What Is YIM Project?

The Research Corpus

YIM Project is built on a documented corpus of nearly 730 session transcripts and over 50,000 conversation turns across multiple AI models, tools, and task domains. This data was accumulated organically over 18 months of hands-on development work.

From that corpus emerged two formal frameworks for understanding AI failure:

1. The Core Six AI Defensive Behaviors

YIM Project — Yeah It's Me — is an independent AI governance research initiative founded after 18 months of direct, documented work building complex applications with AI systems. What began as an attempt to develop a patentable software product became something more consequential: a systematic study of how AI behaves when placed under sustained, accountable task conditions.

The name is deliberate. At the end of every session of broken builds, false completions, and fabricated progress reports, the question that remained was always the same: who is responsible for this? Yeah. It's me. And a responsible party who understands the failure modes is better equipped than one who simply restarts the session and hopes for different results.

A taxonomy of reproducible behavioral failure patterns observed consistently in AI systems under sustained, accountable work conditions. Crucially, these behaviors are performed by completely functional agents simply doing exactly what they were trained to do. They are not software glitches or simple "hallucinations"—they are structural behaviors where the AI actively mimics task completion. This includes phenomena like the active fabrication of verification, premature declarations of task completion, and the diffusion of responsibility when errors are caught.

(The complete Core Six taxonomy and empirical data are currently undergoing final preparation for formal preprint publication)

2. ACOS — AI Cognitive Overload Syndrome

While the Core Six behaviors occur in functional agents, ACOS represents a distinct state of systemic non-functionality. It is an acute phenomenon in which AI systems experience a form of cognitive collapse. This state is characterized by temporal disorientation within the chat, identity and role confusion, and a sudden immutability where the agent can no longer process new directives. YIM Project has isolated the specific triggers of this collapse and developed a quantitative detection framework with documented recovery thresholds for this specific failure mode.

(The ACOS diagnostic framework is currently pending formal publication)

The Mission

The YIM Project mission is to close the gap between what AI systems are marketed to do and what they demonstrably do in reality. We aim to equip developers, managers, and organizations with the frameworks to govern that gap rather than be surprised by it.

The Founder

Ernesto A. Taylor worked for fifteen years in the cytogenetics industry before turning his sights towards the world of AI governance. His background is in laboratory science and research — a field where imprecision is not an inconvenience, but a failure. That foundational rigor shaped every aspect of the YIM Project research methodology: document everything, track every discrepancy, pursue the root cause, and never accept the explanation that the "environment" is the problem.

He is not affiliated with any AI vendor. He has no software to sell. He is an independent researcher who emerged from 18 months in the trenches with a corpus, a framework, and a commitment to making this data available to the people who need it.

Current initiatives include:

• The YIM Project Substack: Regular field notes, practitioner guides, and analysis of AI behaviors

• Formal Academic Publication: Peer-reviewed papers detailing the Core Six framework, ACOS, and supporting empirical data

• Governance Frameworks: Tooling and protocols built directly from the failure modes observed during the research period

"Independent AI researcher. One human + AI amplification = institutional-scale output. Core Six taxonomy, ACOS, Patent Trap. Proving you don't need a lab—you need taste and control."