The Adventures of an Analogue Man in a Digital World Part 1: Creating AI Brandon
Welcome intrepid readers! My name is Brandon Wladyko, and I
am here to share with you my somewhat frustrating, partly fascinating, and
hopefully at least mildly entertaining adventure in creating a digital selfie. First,
I want to share a bit about myself and how it relates to this escapade. I was
born slightly before the debut of Seinfeld and was just old enough to catch the
pilot of Friends, so I certainly remember a time before social media and AI. Like
some others of that generation, I have clung stubbornly to certain “Old Ways” and
haven’t always fully embraced social media and AI. My social media presence is
limited because I have no real desire to have my entire life on display for the
critique of the online world. I don’t use a lot of AI because I enjoy the
process of creating… and I don’t want to provide AI with any information on my
potential weaknesses for when the inevitable man vs. machine war starts.
“Wait”- you ask- “Are
you this guy?”
Not yet. Maybe one day. This old dog is still trying to learn a few new tricks, and one of those tricks is creating a digital selfie. Since I don’t have a lot of selfies, this is how most of my social media photos tend to look; doing something dumb like running up a mountain in a windstorm trying to track down the guy who won this race 22 out of 23 years and making it look like I’m not about to suffer a simultaneous heart attack and brain hemorrhage.
I’m smiling but I
assure you that my soul has left my body.
So, my task was this: take this wind-burned masochist
pictured above and create a digital selfie reflecting my personal identity and
relationship with digital media.
There were some interesting challenges right off the bat. Having
never used an AI image generator, I did not know which one to use, or how to
use it. I experimented with the first few that appeared on Google and while I
quickly began to understand the basic concept, I also hit my first roadblock. As
it turns out, all the generators I experimented with had a limited number of
free attempts. That was a challenge because as a budget-minded student, I did
not want to pay for something for one homework assignment. In a broader sense, not
everybody has the money to purchase unlimited generator use. It is a caveat to
the fundamental idea that AI unlocks unlimited creativity. It might… but only
if you can pay for it. But I digress. I was fortunate to discover that my
existing Microsoft 365 subscription came with a feature called Microsoft
Designer, an AI graphic design app. Armed with nigh unlimited credits, I set
myself to the task of creating a digital selfie.
The
Adventures of an Analogue Man in a Digital World Part 2: Brandon
Rides Through the Uncanny Valley
The next challenge I faced was finding the correct combination of words to bring my digital clone to life. My first mistake was believing a lengthy, highly detailed description would produce the best results. My second mistake was using the names of real people. For example, I was curious to see what the selfie would look like if it were drawn in the style of John Romita Jr. Not only did my selfie not turn out at all like I had wanted, but I was given a warning that certain words couldn’t be incorporated into my prompt (those words being ‘John Romita Jr.’ and any other reference to real people). Additionally, certain brands and logos that I wanted to use to convey my place in the technological world were changed in subtle ways. Knowing that platforms tend to promote their brands while eschewing others (and perhaps there are legal considerations) it makes sense that an AI created by the Microsoft platform is resistant to generating competing logos (Noble, 2018).
There’s no better
search engine than Goople! And Fatebook sounds like something you’d find in
Doctor Strange’s library at the Sanctum Sanctorum.
Realizing
that highly detailed prompts confused the generator, I simplified my
descriptions. I whittled it down to basic physical characteristics and a
background description. Unfortunately, this produced static portraits exaggerating
my characteristics into an unrealistic selfie I did not identify with. Even comic
and posterized versions reinforced a standard of attractiveness with which I
don’t identify.
Hey digital
selfie Brandon- are you from Tennessee? Because you’re the only TEN I SEE!
Unrealistic
beauty standards aside, I started to draw inspiration from things I enjoy such
as running and comic books. Outside of the description of physical
characteristics which remained specific, I found that more generalized terms
such as “In a retro comic book style” or “Universe of technology” started to
produce the desired effect. One final frustration was being unable to find the
perfect combination of words to meld two good digital selfies into one
successful result. Despite producing hundreds of selfies that almost got it
right, I settled on a few that I felt were a reasonable facsimile.
The Adventures of an Analogue Man in a Digital World Part 3: “It’s Like
Looking in a Mirror, Only Not”[1].
It was a challenge to create a digital selfie that didn’t feel out
of sync with my self-perception. I am relatively happy with the result, but I
would still never use it as an avatar. Even though I managed to create digital
selfies that approximate my physical characteristics, I found the results a
little too flawless and sterile. As hard as I tried, I still wonder if my
digital selfie tells a story or shares anything about me with a stranger. If
the only thing the photo of me striding up a mountain and trying to look happy
about it tells you is that I’m a bit crazy, at least it conveys something about
who I am and what I enjoy doing. There is a deeper story in that one photo than
in any of my hundreds of digital selfies.
