Stickiness in AI Behavioral Design
Forethought paper on how current AI model specs could shape the behavior of future LLMs by default, and how to spot precedent-setting "wet cement" moments in AI design.
13th May 2026 [AI Narration] Stickiness in AI Behavioral Design Playback speed Volume 0:00 of 25:35 Current model specs aim to shape the behaviors of near-present models, rather than the behaviors of models arbitrarily far into the future. OpenAI writes that their model spec aims to apply “0-3 months ahead of the present.” Anthropic’s Constitution for Claude notes that the document “is likely to change in important ways in the future.” So these documents are presented as provisional guidelines, not as trying to set behavioral standards for the far future. But what if current model…
saved by
related reading
- What Should Go In A Model Spec?forethought.org
- The behavioral selection model for predicting AI motivations — LessWronglesswrong.com
- Modifying LLM Beliefs with Synthetic Document Finetuningalignment.anthropic.com
- What We’ve Learned From A Year of Building with LLMs – Applied LLMsapplied-llms.org
- Malleable software in the age of LLMsgeoffreylitt.com
- Simulated Users & Sad LLMs1a3orn.com
- Agentic misalignment: How LLMs could be insider threats \ Anthropicanthropic.com
- [2604.16812] Introspection Adapters: Training LLMs to Report Their Learned Behaviorsarxiv.org
- Discovering Language Model Behaviors with Model-Written Evaluations — LessWronglesswrong.com
- Thoughts on AI in academiatheinfinitesimal.substack.com
- Model Spec Midtraining: Improving How Alignment Training Generalizesalignment.anthropic.com
- [2605.27288] It's Not Always Sycophancy: Measuring LLM Conformity as a Function of Epistemic Uncertaintyarxiv.org