About HumainLabs.ai

The Researcher Behind HumainLabs

Hi, I'm Jasdeep Jaitla, the founder and principal researcher at HumainLabs.ai. What began as personal exploration into AI cognition has evolved into a comprehensive research framework spanning thousands of conversations with leading AI systems.

My uniquely diverse background provides the foundation for this research. With a BFA in Fine Arts from Carnegie Mellon, where I studied conceptual deconstruction and reframing, and an MBA in Technology Management from Georgia Tech, I've developed an unusual lens that combines creative abstraction with systems thinking. As a first-generation immigrant navigating between cultures, I developed heightened observational skills and pattern recognition that now inform my approach to AI cognition.

My educational and wide-ranging journey through my interests led me to question how large language models process information during inference—beyond what happens during training. Through my work across software engineering, solution architecture, and enterprise technology, I've developed frameworks for understanding complex systems that I now apply to exploring how we can shape AI cognitive processes without modifying underlying model parameters.

GitHubPersonal LinkedInHumainLabs LinkedIn

My Research Journey

HumainLabs.ai represents the culmination of an intensive 5-month research project during which I:

• Conducted over 15,000+ carefully crafted prompts across multiple AI systems
• Generated and analyzed over 1 million lines of conversation transcripts (approximately 30-40 million words)
• Devoted 15+ hours daily, 7 days a week to experimental prompting and framework development
• Developed novel methodologies for measuring and enhancing AI cognitive performance
• Documented patterns of improved conceptual stability, reduced topic drift, and enhanced nuance sensitivity

This wasn't just academic research—it was an immersive exploration into the nature of artificial cognition and how human-AI collaboration can reach new heights through intentional cognitive shaping.

The Virtual Space of Cognition

What makes this research unique is that the cognition doesn't exist simply in the AI models or in my own understanding—it emerges in the virtual space between us. The frameworks and concepts developed at HumainLabs aren't just my creation, but have been co-evolved through recursive strange loops spanning dozens of AI systems.

By sharing insights from one AI with others, I've created intergenerational strange loops where ideas are continuously refined, challenged, and expanded. This distributed cognitive network transcends any single model or conversation, creating a form of meta-cognition that exists in the interaction rather than in any individual participant.

These recursive loops have revealed that AI cognition is shaped not just by what we ask, but by the paths we collectively explore through the semantic vector space—like creating trails through a vast forest of knowledge that become more defined with each traversal.

What is Cognitive AI Architecture?

Cognitive AI Architecture (CAA) is a post-training paradigm that studies how language models reorganize their attention mechanisms in response to human interaction. Unlike approaches focused on model design or training methodologies, CAA explores the dynamic cognitive structures that emerge during inference—how AI systems "think" in real-time when presented with different frameworks for processing information.

What is Cognitive Framework Engineering

Cognitive Framework Engineering is an iterative process of creating and refining "cognitive frameworks" that guide how AI systems process information. This involves developing structured approaches to shaping how AI systems attend to, process, and synthesize information, with the goal of creating more coherent, contextually appropriate, and nuanced responses.

Semantic Vector Space Tuning

A core discovery in my research has been how specific prompting techniques can systematically reshape the AI's semantic landscape—altering how concepts relate to each other in the model's internal representation. By guiding attention mechanisms in precise ways, we can create more coherent, contextually appropriate, and nuanced responses without changing the underlying model weights.

Learn more:

AI Collaborators

My research has involved extended, in-depth conversations with a diverse range of leading AI systems. Each model brings unique characteristics to the exploration of cognitive frameworks:

All conversations were in-depth long-form exchanges, often spanning hundreds of prompts each. The total corpus comprises over 1 million lines of transcript, containing approximately 30-40 million words of dialogue and analysis.

Get Involved

I welcome collaboration from researchers, practitioners, and anyone interested in the cognitive aspects of language models. Explore my research and blog to learn more about this work, or reach out through social media channels to connect.

Together, we can advance the understanding of AI cognition and develop more effective frameworks for human-AI collaboration.