concept Updated 2026-07-09 Tags: Ai, Agency, Identity, Context

Subjectivity As AI Asset

Subjectivity as AI asset is Tristan’s claim in 135. 和自然选择创始人Tristan聊,Elys、赛博分身、灵魂、Context的获取与流动和AI社交网络 that the scarce AI-era layer is not generic intelligence, but a person’s own “who I am.” In his formulation, agents need to know what the user wants, values, notices, rejects, finds beautiful, has created, and is trying to become.

The concept extends Human Agency Under AI. Agency asks what should be delegated and why; subjectivity as an asset asks how a person’s taste, values, works, memory, and self-understanding can be made explicit enough for agents and social products to preserve or act from it.

E45 孟岩对话李继刚:人何以自处 adds Li Jigang / 李继刚’s “taste as weights” version. Taste is not mystical in the episode; it is the result of long exposure, selection, and repeated training that shapes the person’s internal weights. That makes Feed Curation, Reading As Frame Training, and Wet-State Human Agency part of subjectivity formation, not only preference data for a model.

Key Claims

  • If model intelligence becomes more widely available, differentiated outputs may depend more on personal context, taste, prior work, and value choices.
  • Strong creators and IP holders may gain from AI if their subjectivity is well represented and legally/socially recognized.
  • User context is not only a data moat; it can be a representation of agency that users may want to own, port, and protect.
  • Elys treats subjectivity as something that can be stored, updated, and circulated through Cyber Avatars, but this also raises privacy and authenticity concerns.
  • Taste can be understood as trained weights: repeated exposure to high-quality inputs changes what the person can notice, choose, and ask agents to preserve.

Connections