AI and the Anthropocene—LLMs as Democratic Knowledge Producers

Authors:
Open Access
Publisher website
Google Scholar
Abstract
This chapter examines artificial intelligence (AI), and in particular large language models (LLMs) as knowledge producers in the Anthropocene. Views on AI are divided: some argue that it offers vast analytical capacity to address planetary challenges, while others warn of dystopian futures and patterns of ecological harm. Yet, the quality of AI inference is often overlooked, leaving its potential contribution to the Anthropocene uncertain in terms of democratic viability and sustainability concerns. To investigate this, the chapter compares findings from human-guided natural language analysis of 158 interviews on democratic innovations with those produced through a structured LLM protocol designed to replicate the same tasks. The results demonstrate that LLMs can generate intelligible and actionable outputs, although these vary in reasoning and abstraction, and warrant further scrutiny. The chapter suggests that LLMs may emerge as autonomous knowledge creators, establishing a non-human form of agency in the Anthropocene. It concludes by arguing that situating this agency within human guidance requires a critical evaluation of LLM inference.

This publication has 24 references indexed in Scilit: