AI Aesthetics

1) Selecting content from larger collections: search, discovery, curation, recommendations and filtering.

1) Selecting content from larger collections: search, discovery, curation, recommendations and filtering. (Asrar, 2016).
2) Targeting content (e.g., one-to-one marketing, behavioral targeting, and market segmentation).
3) Assistance in creation/editing of new content. (If we are to think of AI as intelligent in the biological sense, we can cal this “participation” in content creation.)
4) Fully autonomous creation (e.g., AI composing music tracks in a particular style, writing business and sports news articles (Kafka, 2016), creating visualizations from given datasets, designing websites, generating email responses, etc.)

5. Huawei ran a photo contest where submitted photos were judged by AI: “Trained using 4,000,000 images taken by professional photographers and picture editors, the AI will then give each photo a personalized AI score based on parameters such as focus, jitter, deflection, color, and composition” (Hillen, 2018)

Harold Cohen. Amsterdam Suite C. Lithograph from a drawing generated by a computer program written by Cohen. 1977

AI now plays an equally important role in our cultural lives and behaviors, increasingly automating the processes of aesthetic creation and aesthetic choices.

Let’s look at computer science publications that quantitatively analyze Instagram as an example. To locate only quantitative papers, I usually add the word “dataset” to the name of the social network I want to find research about, and then search for this combination of words on Google Scholar. The search for “Instagram dataset” conducted on July 15, 2017 returned 9,210 journal articles and conference papers.

This is one of the key issues surrounding cultural uses of AI. Are the results of machine learning interpretable, or are they only a black box which is efficient in production but useless for human understanding of the domain? Will the expanding use of machine learning to create new cultural objects make explicit the patterns in many existing cultural fields that we may not be aware of? And if it does, will it be in a form that will be understandable for people without degrees in computer science? Will the companies deploying machine learning to generate movies, ads, designs, images, music, urban designs, etc. expose what their systems have learned?

Today AI gives us the option to automate our aesthetic choices (via recommendation engines), assists in certain areas of aesthetic production such as consumer photography, and automates other cultural experiences (for example, automatically selecting ads we see online). But in the future, it will play a larger part in professional cultural production. Its use of helping to design fashion items, logos, music, TV commercials, and works in other areas of culture is already growing. But currently, human experts usually make the final decisions or do actual production based on ideas and media generated by AI.
The well-known example of Game of Thrones (American fantasy television drama series that premiered in 2011) is a case in point. The computer suggested plot ideas, but the actual writing and the show’s development was done by humans. We can only talk about fully AI-driven culture where AI will be allowed to create the finished design and media from beginning to end. In this future, humans will not be deciding if these products should be shown to audiences; they will just trust that AI systems know best — the way AI is already fully entrusted to choose when and where to show particular ads, as well as who should see them.

the role played by cultural AI is probably not the most significant yet — but it is likely to grow in the future for at least two reasons. First, billions of people who still don’t have access to internet and smartphones will get this access and start using the same AI-driven recommendation engines, automated aesthetic editing of captured media, selfie-beautifying apps, and so on. Second, the automation of aesthetic decisions we have seen so far is still at an early stage, with many more things to come. For example, right now people take photos themselves, cameras apply some aesthetic adjustments at the time of capture, and then a person may use editing software to do further adjustments. But it is easy to imagine a future scenario where cameras will themselves choose what and when to capture to give us the most satisfying photos that fit certain concepts and aesthetic ideals. In fact, the Google Clips video camera released in January 2018 is already doing this. The camera is fully AI-based. It uses computer vision to recognize people and pets and certain emotional expressions, and was trained by professional photographers to make “good” videos with proper composition, interesting actions, etc. (Lovejoy, 2018).

Of course, a number of other trends also influence aesthetic diversity in contemporary culture besides computational technologies. The rise of the world wide web and social networks, growth of international travel, globalization of consumer economies and advertising, zero cost telecommunication, growth of foreign student enrollment, growth of remote work, and the rise of Japan, followed by Korea and then China, as exporters of cultural products and images are just some examples among many other developments all playing a role. On the one hand, they are making the world into a single global village — or if you like, a single cultural marketplace, where certain im-ages, ideas, values, narratives, products, and styles are marketed to everybody and available everywhere, and this may decrease diversity. On the other hand, the same trends may also be increasing diversity because local cultural DNAs become available globally.