MindHYVE.ai

First Editionv1.0March 2026Lexicon

The AI Practitioner Lexicon.

76 terms for what AI practitioners experience but cannot yet name.

Author
Bill Faruki
Founder & CEO, MindHYVE.ai
Chair, California Institute of AI
Published by
California Institute of AI (CIAI)
In partnership with MindHYVE™
46 pages, English, PDF
Download the Lexicon (46 pages, 334 KB)English · PDF · First Edition
01Why this exists

There is something peculiar about the way the best AI practitioners talk about their work. They sound like philosophers. Not by choice. They sound like philosophers because they are trying to describe experiences for which no precise language exists, and philosophy — with its tolerance for abstraction, its comfort with the unnamed — is the closest available register.

Modern inference AI is roughly three years old in any meaningful public sense. Three years is not enough time for a culture to develop the linguistic infrastructure to describe a fundamentally new category of human experience. Practitioners reach for metaphor. They borrow vocabulary from adjacent domains. They drift, inevitably, toward philosophy — because philosophy is the discipline that has always handled the unnamed.

Language shapes what we can think. The vocabulary we build for AI will shape what AI becomes. Let us build it deliberately.

02The six domains

Six loci of experience. Seventy-six terms. One vocabulary.

Each entry contains four components: the definition, an in practice scenario, the why it matters rationale, and a common misidentification clause that says what the term does not mean. Cross-domain connections are frequent and intentional. The lexicon is a network, not a list.

IDomain

Perception

What practitioners detect in model behavior. Inference drift, coherence lock, flattening, register shift, depth ceiling. Felt qualities of AI output that experienced users read instinctively, but had no precise language for.

IIDomain

Craft

What practitioners actually do. The skills, techniques, and disciplines that constitute AI proficiency. Scaffold loading, temperature reading, exit recognition. Names for moves that proficient users perform routinely and reflexively.

IIIDomain

Cognition

What changes inside the practitioner. The cognitive shifts, biases, and transformations that sustained AI use produces. The ways your own thinking is reshaped by the work — frequently invisible to the person doing it.

IVDomain

The Between

Phenomena in the interaction space. Dynamics that exist in the relationship between human and model, belonging fully to neither. The loop. The feel. Things that are real but not located inside either party.

VDomain

The Commons

Social and communicative phenomena. What happens between practitioners, between practitioners and non-users, and in the broader AI discourse. The vocabulary gap. The experiential gap. The double compression.

VIDomain

Terrain

Systemic and field-level conditions. The large-scale patterns that characterize the organizational and societal landscape of AI. The ground we all stand on while working with these systems.

03A sample entry

One representative entry from Domain I (Perception). Every term in the lexicon follows this four-part structure: definition, in practice, why it matters, common misidentification.

03

Coherence Lock

Domain I · Perception

Definition
The perceptible moment when a model fully captures your intent and begins generating output that tracks not just the surface of your request but its underlying logic, context, and purpose. A felt sense that the model is building from the same foundation you are.
In practice
After two rounds of refining your prompt for a technical architecture document, the model's third response suddenly feels different. It is not just following instructions — it is anticipating implications, connecting elements you had not explicitly linked, and making choices that align with a design philosophy you described but did not name. Coherence lock has been achieved.
Why it matters
Coherence lock is the state that distinguishes productive AI sessions from frustrating ones. Practitioners spend significant effort trying to reach it and can feel immediately when it arrives. Without a name, this state cannot be discussed, targeted, or taught as an objective.
Common misidentification
Not the same as the model agreeing with you. Coherence lock is about depth of comprehension, not alignment of opinion. A model in coherence lock may push back on your ideas — but it pushes back from a place of understanding rather than misinterpretation.

Seventy-five more terms in the full lexicon.

04Read further
Download the full Lexicon (PDF, 46 pages)

The AI Practitioner Lexicon is published by the California Institute of AI (CIAI) in partnership with MindHYVE™. CIAI is a wholly-owned subsidiary of MindHYVE.ai, Inc. and operates The Dawn Directive™ AI fluency program through the ArthurAI™ Vocational Learning Edition platform. This page hosts the first-edition v1.0 release with the publishers' permission.