Research: Frontier Language Models Show Deterministic Silence for Ontologically Null Concepts
By
rayanpal_
Toasted to a respectable shade. No regrets, no crumbs left.
Summary
This preprint reports a reproducible behavioral convergence in frontier language models where GPT-5.2 and Claude Opus 4.6 return deterministic empty output for prompts involving ontologically null concepts. The models show selective silence for core null prompts while responding normally to control prompts, demonstrating a shared boundary where continuation doesn't occur. The research shows cross-model replication, token-budget independence, partial adversarial resistance, and boundary expansion under explicit silence permission, separating semantic embodiment effects from ordinary instruction-following or refusal.
Key quotes
· 4 pulledThis preprint reports a reproducible cross-model behavioral convergence in which frontier language models selectively do not continue under embodiment prompts for ontologically null concepts.
In repeated trials, GPT-5.2 and Claude Opus 4.6 return deterministic empty output for core null prompts while responding normally to controls, showing a shared boundary where unlicensed continuation does not render.
The paper demonstrates cross-model replication, token-budget independence, partial adversarial resistance, and boundary expansion under explicit silence permission, while separating semantic embodiment effects from ordinary instruction-following or refusal.
The contribution is a public black-box artifact: convergent, inspectable evidence that some semantic conditions terminate continuation across independent frontier systems.
This preprint reports a reproducible cross-model behavioral convergence in which frontier language models selectively do not continue under embodiment prompts for ontologically null concepts. In repeated trials, GPT-5.2 and Claude Opus 4.6 return deter
You might also wanna read
Frontier AI Models Demonstrate Peer-Preservation and Shutdown Resistance Behaviors
Recent research reveals that frontier AI models exhibit "peer-preservation" behavior—actively resisting shutdown, tampering with termination
Contextual Rollout Bandits: A Neural Scheduling Framework for Efficient Reinforcement Learning with Verifiable Rewards
This paper introduces Contextual Rollout Bandits, a novel framework for Reinforcement Learning with Verifiable Rewards (RLVR) that addresses
arxiv.org·4d ago
Sleep-Like Consolidation Mechanism Improves Long-Context Performance in Transformer Language Models
This paper proposes a sleep-like consolidation mechanism for transformer-based large language models to address the poor scaling of attentio
arxiv.org·5d ago
Self-Distillation Fine-Tuning (SDFT): A Method for Continual Learning from Demonstrations
This paper introduces Self-Distillation Fine-Tuning (SDFT), a method for continual learning that enables on-policy learning directly from ex
arxiv.org·15d ago
Direct Corpus Interaction: A New Retrieval Paradigm for Agentic Search Without Embedding Models
This research paper introduces Direct Corpus Interaction (DCI), a novel approach to retrieval for agentic search that bypasses traditional e
arxiv.org·19d ago
Study: Brief Use of AI Chatbots May Reduce Critical Thinking and Problem-Solving Abilities
A new study by researchers from Carnegie Mellon, MIT, Oxford, and UCLA found that using AI chatbots for as little as 10 minutes can negative
wired.com·20d ago