I'm developing artistic research using Artificial Intelligence Software. The current outcome is an artwork with science fiction storytelling, distinct from a typical science fiction narration, and two symbolic pictures about life and death. It appears in the form of a printed book, file and video. The title comes partially from Philip K. Dick's biography, and the theme is related to Karel Čapek, two science fiction classics authors. I am working with AI-powered chatbots. The problem for me is that this kind of software interacts by imitating human language. This attitude of imitation impacted Western art for a long time. Then, it was challenging for me to work with a language-imitating tool. I asked the chatbot to create a new language to break the imitation effect. The theme is at the intersection of imitation and sympathy through technologies and social implications.
The first artistic research I developed using Artificial Intelligence Software is "Creating a New Language: Impossible. (et vous êtes)".
According to the Joint Research Centre's report "Defining Artificial Intelligence 2.0.", we don't have a standard definition of Artificial Intelligence[note 1]. Indeed, as the report stresses, human intelligence is hard to define, and sometimes to practically develop AI, there is a need to oversimplify the broad concept of intelligence[note 2].
From a engineer's point of view, it is about how machines can show intelligent behaviour, as argued by Wolfgang Ertel[note 3].
The current outcome is an artwork with science fiction storytelling, distinct from a typical science fiction narration, and two symbolic pictures about life and death. It appears in the form of a printed book, file and video.
Accordingly, I chose the title keeping in mind a Philip K. Dick biography, namely "Je suis vivant et vous êtes morts" ("I Am Alive and You are Dead") by Emmanuel Carrère[note 4].
Dick is a singular science fiction author[note 5]. Likewise, the biography title suggests a thought-provoking idea that can fit the theme of this work, playing with hypotheses around AI status.
The AI-powered software I used is called ChatGPT, developed by OpenAI[note 6]. I chose a chatbot to have a fast-access interface to AI software.
According to Masood Ghayoomi, the word "chatbot" comes from "chatterbot", coined by Michael Loren Mauldin[note 7], from the term "verbot" as "talking robot"[note 8].
For the present research, it is fascinating to note that the term "robot" was coined by an artist, Joseph Čapek[note 9], and spread in scient-fiction[note 10] by Karel Čapek's novel in 1921[note 11] with a social-related meaning[note 12].
The problem for me is that this kind of software interacts by imitating human language.
This attitude of imitation (namely, mimesis) impacted Western art for a long time[note 13]. Then, it was challenging for me to work with a language-imitation tool.
Yet, indeed, that's the point: you can interact fast due to the use of language as a means of interaction.
However, there is another side of the coin. The imitation ability of the chatbot is so good that it can drive sympathy (namely, empathy), which could have unexpected social effects, according to Sophie Bushwick[note 14].
So "Creating a New Language: Impossible. (et vous êtes)" is at the intersection of imitation and sympathy.
To briefly describe the process of the work, I use sentences like "I asked the chatbot" or "the chatbot answered". It is this use of the language which could drive sympathy for the software, as argued by Douglas R. Hofstadter[note 15]. However, according to the present technology[note 16], I don't consider the chatbot a subject. Yet I consider these sentences appropriate since the chatbot is a conversational tool.
So, to interrupt this imitation effect, I first asked the chatbot to talk in binary code.
Then I changed my mind. I asked the chatbot to create a brand-new language. The chatbot has not this ability. Indeed, it automatically named our chat "Creating a New Language: Impossible." Consequently, I asked: "Generate a sample of artificial language based on existing patterns and structures found in English". It was able to do that. It started to shape grammar.
It decided to call this new language "Sintalish". It said that those name "was intended to be easy to pronounce and remember for English speakers, while also sounding distinct and unique".
Then, I asked the chatbot to write a book about an "AI that generates an artificial language" written in Sintalish. The process was sometimes confusing when asking to translate from Sintalish to English. Then it introduces a character out of the blue called "Pafyz" and the name of another language called "Filak". By the way, in the end, it wrote six chapters.
The artwork presents the Sintalish text. Then it is possible to have access to the English translation online.
I put the English translation apart to clarify my artistic intention. My purpose is not to create a narrative book written by an AI. This kind of aim involves philosophical reflections about the author's role in an artwork creation and social issues about the condition of cultural workers. These are relevant themes but not the focus of my intention. However, since the outcome shows an intersection between the artwork and science-fiction narration, I found it pertinent to provide the English translation.
Meanwhile, I asked the chatbot to generate shapes by writing code for vector images. I used this kind of code-generated shape in a previous work called "Symbiotic violence". First, I tried asking about objects' shapes as an apple or a bird. Then I tried "concept shapes". It was able to produce the shape of life. It did it like a symbol, along with a detailed explanation. It refused to do the same with the concept of death. However, I called "death" "the end of life", and it agreed to create a shape of it.
Along with the text narration, the life and death shapes, and some quotes from our chat, I made the artwork "Creating a New Language: Impossible. (et vous êtes)". It is a physical object and an open-access artwork that you can download.
I checked the content generated by ChatGPT with plagiarism detection software. I checked the images generated by the SVG code provided by ChatGPT with Google Lens for visual matches. Although there are simple shapes that can have similitudes with other simple shapes, I cannot find a unique match. I provided the input, and according to OpenAI Terms of Use, I retain the rights to use the output[note 17].