Podcast Atlas
concept Updated 2026-08-24 Tags: Ai, Work, Agency, Judgment, Relationships

Non-Algorithmic Capabilities / 非算法能力

Non-algorithmic capabilities / 非算法能力 is 陈明霞’s distinction in EP275 Token 通胀时代，谁还能“不可替代”？丨“人在中流”特别策划01. Algorithmic capabilities include search, information lookup, tool operation, and other trainable procedures. Non-algorithmic capabilities include knowing who one is, what one wants, how much one wants it, what price one will pay, how to judge, how to connect with people, and how to build trust in ambiguous situations.

The concept extends Human Agency Under AI and Human Judgment Under AI. As AI gets better at repeatable, standard, and instruction-following work, the scarce layer moves toward purpose, taste, responsibility, relationship, aesthetic discrimination, decision-making, and the willingness to live through a process rather than skipping directly to an answer.

Key Claims

  • AI can raise the value of non-algorithmic capabilities by reducing the scarcity of routine cognition and standard execution.
  • Sales, interviewing, care work, job seeking, and intimate relationships all require connection and trust that cannot be reduced to correct information.
  • Process matters because reading, talking, observing, preparing, and failing are part of how judgment forms.
  • Non-algorithmic capability is not anti-technology; it decides which technology use is worth starting, continuing, slowing, or refusing.
  • Education and family systems that over-train obedience or correctness may leave people weaker at self-definition when AI can perform many knowledge tasks.
  • The concept overlaps with Subjectivity As AI Asset: personal taste, values, memory, and direction become practical assets when agents need instructions.

Connections