Embedded Files
Contact Information
Contact Information
Researcher: Spencer Anderson
Class of 2023
Survey Link
Survey Link
Survey Title: Emotion in AI and Human Art (Survey is Closed)
Artwork viewable at bottom of page
6 Primary Emotions Experienced by Humans
6 Primary Emotions Experienced by Humans
Anger: A negative feeling that is typically associated with hostile thoughts or physiological arousal. Develops in response to unwanted actions or scenarios around you.
Sadness: A negative feeling of downness or unhappiness in response to grief, discouragement, or disappointment.
Joy: A positive feeling of great pleasure, contentment, and happiness. Often associated with the feeling of love.
Fear: An unpleasant feeling that is triggered in response to danger or a perceived threat, usually causes anxiety or worry as well as nervousness.
Disgust: A negative feeling triggered in response to revulsion, disapproval, or rejection of something. The feeling is unpleasant and distasteful.
Surprise: A feeling of mild astonishment or shock.
Problem Statement and Question
Problem Statement and Question
There is a problem in the art industry regarding the rise of AI. Despite the fact that AI is not advanced enough to experience its own emotions or request payment from customers, concerns from artists are occurring because they believe that AI can capture emotion just the same as them and do it much faster without pay. This issue has negatively impacted the art field and children who have wished to enter the future art field. A possible cause of this problem are the new AI engines that are being created that allow a person to tell the AI to create a drawing based off their prompt for absolutely free. Perhaps a study regarding human perception using a survey of AI drawings as compared to human drawings with the same prompts can help us to learn whether or not AI can capture the same authenticity as a human and eventually replace artists all together. So this begs the question, "What are the perceptual differences (if any) of Generation Z towards AI-generated artwork in terms of its ability to convey emotions and intentions, as compared to human-created artwork?"
Gap and Importance
Gap and Importance
The gap in the information is that AI emotion has never been tested in an artistic sense, it has only ever been tested from a psychological standpoint. In that sense it has been observed that AI cannot feel emotion nor experience life, it can merely understand the dictionary definitions of experiences. Therefore I would like to fill in the gap by testing its ability to create emotion through pieces of artwork to see if it has different results as compared to the negative test results in psychology.
The value of my project is that it will hopefully help struggling artists learn that their art cannot be replaced with an AI and they won’t lose their jobs, due to the fact that an AI cannot create true emotion because it has never had any life experience to gain knowledge of that emotion. I also hope that it helps to encourage more children to go into the art field, rather than it being seen as very implausible because AI dominates it. And If I come to find out that AI is better at conveying emotion then I can use my research and apply what I have learned about emotion to write down some valuable lessons all artists could use, and we could potentially investigate the laws around AI art in order to not allow them to completely shut humans out of the industry entirely.
Hypothesis and Biases
Assumptions and Biases are present in my research even before beginning to work on the artwork. I have the bias as an artist myself that the AI artwork is not ethical nor as high quality as a traditional piece of artwork. With those biases present, I am also making the assumptions that anybody who takes the survey will be able to understand based off of my description and instruction: what is meant by the emotion and intention in the piece rather than looking at quality and favourites. I am assuming that AI can replicate emotions, that art has viewable emotion, and that Generation Z will yield the most accurate results considering that they are the next group to take over the working industry.
My hypothesis of the experiment is that the AI artwork will have a higher level of technical skill in the artwork based off of the ability that it has an endless expanse of knowledge and can instantly learn any new skills. While the AI may have better technical skills, those do not apply to the experiment and the results of the experiment is that the human artwork will be able to capture emotion and intention of a piece better than an AI because it cannot experience any emotions itself and merely understands the definition for any given feeling.
Method of Research
Using a correlational hermeneutic comparative survey, to gather quantitative data based off of a likert scale. The survey will have them look at an artwork and answer 2 questions per artwork (6-10 pieces). There will also be consent questions at the beginning as well as questions to verify that they match the correct cohort that I wish to measure. Questions for the artwork will ask the person what primary emotion they feel when they see the piece, and then ask them to rate it on a likert scale for how strong the emotion is, 10 being the highest and 1 being the lowest. I will use emails to send out the survey to principals in the sumner school district and asking them if they would be willing share it with their students. or teachers if applicable. The data on the surveys will be accessible for them to view after being taken and will be anonymously viewed on my end in order for me to compile the results into a chart to use for my conclusion of data. One compiled I will use percentages to show accuracy of the emotion shown in a piece and then the percentage of the strength of the emotion as well and compare the data between the AI artworks and the human made.
