Initial results of the Digital Consciousness Model

TL;DR

The paper introduces the Digital Consciousness Model (DCM), a Bayesian hierarchical framework that synthesizes 13 diverse stances and 206 indicators to assess consciousness across AI and biological systems. By collecting expert credence on indicators for four target species (2024 LLMs, chickens, humans, and ELIZA) and mapping these through a three-tier structure (indicators—features—stances), the authors generate stance-specific and aggregated posterior probabilities of consciousness. Results show strong evidence against 2024 LLM consciousness but with non-decisive certainty, while humans show robust signals and chickens show mixed but often substantial signals depending on the stance; ELIZA is consistently disconfirmed. The work emphasizes principled uncertainty handling, sensitivity analyses to priors and dependencies, and the need for expanded data and stances, offering a transparent framework to track evolving evidence about AI consciousness and informing policy, ethics, and future research.

Abstract

Artificially intelligent systems have become remarkably sophisticated. They hold conversations, write essays, and seem to understand context in ways that surprise even their creators. This raises a crucial question: Are we creating systems that are conscious? The Digital Consciousness Model (DCM) is a first attempt to assess the evidence for consciousness in AI systems in a systematic, probabilistic way. It provides a shared framework for comparing different AIs and biological organisms, and for tracking how the evidence changes over time as AI develops. Instead of adopting a single theory of consciousness, it incorporates a range of leading theories and perspectives - acknowledging that experts disagree fundamentally about what consciousness is and what conditions are necessary for it. This report describes the structure and initial results of the Digital Consciousness Model. Overall, we find that the evidence is against 2024 LLMs being conscious, but the evidence against 2024 LLMs being conscious is not decisive. The evidence against LLM consciousness is much weaker than the evidence against consciousness in simpler AI systems.

Figures (20)

• Figure 1: Structure of the DCM
• Figure 2: Individual stance judgments about the posterior probability of consciousness in 2024 LLMs, starting from a prior probability of $\frac{1}{6}$ (dashed blue line). The variation in probability outcomes across model runs results from the different ways of resolving uncertainty about the presence of individual indicators.
• Figure 3: Individual stance judgments about the posterior probability of consciousness in chickens, starting from a prior probability of $\frac{1}{6}$ (dashed blue line). The variation in probability outcomes across model runs results from the different ways of resolving uncertainty about the presence of individual indicators.
• Figure 4: Individual stance judgments about the posterior probability of consciousness in humans, starting from a prior probability of $\frac{1}{6}$ (dashed blue line). The variation in probability outcomes across model runs results from the different ways of resolving uncertainty about the presence of individual indicators.
• Figure 5: Individual stance judgments about the posterior probability of consciousness in ELIZA, starting from a prior probability of $\frac{1}{6}$ (dashed blue line).