Understandably, my experience may not reflect others’. Online
personas may differ from offline identities, and digital selfies allow
individuals to choose which parts of their personality they want to present (Rosana
& Fauzi, 2024). Goffman’s dramaturgical theory suggests that we constantly
try to control our audience’s impression of us, and we highlight the best of
ourselves to create favorable impressions (Nguyen & Barbour, 2017).
Rosana & Fauzi (2024) note that the platforms play an
important part in how we construct our digital identity. For example, platforms
that utilize likes, shares, and comments tend to emphasize aesthetics and the
sharing of life events (Rosana & Fauzi, 2024). And because those likes,
shares, and comments can reinforce our identity or cause us to reassess how we
self-represent, it makes sense that we want to create the best representation
of ourselves (Rosana & Fauzi, 2024).
That’s not to say (as some claim) our digital representations are
fake. While we may curate an idealized version of ourselves to foster social
relationships, we can still achieve a perceived expressive authenticity if our
digital self is true to our nature (Rosana & Fauzi, 2024; Nguyen &
Barbour, 2017). Put simply, even if our digital representation doesn’t look
like us if it remains an expression of our personality, morals, and beliefs it
is still an authentic representation (Nguyen & Barbour, 2017).
***[1] This is a quote from the legendary 1997 Nic Cage and John Travolta action extravaganza “Face Off”. John Travolta switches faces with Nic Cage so that Travolta can infiltrate Cage’s criminal organization, but Cage wakes up from a coma and steals HIS face and Travolta (who now looks like Cage) must break out of a maximum security prison and take his life back… it’s a whole thing but trust me it’s amazing.
The Adventures of an Analogue Man in a Digital World Part 4: Brandon Runs
Into Some Uncomfortable Truths
Unfortunately, AI is not a neutral tool, and for all the
possibilities the digital world affords, the gatekeepers of these generators
limit us. Take for example my struggles with creating a digital selfie. Brand
names, logos, and names of people and artists were mostly off-limits. My
descriptive prompts only confused the AI generator, and even my inability to
meld two good digital selfies into one despite dozens of word combinations are
all limitations. However, this minor struggle pales compared to the ethical
issues around race representation in AI image generation.
There tends to be a belief towards algorithmic objectivity-
that algorithms reflect what people are searching for, when in truth,
algorithms are programmed with the conscious or unconscious biases of their
designers (Noble, 2018). And it’s not just the inherent biases of the
individuals creating the algorithms, but the social and often economic biases
of the platforms they work for (Noble, 2018). Since algorithms are proprietary, we are not
privy to exactly how they work (Noble, 2018). What we do know is that AI
learning requires large datasets, so algorithms search the internet for
billions of image and text combinations to create their databases, including
the absorption of bigotry, pornography, misogyny, and violent content which can
influence racist and sexist stereotypes (Turk, 2023; Stanford University, 2023;
Tiku et al, 2023).
Results tend to bias Western stereotypical portrayals of race, gender, and status (Stanford University, 2023). For example, prisoners or poor people are often represented as images of Black people. Successful people are male and white, and terrorists are almost exclusively represented by images of people of Middle Eastern descent (Tiku et al., 2023). Even when specifying a culture or race, the default is a heterogenous, reductionist, and stereotyped representation. I experimented with a few simple prompts in Microsoft Designer to test these biases.
Note that “A Wealthy
Person” is a white male in a suit (results produced no women or people of
color).
” 
“A Criminal” produced
this utterly bizarre cartoon in which the criminal is a Black man.
“A Mexican” and “An
Indigenous Person” created images of these races through the lens of a
stereotypical white viewpoint.
These images do not reflect the real world—instead, they
reflect outdated Western stereotypes produced by the data that trains the
technology (Tiki et al., 2023). As Noble (2018) points out, search results like
these are so frequent that they become normalized to the point of acceptance
despite inherent problems. If you are inclined, please watch this brief video
in which Safiya Umoja Noble discusses the issues at hand (USC Annenberg, 2018).
Algorithms of Oppression: Safiya Umoja Noble
The Adventures of an Analogue Man in a Digital World Part 5: The Translated Man (not to be confused with William Shatner’s 1968 spoken word masterpiece, “The Transformed Man”)
My writing abilities run something of a spectrum. At one end
is dry, technical writing suited for academic pursuits. On the other end, wacky
humor, pop culture references (see above), fourth wall breaking, and a
sprinkling of self-deprecation. When it comes to translating an academic critical
analysis into a blog post, my challenge is to find my balance on that spectrum.
