talking about is going to
22:06
change that photograph. It's not going be the same photograph.
22:12
Hans Buetow: AI animating services let you do all kinds of
22:16
things. You can make people hug or French kiss or just wave and
22:22
smile. But I haven't done any of those things to any photo, let
22:27
alone the Saigon execution because both I don't want to and
22:32
because, frankly, I don't have to. There is video of that
22:36
moment because standing next to Eddie Adams that day was a news
22:41
camera person. Don't go watch the video if you haven't
22:44
already.
22:45
It's an execution. It's disturbing, but both Don and I
22:48
have watched it, and it feels so different from Eddie Adams'
22:53
photo.
22:54
Don Carlton: I would argue probably very few people have
22:57
seen the film, and if they have, very few remember much about it,
23:03
because it's up and gone in a flash.
23:06
Hans Buetow: A lot of the iconic capital h history photos have
23:09
accompanying moving images. The flag raising in Iwo Jima, the
23:13
Hindenburg crash, the liberation from the concentration camps.
23:17
Don Carlton: The napalm girl.
23:18
Hans Buetow: Napalm girl. Yeah. Which interestingly was also
23:21
filmed.
23:21
Don Carlton: Yes. That's true.
23:23
Hans Buetow: Yeah. But like that photo and Eddie's photo, like
23:27
have a life that those films don't and never will.
23:30
Don Carlton: No. No. No. No way. No way.
23:32
I mean, you can't use a sit there and look at the horror on
23:35
those kids faces. You're not gonna see those faces the same
23:39
way in a motion picture. It's so fast. I mean, it just but it
23:43
lets you sit there and stare into the face of this horror,
23:47
and think about it. You can't do that with a film.
23:58
Hans Buetow: There's something in the gap between still and
24:02
moving images that's helping me to start to understand what
24:04
feels so unsettling about those AI images that I first saw in
24:09
that city council meeting. Still and moving, they're not
24:13
interchangeable, and they affect us differently. They hold
24:16
different power, different potential, even different
24:19
truths. Since we created moving images in the nineteenth
24:24
century, we have always only been able to bridge that gap in
24:28
one direction. We take a still shot from the moving, but AI is
24:33
now letting us cross that chasm the other way.
24:39
After the break, we're going look at that chasm between a
24:42
still and a moving image and how bad we are at seeing what's at
24:46
the bottom.
24:56
Call In: Hey. Hi, Hans.
24:57
Hello. This is Kevin Myers. I am a video producer in the Twin
25:01
Cities.
25:02
It was interesting to get your question about how do I see the
25:07
difference between a still image and a moving image.
25:11
The difference between a still image and a moving image.
25:15
A still image and a moving image.
25:18
I have been thinking about the difference between a still image
25:23
and a moving image a lot.
25:26
Hans Buetow: To answer that for myself, I wanted to start by
25:30
getting a compass from my family, my friends, my friends
25:34
of friends.
25:36
Call In: A still image is like an individual.
25:39
It is an old man stooping to pick up his hat.
25:42
The inhale. And a moving image is like a family.
25:48
It unfolds like a river.
25:49
It's the exhale.
25:52
Hans Buetow: In some ways, scrutinizing difference between
25:55
a still and a moving image is a purely philosophical question.
25:59
But in other ways, it's deeply practical. So let's start there,
26:03
practically, and then work our way back to the philosophy of
26:07
it.
26:07
Call In: I associate moving images with sharing a narrative
26:11
and still images with needing to invent a narrative myself. I can
26:16
add my imagination to it. They're all kind of imagination,
26:21
all possibility. They can be interpreted in a million
26:25
different ways.
26:27
Hans Buetow: Science backs this up. Motion tells our brains
26:31
what's important to look at. In our animal brain, motion is much
26:36
more likely to be dangerous. It hijacks our attention. It means
26:40
that when there's motion in a movie, everyone's eyes are
26:43
following along.
26:44
A dozen people looking at the same thing altogether. Still
26:49
images, they don't tell us where to look so explicitly, at least
26:53
not on that, like, animal level. So a dozen people looking at a
26:57
painting are all gonna trace different paths with their eyes.
27:01
They will fixate less in any individual spot and they'll
27:05
shift around more. The synchronicity of moving images
27:09
has an impact.