Results
After the survey concluded, we could see a strong correlation between emotion being present within the art. The averages were taken for each emotion presented, and an overall average for all the combined data. Using the overall data graph , the human accuracy was 7.2% higher, and the perceived strength was 12% higher than the AI, and these two data sets prove that my hypothesis was correct in that the test would tell us that humans are better at achieving accuracy and strength. While this may be true in general as an average, I broke down the data to closely examine each emotion, human art was found to be more accurate and perceived to be stronger than AI in all categories except that it was found to be 8% less accurate for both fear and anger, as well as the strength 10.2% weaker for anger. This data proves that AI can understand emotion and has enough emotional intelligence to capture and share its intention with a viewer stronger than a human. However, in general, it is at a slightly weaker level than a human, but in some specific circumstances, it can perform better than a human (Data graphs available to view at bottom of the page).
Conclusion and Implications
Gen Z’s perception of AI made artwork shows that it is almost at the same emotional understanding and intelligence as a Human in terms of representation in art. Open the door for many new artists. This research focused on filling the gap of whether or not AI could replicate and capture emotion with artwork and on trying to help artists learn not to fear AI. This new research has proven that AI can replicate emotion and has some understanding of emotional intelligence within an artwork. It also shows that artists have not been replaced and can continue to have hope and improve. I hope that this paper can be used to further future research and fill more of the gap by being used as a template for how to run more studies involving using different artists, AI programs, and prompts to find sufficient data and determine whether or not ai is above or below the average artist's emotional intelligence and intention.
Sources
Baillie, M. (1960). Early Picasso. New Left Review, 6, 58.
Barrett, L. F. (2017). How Emotions are Made: The Secret Life of the Brain. Pan Macmillan.
Bostrom, N. (2014). Get ready for the dawn of superintelligence. New Scientist, 223(2976), 26-27.
Burrell, J. (2016). How the machine ‘thinks’: Understanding opacity in machine learning algorithms. Big data & society, 3(1), 2053951715622512.
Cabanac, M. (2002). What is emotion?. Behavioral Processes , 60(2), 69-83.
Cellan-Jones, R., & Hawking, S. (2014, December 2). AI Could Spell the End of the Human Race. Stephen Hawking warns artificial intelligence could end mankind. other, BBC. Retrieved March 21, 2023 from https://www.bbc.com/news/technology-30290540.
Crevier, D. (1993). The tumultuous history of the search for Artificial Intelligence. Basic Books.
Dissanayake, E. (1990). What is art for?. University of Washington Press.
Hjort, M., & Laver, S. (Eds.). (1997). Emotion and the Arts. Oxford University Press on Demand.
Matravers, D. (2011). Art and emotion. Clarendon Press.
Pons, F., Harris, P. L., & Doudin, P. A. (2002). Teaching emotion understanding. European Journal of Psychology of Education, 17(3), 293-304.
Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022). Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125.
Ravin, J. G., & Perkins, J. (2004). Representations of blindness in Picasso's blue period. Archives of ophthalmology, 122(4), 636-639.
Spitznagel, E. (2020, January 25). Generation Z is bigger than millennials - and they're out to change the world. New York Post. Retrieved March 21, 2023, from https://nypost.com/2020/01/25/generation-z-is-bigger-than-millennials-and-theyre-out-to-change-the-world/
The state of AI in 9 charts. Stanford HAI. (n.d.). Retrieved April 16, 2023, from https://hai.stanford.edu/news/state-ai-9-charts
Tolstoy, L. (2021). What is Art? . Standard Ebooks.
Z-score to confidence calculator: A/B testing confidence level. Evolytics. (2020, June 24). Retrieved March 21, 2023, from https://evolytics.com/resources/calculators/abtesting-zscore-to-confidence/
Data Charts
Artwork Created for Experiment
AI Made (Disgust and Sadness)
AI Made (Joy and Surprise)
AI Made (Fear and Anger)
Human Made (Joy and Surprise)
iHuman Made (Fear and Anger)
Human Made (Disgust and Sadness)
Report abuse