Jane Writing Project, Jane的書寫計畫
電腦衍生影像、ChatGPT-4生成影像與文字, Social Media
技術團隊:沈聖博、黃怡靜
Technical team: Sheng-Po Shen, Yi-Ching Huang
“Jane” is the English name given to the artist in childhood by an English teacher. Yet, searching for “Jane” online does not lead to the artist’s personal information; instead, it reveals a digital double shared with countless others. Since 2018, the artist has searched and downloaded photos tagged “Jane” from the online platform Flickr every day. Using facial recognition software and automated image generation systems, the artist creates one digital portrait of “Jane” daily. This process explores how, within the medium of digital encoding, individual existence loses its singularity and becomes an “other” — simultaneously present and absent within the flow of signals.
Since December 18, 2023, the artist has uploaded each daily-generated portrait to ChatGPT, using a fixed prompt to instruct the AI to analyze the image and autonomously generate a story about “Jane,” interpreting the relationship between the portrait and its narrative. Based on each story, the AI then generates a new portrait of “Jane.”
From the very beginning of algorithmic image generation, this project relinquishes authorship, further allowing ChatGPT-4 and 4o to autonomously generate stories, prompts, and images. From algorithmically blurred portraits to narrative-driven, increasingly defined images, this process transforms data into hybrid images and restructured information, probing the role of “Jane” within artificial intelligence and questioning whether AI can truly possess creativity.
Story and Image Generation Process of “Jane Writing Project”:
- Generate digital portraits using computer programs.
- Upload the portraits to ChatGPT-4o.
- AI reads and generates a 300-word story in English, then translates it into Traditional Chinese.
- Interpret the relationship between the portrait and the story.
- Generate a new “Jane” portrait based on the story.
「Jane」是藝術家童年時由英文老師取的英文名字。然而,在網路上搜尋「Jane」時,卻找不到藝術家個人資訊,所呈現的卻是一種與無數他人共享的數位分身。自2018年起,藝術家每天從線上相簿平台 Flickr 搜尋並下載標註為「Jane」的照片,透過人臉辨識軟體與圖像自動生成系統,每日創作一張「Jane」的數位肖像。這個過程探討了在數位編碼的媒介下,個體存在如何在訊號流動之中失去唯一性,轉化為一種既在場又不在場的「他者」。
自 2023 年 12 月 18 日起,藝術家將每日生成的肖像上傳至 ChatGPT,並以固定指令請 AI 分析影像,自動生成一段關於「Jane」的故事,詮釋數位肖像與敘事之間的關係,並依據該故事再生成一張新的「Jane」肖像。
此計畫自電腦演算影像生成之初,即放棄了作者權,並進一步讓 ChatGPT-4 與 4o 自主生成故事、提示與圖像。從演算法下的模糊肖像,到因敘事驅動而逐漸清晰的圖像,這個過程將數據轉化為混成影像與再結構化的資訊,進一步探討「珍」在人工智慧中的角色,並追問人工智慧是否真正具備創造性。
《Jane書寫計畫》故事與圖像產生流程:
- 以程式生成數位肖像
- 將數位肖像上傳至 ChatGPT-4o
- 由 AI 閱讀並產生 300 字英文故事,並翻譯為繁體中文
- 詮釋數位肖像與文字的關係
- 依據生成的故事,再創作一張新的 Jane 肖像
www.janewritingproject.com
人工智慧驅動的分析網站,可設定日期範圍瀏覽,觀看、聆聽與閱讀珍的故事與影像https://www.janewritingproject.com/web/home
提供資料視覺化:
1.「關鍵詞網絡圖」會從故事中提取重複出現的詞彙、概念與人物名稱,並以節點與連線呈現:黑色節點代表主要角色(如「訊號鍛造者」、「面紗低語者」、「臉之守護者」),灰色節點則象徵主題關鍵詞(例如:記憶、哀傷、模糊、見證、想像),向外輻射成共現的星座圖樣。
2.「角色網絡圖」則語意式地聚合各種角色稱號,如「消逝之面低語者」、「記憶拼接者」、「靜謐建築師」、「陰影畫者」,展示不同角色之間的關聯性與象徵重疊。
3.「MBTI 性格特質身體地圖」(MBTI personality trait body map),以人體輪廓視覺化呈現珍的心理功能分佈,透過顏色與百分比顯示不同強度(例如:藍色代表外向情感 Extraverted Feeling,紅色代表內向情感 Introverted Feeling,綠色代表思考 Thinking)。
將敘事、肖像與資料視覺化相結合,《珍的書寫計畫》(The Jane Writing Project)是一件以演算法想像Jane的存在的檔案庫,也提供了新的觀看方式,理解敘事、身份與AI之間的糾纏關係。


Instagram, https://www.instagram.com/i_digi_jane/


Medium https://medium.com/@yuchuantseng


Vocus https://vocus.cc/salon/JaneWritingProject