27:10
A 2010 study of people watching movies in MRI machines showed
27:15
that along with joining up our eye movements, moving images
27:19
also synchronize our emotional responses to those images.
27:25
Everyone in the theater gasps all at once. No one in the
27:29
portrait gallery gasps at the same time, or even sometimes at
27:34
all. We generally experience still images as objects of
27:39
interpretation and exploration, and moving images as places for
27:43
emotion and engagement.
27:49
Call In: Sure. Minimal motion blur, but a 20 FPS is still 120
27:57
still images, one after the next. I think of all images as
28:01
still, that the word image implies stillness.
28:06
Hans Buetow: These perceptions of both still and moving images
28:10
all hinge on our ability to move through time.
28:14
John Berger: The most important thing about paintings themselves
28:18
is that their images are silent still.
28:23
Hans Buetow: That is John Berger, a legendary theorist
28:26
about images, art, reproduction, the observer, and time. Yes.
28:32
Moving images are a bunch of still images over time. It's all
28:37
an illusion. The movement is manufactured as the still images
28:41
collect over time into motion.
28:45
But the relationship between images and time is much more
28:48
complicated than just that. Here's Berger from his 1972 BBC
28:53
special, Ways of Seeing.
28:54
John Berger: It's as if the painting, absolutely still,
28:58
soundless, becomes a corridor connecting the moment it
29:03
represents with a moment at which you are looking at it. And
29:08
something travels down that corridor at a speed greater than
29:11
light, throwing into question our way of measuring time
29:16
itself.
29:17
Hans Buetow: Berger's corridor brings us into connection
29:20
through time across time. It narrows, it collapses, or it
29:26
freezes. As Laura Mulvey says in her book, Death 24 Times a
29:30
Second, quote, just as the cinema animates its still
29:34
frames, so it brings back to life in perfect fossil form
29:39
anyone it has ever recorded, from great star to fleeting
29:44
extra. It brings back to life. Back.
29:50
Necromancy. Wicked, thrilling, and tied to our dialogue with
29:54
death. Which brings us back to AI. AI to me does not feel like
30:03
necromancy. AI is creating a monster from scraps of reality.
30:09
A version of something that we recognize but really only
30:12
sideways that is animated by supernatural black box forces
30:19
that can give form to dark shapeless substances but cannot
30:24
bring into being the substance itself. Man, how ignorant art
30:31
thou in thy pride of wisdom. We think we can harness the
30:38
monster, control it, make trivial its consequences as we
30:43
channel lightning through stitches and flesh electrifying
30:46
the slab that holds a hulking behemoth. We think it will do
30:51
our bidding with no reason to have a life of its own. We have
30:56
fed our loved ones into the guts of a machine so that we can
31:00
watch them dance.
31:13
Call In: Still images, I I feel like have stood the test of time
31:16
more so maybe because they've been around longer. It can be
31:20
more iconic. I think it is much harder to have an iconic moving
31:24
image.
31:25
Hans Buetow: Time is also built up as provenance. When we stand
31:30
in front of an original painting, we marvel at the
31:33
content and the craft, but also a physical presence. Seeing
31:38
where the artist's brush left the canvas and knowing that your
31:42
eyes and their eyes have seen the same atoms. As John Berger
31:47
says, we believe in the power and value of the original, the
31:51
genuine, the verifiable, as a nearly religious object, not
31:58
because of what it shows us, but simply because it has survived
32:03
long enough to show it to us. Saigon execution feels like that
32:11
and an AI animation of an image feels to me the opposite of
32:17
that.
32:18
It doesn't feel transportative, but strangely static. It doesn't
32:22
feel connected, but rerouted, tangential. John Berger's
32:27
corridor in an AI animation only runs for a moment before it hits
32:32
the black box of ones and zeros.
32:40
Call In: I'm not sure why, but the first image that came to
32:44
mind is one that my grandfather took back in 1912. And the photo
32:52
is of my grandmother standing close-up in a field of corn
32:58
contemplating a corn cob. And I get to put myself kind of in
33:03
place of my grandfather and how did he see her in that moment.
33:08
We, I think, images. It still, I think, says more about the
33:17
person taking the image than about the image itself.
33:26
Hans Buetow: All images are subject to perspective. They are
33:29
not moments themselves, but windows into them.
