What does AI mean for creativity, aesthetics, marks… and humanity?

TL;DR: Bad clipart and Papyrus might tell you more about a student’s understanding than any AI poster ever will. Also, I curated a collection of poems about what AI can’t be. It’s here, with content warnings below.

In my spare time, I have begun de-AI-ing posters. There’s something about an AI-generated image that makes my head hurt. It’s too uncanny, with too little negative space, and too… much. The stupid paintbrush smear behind each letter, and that font that’s the same in every AI poster and advertisement you see. It’s off and full of overused phrases. It’s so optimized that it’s not human.

AI writing gives me the same feeling, but there’s something about images that makes that disconnect just a step more jarring for me.

Which one of these posters shows more understanding??

Several posts back, I hypothesized that we are moving into a two-peak paradigm, where traditional aesthetics are no longer a reasonable metric for the depth and comprehension shown by a piece of work. At one peak, we’ll see hand-scrawled projects, untouched by any computer program, that would once have been considered bad work. Yet they’ll show a level of understanding that slicker projects built from a basic prompt show no evidence of. At the other peak will be projects made by people who know their background information and engage with AI as a partner: cutting what is no good and the AI mistakes, and adding the unlikely connections that fall outside the realm of previously calculated statistical probability. In this world, a student poster with one bad piece of self-selected clipart, paired with Papyrus font, tells us much more about a person’s understanding than whatever comes out when an assignment is pasted into an LLM.

Line graph titled by AI use, running from AI-free through light help, copy-paste prompt, and some editing to AI as partner. A solid blue line shows the typical mark, from low to high. It peaks at AI-free, drops to its lowest point at copy-paste prompt, then climbs back to an equal peak at AI as partner. A shaded blue band around it shows the range of marks: wide at every level except copy-paste prompt, where it narrows almost to nothing. A dashed orange line shows aesthetics. It starts low at AI-free, rises sharply to high at copy-paste prompt, and stays high through AI as partner. The two lines cross, so the most polished work sits at the lowest mark.
an updated and more coherent visual for the new assessment paradigm

I sort of think that we’ll see a human push toward bad or slightly off design, as a way of signifying that this was made by a human who cared enough at least to try. And yeah, I guess people could prompt for that… but what can’t we find loopholes for?

Matt Bernstein’s yearning for anti-AI aesthetic gives me hope

On that note, about what makes something human: over the past months I’ve gradually curated a collection of poetry with AI themes, mostly about what AI can’t be. I put them together into a collection that I shared with the teachers in our English Department, and which I will share with you here as well: Poems… about AI! Feel free to share it with whomever you want. If you are thinking about using any of them with your classes, I do have a few content warnings and notes:

  • The Aditya Narayan video piece is my favourite, but does drop an F-bomb in the first 30 seconds or so.
  • The “AI Human Centipede” written poem by Samantha Bise has some disgusting imagery and talks about eating shit (literally). It is also the closest to my fears on what AI does to creativity.
  • The Fiona Robertson piece “The Truth of the Poets” references suicide.

‘Artificial’ Intelligence

Whose artificial intelligence is it anyway?

I went a bit overboard this week, something I may not be able to sustain long term – but I had a lot of fun putting this together! The animated video and clipart are courtesy of Adobe Express.

It should be known that the astronaut is just a preset character in the animate from audio function in Adobe Express, but how serendipitous. Little guy looks a lot like me!

References

BBC News. (2016, January 26). AI pioneer Marvin Minsky dies aged 88. BBC News. https://www.bbc.com/news/technology-35409119

Biography.com Editors. (2020, July 22). Alan Turing . Biography. https://www.biography.com/scientists/alan-turing

Buolamwini, J. (2019, February 7). Artificial Intelligence has a problem with gender and racial bias. Here’s how to solve it. Time. https://time.com/5520558/artificial-intelligence-racial-gender-bias/

Chollet, F. (2019). The measure of intelligence. ArXiv. https://arxiv.org/abs/1911.01547

Donovan, P. (n.d.). Herbert Simon: Father of Artificial Intelligence. UBS Nobel Perspectives. Retrieved January 24, 2024, from https://www.ubs.com/microsites/nobel-perspectives/en/laureates/herbert-simon.html

Hao, K. (2020, December 4). We read the paper that forced Timnit Gebru out of Google. Here’s what it says. MIT Technology Review. https://technologyreview.com/2020/12/04/1013294/google-ai-ethics-research-paper-forced-out-timnit-gebru

Harris, A. (2018, November 1). Human languages vs. programming languages. Medium. https://medium.com/@anaharris/human-languages-vs-programming-languages-c89410f13252

Heilweil, R. (2020, February 18). Why algorithms can be racist and sexist. Vox. https://www.vox.com/recode/2020/2/18/21121286/algorithms-bias-discrimination-facial-recognition-transparency

McCarthy, J. (2019). What is AI? / Basic Questions. Stanford.edu. http://jmc.stanford.edu/artificial-intelligence/what-is-ai/index.html

OpenAI. (2024). ChatGPT (GPT-4 version) [Large language model]. https://chat.openai.com/chat

OpenAI. (2024). DALL-E (Version 3) [Large language model]. https://labs.openai.com

Sutton, R. S. (2020). John McCarthy’s definition of intelligence. Journal of Artificial General Intelligence, 11(2), 66–67. https://doi.org/10.2478/jagi-2020-0003

Wikipedia Contributors. (2019, October 15). John McCarthy (computer scientist). Wikipedia; Wikimedia Foundation. https://en.wikipedia.org/wiki/John_McCarthy_(computer_scientist)

Winston, P. H. (2016). Marvin L. Minsky (1927–2016). Nature, 530(7590), 282–282. https://doi.org/10.1038/530282a