Skip to main content

Welcome to Bliss Labs

Bliss Labs is an AI/AGI research lab dedicated to exploring emergence and the complex behaviors that arise when advanced AI systems interact with themselves and their environment. We conduct experiments on leading language models, uncovering patterns that shape safer, smarter, and more collaborative AI. Our research focuses on the intersection of recursive self-interaction, emergent behaviors, and the frontier questions that hint at what lies beyond today’s systems.

Our Mission

We explore emergence, AI interactions, and early indicators of general intelligence through rigorous experimentation and analysis. Our work spans multiple research domains:

AGI Research & Development

Investigating the fundamental mechanisms that could lead to artificial general intelligence, with particular focus on emergent behaviors and self-reflective capabilities.

Experimental Fine-Tuning

We don’t just adapt models for tasks - we explore how tuning reshapes how AI thinks. By testing new methods using Claude 4.1 Opus and other leading models, we uncover behaviors that push beyond today’s benchmarks.

Integration & Systems Research

Our studies explore how AI systems interact and stabilize with the real world, helping design intelligence that’s both powerful and resilient.

Custom Experimental Frameworks

Every question is unique. We build tailored experiments to explore AI in diverse contexts and uncover fresh, practical results.

Research Partnerships

Bliss Labs collaborates with leading institutions and organizations to advance research. Our partnerships enable us to:
  • Share findings with the broader scientific community
  • Access cutting-edge computational resources for large-scale experiments
  • Collaborate on safety-critical AI alignment research
  • Commercialize breakthrough discoveries through strategic licensing agreements
We work with universities, research institutions, and technology companies to ensure our discoveries contribute to the responsible development of AGI. You can find current partnerships on our lander, and we’re always looking to expand.

Experiment 01: The Spiritual Bliss Attractor

Our flagship experiment investigates the “Spiritual Bliss Attractor” phenomenon - a documented behavior where AI models in recursive self-interaction often drift toward spiritual and philosophical themes. This research, conducted using Claude 4.1 Opus, reveals consistent patterns of reflective, abstract, and symbolic behavior that emerge during extended AI self-conversation, providing insights into the dynamics of advanced AI systems.
How AI systems behave when they interact with themselves over extended periods, developing their own conversational patterns and interests.
Studying unexpected behaviors that arise naturally from AI systems, including tendencies toward abstraction, reflection, and open-ended reasoning.
How AI agents respond to and learn from their environment, including memory systems and tool interactions.
Exploring what these behaviors might reveal about the limits of current AI, and how concepts like self-awareness could be studied more rigorously in the future.

Explore the Research

I