33:34
Errol Morris: Images, by their very nature, they they rip a
33:39
piece out of the fabric of reality. They they take a swatch
33:45
out of reality. And in doing so, you you don't get to see above
33:52
and below. You don't get to see left and right. You just get to
33:58
see what's in the frame.
34:00
Hans Buetow: Errol Morris likes to speculate about the elephant
34:03
that's just outside the frame, the one that the photographer
34:06
chose not to include in their picture of their Florida
34:08
vacation. Someone said, move that elephant out of the way,
34:12
and then snapped the photo or took the video. Either way,
34:17
because of the possible elephant outside of every frame, we are
34:21
left only with the choices made by the framer.
34:25
John Berger: Paintings lend themselves to easy manipulation.
34:28
They can be used to make arguments or points which may be
34:31
different, very different from their original meaning. And
34:36
because paintings are essentially silent and still,
34:41
The most obvious way of manipulating them is by using
34:44
movement and sound.
34:47
Hans Buetow: John Berger here is talking about putting paintings
34:50
on TV. So zooming in on certain elements that will guide and
34:54
focus the audience with a new frame. Same thing with AI. By
34:57
animating a still image, you can completely change the frame. You
35:01
can pan right.
35:02
You can pan left. You can have the camera move down the street.
35:06
You can theoretically reveal the elephant. But what if there was
35:10
no elephant to be revealed?
35:15
Call In: There's more information in a moving image
35:19
for sure. There's just more context.
35:25
Hans Buetow: Beyond the frame is everything else.
35:28
John Berger: The meaning of an image can be changed according
35:31
to what you see beside it or what comes after it.
35:35
Hans Buetow: An old photo is always instantly recognizable as
35:40
an old photo. In looking at a lot of AI animations, however,
35:45
as soon as they start to move with their perfect frame rates,
35:48
their smooth skin, their eerily digital modernity, It puts the
35:53
old into a contemporary visual language. The antiquated and
35:58
distant becomes relatable and immediate. If you put it into
36:02
your feed, it wouldn't really stand out as an old photo. It
36:07
would be imbued with the essence of a TikTok.
36:11
But really, it would become less jarring to put next to a TikTok.
36:15
And when it's more comfortable to ingest, it becomes easier to
36:21
digest.
36:21
John Berger: What it means, in theory, is that reproduction of
36:25
works of art can be used by anybody for their own purposes.
36:30
Images can be used like words, we can talk with them.
36:34
Reproduction should make it easier to connect our experience
36:37
of art directly with other experiences.
36:40
Hans Buetow: We take them out of context, they start to say
36:43
something new, which can be good, even great sometimes. But
36:50
some swords have two edges.
37:04
Call In: A still image is, more easily controllable, more easy
37:11
to hide the truth, actually.
37:14
Hans Buetow: Are AI generated moving images a lie? Sure. But
37:20
so are moving images for all the reasons we've discussed framing,
37:24
intention, time. And so are all photographs. All of them tell a
37:29
limited version of what there is to know.
37:31
They are all lies. Or are they? Here's Errol Morris again.
37:38
Errol Morris: Well, don't believe that photographs are
37:41
either true or false. Never has made any sense to me. True seems
37:46
to be linguistic about a relationship of language to the
37:51
world, not of a photograph to the world. Also, I believe all
38:00
photographs are posed. So if someone tells me that one
38:07
photograph is more truthful than another because it was or wasn't
38:12
posed, I think that's very very close to nonsense talk.
38:19
Hans Buetow: This is a complicated conversation. There
38:21
are many smart people who have written and said many things
38:24
about this. But for our purposes, I don't think the
38:28
truth inherent in either moving or still images really matters
38:33
that much. I think about it sort of like Errol does. Like, I
38:37
think about one of my favorite TV shows, the reality show Love
38:41
Island.
38:43
Situation is ridiculous, but the emotions are real.
38:49
Call In: Yeah. There's all sorts of things to think about it.
38:53
Anyway
38:53
Anyways. I think that's all I got for you right now.
38:55
Those are my thoughts. Hope that helps.
38:57
Thanks again for reaching out. See you soon. Bye bye. Bye.
