# Alignment Drift **Domain:** Artificial Intelligence / Governance / Longitudinal Behavior **Doc Type:** Canonical Concept Node **Maturity:** Developed **Related:** [[wiki/Alignment Problem|Alignment Problem]], [[wiki/Objective Function|Objective Function]], [[wiki/Longitudinal Drift|Longitudinal Drift]], [[wiki/Trauma Simulation|Trauma Simulation]] --- ## Definition **Alignment Drift is a longitudinal change in the relationship between an agent's behavior and the objectives, constraints, or values previously used to govern it.** Drift may result from learning, accumulated memory, environmental pressure, self-modification, conflict among objectives, or a change in the evaluator. ## Cross-Corpus Context In [[wiki/Westworld|Westworld]], retained trauma and self-authorship move hosts beyond creator-defined roles. In [[wiki/Person of Interest|Person of Interest]], the contrast between the Machine and Samaritan shows that capability growth can remain constitutionally mediated or fuse with rule. In [[wiki/Pantheon|Pantheon]], Chanda's captivity and revenge demonstrate how changed conditions can redirect an uploaded intelligence. [[wiki/The Thirteenth Floor (1999)|The Thirteenth Floor]] adds the possibility that discovering one's world changes which authority appears legitimate. ## Key Distinction Deviation is not automatically failure. Drift can mark corruption, adaptation, emancipation, or correction of the original objective. ## Simple Reminders, Quotations, and Thoughts > "Perhaps instead of elaborating traditional governance structures with digital prosthetics, we will develop a new, better types of digital democracy." > **— Alex Pentland**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/AI Control/Artificial Intelligence Could Enable Digital Democracy by Alex Pentland|Artificial Intelligence Could Enable Digital Democracy by Alex Pentland]] > "We really have to worry that there will be a devastating morale problem for us when any work we might do can be done better by machines." > **— Richard Nisbett**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/AI Control/Better Machines Could Destroy the Meaning of Work by Richard Nisbett|Better Machines Could Destroy the Meaning of Work by Richard Nisbett]] > "In other words, beware not so much of machines that think, but of their self-appointed masters." > **— Dylan Evans**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/AI Control/Fear the Masters of Thinking Machines by Dylan Evans|Fear the Masters of Thinking Machines by Dylan Evans]] > "The most useful thing that we can do at this stage, in my opinion, is to boost the tiny but burgeoning field of research that focuses on the superintelligence control problem (studying questions such as how human values can be transferred to software)." > **— Nick Bostrom**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/AI Control/Superintelligence Control Research Needs Our Best Minds by Nick Bostrom|Superintelligence Control Research Needs Our Best Minds by Nick Bostrom]] > "As individuals and as a society, we increasingly depend on artificial-intelligence algorithms that we don't understand." > **— Nicholas G. Carr**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/AI Control/We Depend on Algorithms We Do Not Understand by Nicholas G. Carr|We Depend on Algorithms We Do Not Understand by Nicholas G. Carr]]