Part 1
Reading these articles made me particularly think of what we value in creative processes. For me, authorship is an essential but often overlooked element in the conception of AI-based tools. Following current societies’ business and start-up models, AI-based applications are impact-driven aiming to deliver performance and results. We can take a look at the app Lensa which became viral after the release of their “magical avatars”. Using the recent advancements in generative images with Stable Diffusion, Lensa enables a wide audience to create colorful, stylized, custom cartoon images of themselves such as the ones shown below. They offer “an all-in-one image editing app that takes your photos to the next level” with the mission to “democratize photo and video editing”.
The final results are compelling for certain. But what do they represent? What is their actual value? Monetary speaking, their value is $3.99 for 50 “unique avatars”. I was surprised by this specific claim. Literally speaking, these avatars may well be “unique” since they are generated from some very random seeds by models that are regularly updated. This ensures at least that any other generated avatar will be different pixel-wise and probably distinguishable from a human eye. Besides, on a technical level, they may seem very well and carefully made with particular styles and impressive details. But should we define “uniqueness” as the value of the distinguishability of content or as a measure of technicity? By paying more money, Lensa proposes to produce thousands, if not more, avatars. This does not sound unique to me. I believe that such a massive production lacks intent and genuine creativity. Because these generated images are not the results of a human process, they were not given any meaning or properties that make them creatively valuable.
To me, what gives creative value to a work — or actually any value — is authorship. These articles reminded me how much many AI or, more generally, technological systems keep human intents out of the loop. As explained in the readings, this can specifically lead to a loss in controllability, interpretability, explainability, or transparency. But most importantly, I think the end goal should be that the people behind the machines feel like they are creating and that the works they are making come from their souls. The feeling of authorship provides meaning to a work as it measures how much the ideas, vision, personality, or unique perspective of an individual was put into a work. As Marx believed, work and its products are an extension of people, enabling them to express themselves, realize their potential, and shape the world. Under capitalism, the products of labor are often alienated from their creators, leading to a loss of identity and purpose. In that sense, AI-based tools can be dehumanizing.
Many modern tools sacrifice the feeling of authorship for the sake of process efficiency and productivity. I myself have been working on tools to automate what seems to be tedious parts of video editing to help people create faster and give access to some of these techniques. These readings and class discussions have made me question the underlying premises of my work and the shape that tools should take to respect every aspect of the author’s intent and creative process.
Part 2
List of 10 activities or tasks that could benefit from an interactive AI approach/mindset:
- 1. Filmmaking: Semi-automating video editing with human feedback and controls
- 2. Education: Personalizing the learning experience of each student and adapting it to individual strengths and weaknesses.
- 3. Health: Using chatbots to answer questions, make appointments, and provide feedback on patient health
- 4. Legal assistance: Facilitating legal research and document reading/understanding
- 5. Video games: Creating dynamic and interactive narratives
- 6. Cooking: Discovering recipes and helping to cook with instructions
- 7. Music: Interactive music recommendations based on feedback and inputs (e.g. mood or group recommendations)
- 8. Paper Pieceable Quilts: Helping with the design of paper pieceable quilt patterns (cf. Mackenzie Leake’s work)
- 9. Sports: Tracking and analyzing athlete performance data through user-friendly interfaces
- 10. Architecture: Conceiving digital interfaces for building design and construction
— Jean-Peïc Chou