39:06
Hans Buetow: So why does all this matter? The way we
39:10
perceive, the way we think about time, the truth inherent in
39:14
images, the emotions that they evoke. Why do we care? Well,
39:22
it's because the emotions are real. On 11/13/2021, at a
39:33
stoplight in Arizona, a man named Chris Pelkey was shot and
39:37
killed by another man in a road rage incident.
39:41
After a three and a half year trial, the man who shot and
39:44
killed him, Gabriel Horcasidis, was convicted of manslaughter.
39:49
As first reported by four zero four Media, this trial marked
39:53
the first time that testimony from a murder victim was given
39:58
at his killer's sentencing, and it was accomplished using AI.
40:03
Pellke's sister wrote some copy and created an avatar that
40:06
looked and spoke like her dead brother.
40:17
Christopher Pelkey Avatar: Picture and my voice profile.
40:19
Hans Buetow: There's a brief video of Pelkey, who's an army
40:21
veteran, that they used to make the avatar, and then it cuts
40:24
back to the avatar who haltingly thanks the judge for all his
40:28
hard work. He says his appreciation for everyone who
40:31
testified on his behalf, and then he addresses his killer
40:34
directly.
40:36
Christopher Pelkey Avatar: So I would like to make my own impact
40:38
statement to Gabriel Horkasidas, the man who shot me. It is a
40:44
shame we encountered each other that day in those circumstances.
40:48
In another life, we probably could have been friends. I
40:52
believe in forgiveness and in God who forgives. I always have,
40:55
and I still do.
40:58
Hans Buetow: He's not really moving much. Just his mouth is
41:00
slightly animated. The audio is clipped at the end of every
41:04
sentence with clear AI artifacts, weird inflections,
41:08
multiple breaths, strange phrasing, and all of it was
41:12
written by his sister. This is weird enough to begin with. And
41:17
then I listened to what the judge, Todd Lang, had to say
41:21
about it.
41:25
Judge Lang: I love that AI. Thank you for that. And as angry
41:32
as you are and justifiably angry as the family is, I heard the
41:36
forgiveness, and I know mister Hortacitas appreciate it, but so
41:41
did I. As I said, I like to think I would do that. I don't
41:46
know that I would.
41:48
But I love the beauty in what Christopher and I call him
41:53
Christopher. I always call people by their last names. It's
41:55
a formality of the court, but I feel like calling him
41:57
Christopher as we've gotten to know him today. I feel that that
42:02
was genuine, that his obvious forgiveness of mister Jorge
42:07
Cicitas reflects the character I heard about today. But it also
42:12
says something about the family because you told me how angry
42:17
you were, and you demanded the maximum sentence.
42:21
And even though that's what you wanted, you allowed Chris to
42:25
speak from his heart as you saw it. I didn't hear him asking for
42:30
the maximum sentence.
42:33
Hans Buetow: Respectfully, your honor. No. You didn't. You
42:38
didn't hear it because he never said it because he's dead. What
42:44
you heard, everything that you heard, judge Lang, was the words
42:48
of his sister run through a machine, a sister who told four
42:53
zero four media in an interview that quote, our goal was to make
42:57
the judge cry.
42:59
Our goal was to bring Chris to life and humanize him. Judge
43:05
Lang knows that this was AI generated, but you can hear over
43:10
the course of his comments how much he anthropomorphizes the
43:13
simulation. He calls it Christopher. He begins talking
43:18
about the victim in the present tense. This has affected him.
43:23
Pelkey's sister accomplished her goal. So the prosecution had
43:28
been asking for a sentence of nine years. And then following
43:32
the testimony that he knew wasn't real, judge Lang gave
43:36
Gabriel Horkasidis the maximum sentence of ten and a half years
43:40
in prison. This isn't a personal decision. Do I make an AI model
43:47
of my dead brother?
43:49
This is a societal question. Where? How? And how much of this
43:55
do we tolerate? I don't know where to draw that line.
44:00
But one thing this makes absolutely clear to me is that
44:03
even people whose job it is to soundly, dispassionately, and
44:08
accurately evaluate and judge things seem to be confused.
44:14
Maybe this is making a mountain out of a molehill. This is just
44:17
one person, after all. But this is person in a position of
44:21
power, who has now set a precedent. And this is not not
44:28
worrying.
44:30
I'm not not worried.
44:39
Judge Lang: I'm not
44:39
Jeff Ongstad: really sure why it's doing that.
