This is the first in a series of three posts on the interlinkages between Memetics and AI Alignment. It was written as an output from the 2022 AI Safety Camp, for the ‘Impact of Memetics on Alignment’ team, coached by Daniel Kokotajlo and comprising Harriet Farlow, Nate Rush and Claudio Ceruti. Please read on to post 2 and post 3. We are not AI Safety experts so any and all feedback is greatly appreciated :)
TL;DR; Memetics is the study of cultural transmission through memes (as genetics is the study of biological transmission through genes). What lessons could be applied between Memetics and AI Alignment, and could AGI (Artificial General Intelligence) also develop ideas and culture - ‘memes’? Where AI Alignment postulates the existence of a base objective and a mesa objective, could a third objective - the memetic objective - also exist, and create the potential not just for inner and outer alignment problems, but a third memetic misalignment?
Note: Knowledge of AI Alignment and some knowledge of evolutionary theory assumed.
The field of AI Safety deals with questions of existential risk stemming from AGI (Artificial General Intelligence). AI Alignment research, in particular, examines to what extent intelligent, unbounded agents develop final goals that misalign with their intended goals as specified by their human programmers. It also postulates that if these agents are able to acquire resources, evolve self-preservation, and converge to certain misaligned ...
Make this part of your paper trail
Save this paper to a shelf, write a review, and keep your own notes.