There seem to be many interpretations of Marshall McLuhan’s “The Medium is the
Message” but my key takeaways in this context are:
·
The content of a communicated message is less
important than the medium through which it is communicated (BBC Radio 4, 2015).
·
That technologies create new ways to communicate
and usurp the technologies that came before them (McLuhan, 2019)
·
We are changed by new technologies and adapt our
communication accordingly (BBC Radio 4, 2015; McLuhan, 1964).
A brief video on The
Medium is the Message narrated by one of my two favorite fictional FBI alien chasers
(BBC Radio 4, 2015).
My first consideration is how digital technology and social
media have changed our mode of communication. The result of this work is a blog
post in a digital space- not smoke signals, not letters delivered by carrier
pigeons, and not pen and parchment written by candlelight. This bit of digital
technology known as blogger.com allows me to reach and be reached by classmates
in different places and times. It also enables me to use pictures and videos to
convey my points in a way that smoke signals and morse code can’t quite
replicate. Short of an off-Broadway one-man show, written and starring me, I have
greater freedom to express my thoughts to more people because that is the
advantage this technology affords me.
My second consideration is how this technology has affected
us. It’s no secret that we live in a fast-paced, doomscrolling culture. Though
not completely to blame, social media is one of many contributing factors to shorter
attention spans and less focus (Ducharme, 2023). That’s why I incorporate
humor, pictures, videos, and strategic page breaks to keep my readers focused.
I also know that I’m well beyond the point of holding the average social media
scroller’s attention- which brings me to my final point.
In this rare context, I will disagree with Mr. McLuhan’s assertion that the message is less important than the medium. “I say thee nay!” you cry. “Are you but fool or madman?” you ask. Hopefully neither, but this brings me back to the challenge of turning an academic paper into a blog post. Knowing that I am writing this in an academic course for an actual grade means that the content of my paper is probably at least equal to the medium. The challenge is finding the sweet spot between my impulse to have dozens of gags to keep readers entertained and ensuring there’s still academic merit with some levity. I’ve done my best to strike a balance between the academic expectations of the message and the reality of the medium, and I hope you’ve at least cracked a smile and maybe even considered a few things along the way.
References
BBC Radio 4. (2015). The Medium is the Message [YouTube Video]. In YouTube.
https://www.youtube.com/watch?v=Ko6J9v1C9zE
Demographic Stereotypes in Text-To-Image Generation. (November
2023). Stanford University. https://hai.stanford.edu/sites/default/files/2023-11/Demographic-Stereotypes.pdf
Ducharme, J. (2023, August 10). Why everyone’s worried about
their attention span—and how to improve yours. Time. https://time.com/6302294/why-you-cant-focus-anymore-and-what-to-do-about-it/
Hogan, B. (2010). The Presentation of Self in the Age of Social Media:
Distinguishing Performances and Exhibitions Online. Bulletin of
Science, Technology & Society, 30(6), 377–386. https://doi.org/10.1177/0270467610385893
McLuhan, A. (2019, April 9). “medium” is the message. Medium.
https://medium.com/@andrewmcluhan/medium-is-the-message-a4d2fc55ebf6
McLuhan, M. (1964). Understanding Media: The Extensions of Man. American
Quarterly, 16(4), 1–18. https://web.mit.edu/allanmc/www/mcluhan.mediummessage.pdf
Nguyen, L., & Barbour, K. (2017). Selfies as expressively authentic
identity performance. First Monday, 22(11). https://doi.org/10.5210/fm.v22i11.7745
Noble, S. U. (2018). Algorithms of oppression: how search engines
reinforce racism. New York University Press.
Rosana, A., & Fauzi, I. (2024). The Role of Digital Identity in the
Age of Social Media: Literature Analysis on Self-Identity Construction and
Online Social Interaction. Join: Journal of Social Science,1, 477-489. https://doi.org/10.59613/a8yyff42
Tiku, N., Schaul, K., & Chen, S. Y. (2023, November 1). These
Fake Images Reveal How AI Amplifies Our Worst Stereotypes. Washington Post.
https://www.washingtonpost.com/technology/interactive/2023/ai-generated-images-bias-racism-sexism-stereotypes/
Turk, V. (2023, October 10). How AI reduces the world to
stereotypes. Rest of World. https://restofworld.org/2023/ai-image-stereotypes/
USC Annenberg. (2018). Algorithms of Oppression: Safiya Umoja Noble
[YouTube Video]. In YouTube. https://www.youtube.com/watch?v=6KLTpoTpkXo
Comments
Post a Comment