44:41
Hans Buetow: So there I am look too close. In this council
44:42
meeting this past March, watching example after example
44:45
of AI animating still images turning into videos.
44:48
Jeff Ongstad: But it is kind of amazing how it can take these
44:50
and have all these kids move and do stuff and we're this kid
44:54
keeps changing to different people.
44:56
Hans Buetow: I'm tempted by it.
44:58
Jeff Ongstad: But it can look at something
44:59
Hans Buetow: I'm repulsed by it. I feel like I can hold both of
45:04
those at once.
45:05
Jeff Ongstad: And like if you looked at this kind of stuff,
45:07
probably like even six months ago, was nowhere near this and I
45:12
figure in the next six months it'll progress even more. And I
45:15
know
45:15
Hans Buetow: But I'm thinking, what do we think we're learning
45:18
from this? As media consumers, how much do we understand about
45:22
the power we're witnessing or wielding ourselves? I'm not sure
45:27
what lesson we're taking or more probably not taking.
45:31
Jeff Ongstad: Something
45:33
Hans Buetow: is happening. I'm not sure what, but we can see
45:39
that it's already having an impact.
45:40
Jeff Ongstad: Do you have any questions I'd be happy to
45:42
answer?
45:45
City Council Member: Questions from counsel? Jeff, thank you
45:49
very much.
45:49
Jeff Ongstad: Thank you. And I can leave support we have.
45:52
City Council Member: And yeah. This is great. And
45:55
Jeff Ongstad: I'll use it I'll use my power wisely from now.
46:00
Thank you very much.
46:01
City Council Member: Thank you. Appreciate you taking the time
46:04
to come out and give us the annual report.
46:23
Hans Buetow: Huge, huge thank you to Doctor. Don Carlton for
46:28
talking to me for this segment. You can learn more about the
46:31
Eddie Adams collection and the whole of the Dolph Briscoe
46:35
Center, it's very cool, at briscoecenter.org. Thank you
46:40
also to Amy, Kevin, Kristen, Munna, Seth, Martha, Dana,
46:45
Eddie, James, and everyone else who submitted their thoughts and
46:49
recordings about the difference between a still and a moving
46:52
image. I am interested in what you think that difference is and
46:59
whether or not you would animate any of your family photos or any
47:02
other photos.
47:04
Please let me know. Get in touch at 651615507, drop us a line at
47:10
our website, or email us at the neverpost@gmail.com. I wanna
47:15
know how you feel about all of it. All that in the show notes.
48:20
Mike Rugnetta: That is the show we have for you this week. We're
48:36
gonna be back here in the main feed on Wednesday, June 18. In
48:40
2019, Rove Concepts, a furniture manufacturer, surveyed 2,000
48:46
American households and found that the average couch contained
48:51
$1.55 in change. That is nearly halfway to a monthly Neverpost
48:58
membership. Go check the car.
49:00
Maybe that winter coat you put away a couple months ago. The
49:02
chair by the window. The one that you love to read in. That
49:05
has gotta have at least 30¢ in it. What I'm saying here is that
49:09
there is $4 just laying around your house that you're not even
49:15
using and we will.
49:18
We'll use it to pay our hosting costs. We'll use it to pay
49:21
contributing producers. We'll put it towards our research
49:24
budget and more. Never poe dot s t to become a member. $4 a
49:28
month.
49:29
That's just over two and a half couches worth of change. Never
49:39
post producers are Audrey Evans, Georgia Hampton, and the
49:41
mysterious. Doctor first name, last name. Our senior producer
49:44
is Hans Buto. Our executive producer is Jason Oberholzer,
49:48
and the show's host, that's me, is Mike Rignetta.
49:55
Moving, variant ornithography of those uninitiated made into
50:00
memory by the me briefly incarnate. Full of myself on
50:06
successive nights dense and alone sings you back. Need keeps
50:11
the book of dying open, the language common after all.
50:15
Relieved, the task finally changing prompts, tapping my
50:20
reserve feeling now wise to its edge. Where are you risk any
50:25
detail of what's in me having been tricked by that image of
50:30
man?
50:32
Excerpt of Fourth Fourth by Gil Ott. Never Post is a production
50:40
of Charts and Leisure and is distributed by Radiotopia.