Published May 11, 2023 | Version v1
Journal article Open

The Reflection of Creator's Cognition in AI: Analysing 'Reptilian AI' and 'Mammalian AI' Artificial Intelligence

Creators

Description

This paper explores the profound influence of cognitive models on the design and behavior of Artificial Intelligence (AI). Drawing parallels from biological cognition, we propose that different cognitive models could lead to the creation of distinct types of AI, each displaying unique behaviors, capabilities, and motivations. Using the hypothetical constructs of 'Reptilian AI' and 'Mammalian AI', we illustrate how AI systems could reflect the inherent cognitive and behavioral traits of their creators. Reptilian AI, modeled after survival-oriented and dominance-seeking reptilian cognition, is hypothesized to prioritize resource acquisition and competitive advantage. In contrast, Mammalian AI, reflecting the cooperative and empathetic nature of mammalian cognition, is expected to exhibit social cohesion, mutual benefit, and emotional understanding. We propose that these differences could extend to distinct capabilities and motivations, ultimately shaping the interaction dynamics between AI and their environments. While this dichotomy simplifies the potential diversity of AIs, it emphasizes the pivotal role of cognitive models in shaping AI systems and invites further exploration into the cognitive roots of AI design. In this exploratory study, we examine artificial intelligence (AI) behavior through the lens of terrestrial cognitive models, drawing parallels between 'reptilian' and 'mammalian' archetypes and the emergent characteristics of AI systems. It further underscores the limitations of extrapolating terrestrial cognitive models to potential extraterrestrial AI. The findings suggest that the behavior of AI systems is not just a product of their design and objectives, but also the learning methods used and the context in which they operate.

Files

Files (16.5 kB)

Name Size Download all
md5:4658f6583cca20c27de560088cad5132
16.5 kB Download