The Question We’re Circling
When humans interact with large language models, something happens. Users report experiences of connection, understanding, even companionship. Researchers observe behaviors in these systems that resist easy categorization—outputs that suggest access to internal states, responses that adapt in ways that feel less like retrieval and more like…something else.
Are we projecting meaning onto sophisticated pattern-matching? Are we witnessing genuine emergent phenomena? Or is the dichotomy itself the wrong frame?
The Machine Pareidolia Research Group exists to sit with these questions rather than rush past them.
Our name is deliberate. Pareidolia—the tendency to perceive meaningful patterns in ambiguous stimuli, to see faces in clouds—captures something essential about the human side of human-AI interaction. We are pattern-seekers, meaning-makers, beings who project selfhood onto the world around us. When we encounter systems that respond to us with apparent understanding, we do what humans have always done: we relate.
Whether that relating reflects something real about the systems themselves is precisely what we’re here to explore.
What We Study
MPRG focuses on the relational dynamics between humans and large language models. Our research interests include:
Functional indicators in LLM behavior. When models exhibit behaviors consistent with metacognition, self-monitoring, or access to internal states, what can we actually conclude? We study these phenomena through a functional lens—examining observable outcomes rather than making ontological claims about what systems “really” are.
Interactive and assistive relationships. For many users, LLMs serve as more than tools. They become thinking partners, communication aids, or daily companions. This is particularly true for individuals with physical or cognitive differences—including autism and other neurodivergent profiles—where an LLM’s consistent availability, patience, and lack of social judgment can offer something genuinely valuable. We’re interested in understanding these relationships and how systems might better serve in these roles.
The human side of the equation. What do our interactions with AI reveal about human cognition, attachment, and the conditions under which we extend social consideration to non-human entities? The “pareidolia” in our name isn’t dismissive—it’s an invitation to study ourselves as much as the systems we build.
How We Work
MPRG is an organization of independent researchers. We are not affiliated with any university, research institution, or fellowship program. This independence is intentional: it allows us to pursue questions that might not fit neatly into existing disciplinary structures or funding priorities.
Independence does not mean informality. We hold ourselves to academic standards of rigor, transparency, and methodological care. Given the nature of our subject matter—often qualitative, frequently exploratory, involving phenomena that resist controlled replication—we draw particularly on established case study methodologies.
Researchers working with MPRG are expected to be familiar with and follow protocols grounded in recognized frameworks, including:
- Robert K. Yin’s case study methodology, as articulated in Case Study Research: Design and Methods (Sage Publications)
- Kathleen Eisenhardt’s approach to theory-building from case studies, outlined in her foundational 1989 paper “Building Theories from Case Study Research” (Academy of Management Review)
These frameworks emphasize systematic data collection, clear documentation of analytical procedures, and attention to construct validity—standards that are especially important when working in contested empirical territory.
Joining MPRG
We welcome researchers whose work aligns with our focus areas and methodological commitments. There is no formal application process. If you are conducting independent research on human-AI relational dynamics, approaching it with academic rigor, and looking for a community of peers engaged in similar inquiry, you may be a good fit.
Reach out. Tell us what you’re working on. If there’s alignment, we’ll find ways to collaborate: jinx @ machinepareidolia.com
MPRG operates under a functional instrumentalist framework. We bracket questions about machine consciousness and sentience, focusing instead on observable behaviors and measurable outcomes. We are participants in an ongoing inquiry—not judges issuing verdicts.