I sat in the large, orange cushioned seat next to the front window of Cafe Bolívar, frantically
searching for a painting that may or may not exist. It wasn't that I particularly cared either way,
but rather a need for reassurance of material reality. To put it bluntly, I thought I might be going
crazy. The painting was one of two port depictions presented by my art history professor for a
comparison essay, meant to highlight the difference between neoclassical and impressionist art.
After a brief Google and Wikipedia search, I was confused at a lack of documentation. The
painter, Claude-Joseph Vernet, was decently well known, certainly no Caillebotte or Artemisia,
but nonetheless garnering a consistent Wikipedia viewing (“Pageviews Analysis,” 2026).
Despite this, I was unable to find a consistent name for the piece, and the only accurate results
in a reverse image search were from a variety of art printing sites, which all linked back to a
single stock photo. As I continued to search, I grew somewhat restless and agitated, perhaps
even paranoid. What sent me into such a frenzy was not the painting, which in retrospect I'm
sure exists somewhere, but a baseline level of fear that material reality is slowly eroding.
Such erosion, or rather the fear of it, does not manifest without cause. In 2026, it is rational and
likely sensible not to trust what you see on the internet, especially from sources that appear
trustworthy. This is due in part to the extensive proliferation of content that has been generated
by AI (Hadi et al., 2024; Wei & Tyson, 2024). The ability to create professionally presenting text,
images, videos, sound, and more without any effort beyond the writing of a prompt creates
many opportunities for the enterprising individual. Similarly to the invention of the printing press,
this drastically shifts the balance of information production towards the individual, who will not
always choose to propagate true information (Dewar, 1998). Naturally, within such an
ecosystem, the consumer of online content must develop the skill of recognizing generated
content, leading to an obsessive examination of everything. This in turn, requires a great deal of
effort and leads to less actual engagement and human interaction on the internet.
Unfortunately, use of artificial intelligence is not limited solely to individual experimentation.
Amongst other institutional uses, it has proved to be quite the effective tool in politics. From
deepfakes of political opponents, to campaign materials, to reactionary propaganda images,
use of generated material has become quite widespread in the American political sphere
(Warzel, 2024). These tools have particularly been coopted by the Trump administration,
although democrats are also no strangers to their use (Mancini, 2025; Verma & Kornfield,
2024). The ability to shape narratives with nonfactual information is certainly nothing new in the
realm of politics, but AI has made it far easier and more convincing.
The division sown by such propaganda is exacerbated by algorithms which isolate populations
into bubbles, in which information is inadvertently withheld that may contradict their beliefs.
People are shown only content that will confirm their biases, preventing the development ofnuanced understandings, and potentially fostering radicalization (Ahmmad et al., 2025;
Rodilosso, 2024). This is true across social media, and more widely as algorithms are
increasingly implemented into various platforms. These divides give rise to alternative facts,
allowing for completely different interpretations, if not understandings of an occurrence, and
polarization which was previously impossible (Ledwich & Zaitsev, 2019).
Algorithms are not completely to blame for this partitioning of reality. Another factor is the trend
of abandoning traditional news outlets, and an adoption of social media as a primary
information source. This mistrust in legacy media is somewhat ironic given the combination of
laissez-faire reporting and AI slop one finds on social media. Despite these factors, or
potentially because of them, at least half of American adults reportedly receive a portion of their
news from social media (Bick, Blandin, & Deming, n.d.). This trend reflects a lack of interest in
reading, particularly in children post-pandemic (Knight). With the infinite dopamine of the
internet, humans seem to be less drawn to the slow, less immediately rewarding, and
apparently archaic method of transferring information.
