Why a Human Being Can’t Help But Change

2026-05-06 · Source: Artificial Intelligence on Medium · Field: Science & Research — Social Sciences & Behavioral Studies, Artificial Intelligence & Machine Learning · Depth: Intermediate, medium

Summary

The article posits that human identity is not static but a constantly updating "vector" influenced by various inputs, drawing parallels to machine learning concepts. It illustrates this through the character Alex, whose internal "vector" shifts based on interactions with his boss (high attention weight) and consumption of war movies (lower attention weight). The author explains that humans have "gates" that determine how much new information is integrated, often opening when tired or less defended, making late-night consumption particularly impactful. Confirmation bias is reframed as an attention weighting problem, where agreeable information receives high weight and threatening information receives low weight. The piece connects these human update mechanisms to AI models like Graph Neural Networks, Transformers, and Relational Memory Cores, which also update based on weighted influences from their "neighbors" or contextual elements.

Key takeaway

For professionals seeking personal or professional growth, recognize that your identity is not fixed but constantly evolving. Intentionally direct your attention to positive influences, whether through mentors, educational content, or inspiring narratives. Your "attention weight" determines what shapes you, so consciously choose your inputs, especially when your mental "gates" are open, to foster desired changes and mitigate unwanted ones.

Key insights

Human identity is a dynamic vector, continuously updated by weighted attention to people, stories, and experiences.

Principles

• Attention is an update mechanism.
• We become what we repeatedly attend to.
• Confirmation bias is a weighting problem.

In practice

• Observe who you admire and imitate.
• Identify what stories are shaping your vector.
• Be mindful of what you consume before sleep.

Topics

• Dynamic Identity
• Attention Weighting
• Confirmation Bias
• Machine Learning Analogies
• Personal Permeability

Best for: AI Scientist, Research Scientist

Related on AIssential

• In the Weights is your new AI-centric vanity search — "In the Weights" quantifies an individual's recall prominence within AI model parameters, reflecting a shift from traditional web search.
• The new year — Personal growth is a continuous process, guided by flexible "directions" rather than rigid goals.
• Steering at the Source: Style Modulation Heads for Robust Persona Control — Targeting specific "Style Modulation Heads" in LLMs enables robust persona control with minimal coherency degradation.
• Re-Centering Humans in LLM Personalization — LLM personalization evaluations relying on synthetic data significantly misrepresent real-world human experience and model limitations.
• The Strange Anxiety of People Who Did Everything Right — AI challenges high achievers' identity by decoupling capability from guaranteed value, shaking the foundation of their meritocratic self-understanding.
• Trait, Not State: The Durability of Reading Identity in Social Highlighting — A reader's highlighting patterns form a durable "trait" that consistently predicts future reading choices over long periods.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence on Medium.