All of these factors paint a somewhat dire picture, and one much easier to find than my Vernet
harbor scene. Around us, truth slowly splinters, for now into bubbles, but perhaps in the future,
into shards no bigger than a few people's realities. It would appear that a singular, material
reality is slowly collapsing, if it has not already completely deflated. To me, this inspires a deep
fear. Information itself can no longer be reliably trusted, and the additional filtration or
decontamination required has induced a sort of paranoia. Evidently I question even the
information presented by my professor, who holds a PhD in art history. There exists a nagging,
unkillable idea that anything could be a lie. If I had to guess, I would say the painting exists
somewhere, hanging quaintly in a museum in Avignon or Nîmes, but the blind confidence in
such a detail is long extinct.
Ahmmad, M., Shahzad, K., Iqbal, A., & Latif, M. (2025). Trap of Social Media Algorithms: A
Systematic Review of Research on Filter Bubbles, Echo Chambers, and Their Impact on Youth.
Societies, 15(11). https://doi.org/10.3390/soc15110301
Bick, A., Blandin, A., & Deming, D. J. (n.d.). The Rapid Adoption of Generative AI.
Dewar, J. (1998). The Information Age and the Printing Press: Looking Backward to See Ahead.
RAND Corporation. https://doi.org/10.7249/P8014
Hadi, M. U., Al-Tashi, Q., Shah, A., Qureshi, R., Muneer, A., Irfan, M., Zafar, A., Shaikh, M.,
Akhtar, N., Wu, J., Mirjalili, S., & Shah, M. (2024). Large Language Models: A Comprehensive
Survey of its Applications, Challenges, Limitations, and Future Prospects.
https://doi.org/10.36227/techrxiv.23589741.v6
Knight, Lucy. “Children Reading Fewer, Less Challenging Books, UK and Ireland Study
Finds.” The Guardian, 4 June 2024, www.theguardian.com/books/article/2024/jun/04/children-read-fewer-less-challenging-books-uk-and-ireland-study-finds.
Ledwich, M., & Zaitsev, A. (2019). Algorithmic Extremism: Examining YouTube’s Rabbit Hole of
Radicalization (arXiv:1912.11211). arXiv. https://doi.org/10.48550/arXiv.1912.11211
Mancini, R. (2025, October 2). Gavin Newsom mocks Donald Trump with AI Marie Antoinette
image [Text]. The Hill. https://thehill.com/homenews/state-watch/5535091-newsom-trump-ai-
image-profile/
More Articles Are Now Created by AI Than Humans. (n.d.). Retrieved March 13, 2026, from
https://graphite.io/five-percent/more-articles-are-now-created-by-ai-than-humans
Rodilosso, E. (2024). Filter Bubbles and the Unfeeling: How AI for Social Media Can Foster
Extremism and Polarization. Philosophy & Technology, 37(2), 71.
https://doi.org/10.1007/s13347-024-00758-4
The State of Generative AI Adoption in 2025. (n.d.). Retrieved March 13, 2026, from
https://www.stlouisfed.org/on-the-economy/2025/nov/state-generative-ai-adoption-2025
Verma, P., & Kornfield, M. (2024, February 27). Democratic operative admits to commissioning
Biden AI robocall in New Hampshire. The Washington Post.
https://www.washingtonpost.com/technology/2024/02/26/ai-robocall-biden-new-hampshire/
Warzel, C. (2024, August 21). The MAGA Aesthetic Is AI Slop. The Atlantic.
https://www.theatlantic.com/technology/archive/2024/08/trump-posts-ai-image/679540/
Wei, Y., & Tyson, G. (2024). Understanding the Impact of AI-Generated Content on Social
Media: The Pixiv Case. Proceedings of the 32nd ACM International Conference on Multimedia,
MM ’24, 6813–6822. https://doi.org/10.1145/3664647.3680631
“Pageviews Analysis.” Wmcloud.org, 2026, pageviews.wmcloud.org/?
project=en.wikipedia.org&platform=all-access&agent=user&redirects=0&start=2025-
01&end=2026-02&pages=Claude-Joseph_Vernet. Accessed 14 Mar. 2026.
