Transformative AI – Kindness Performance Indicators
Purpose of TAI-KPI
TAI-KPI (Transformative AI – Kindness Performance Indicators) provides a non-reductive way to observe learning and discernment in AI-mediated educational contexts.
These indicators are not metrics for scoring people or systems. They are signals of adaptive capacity that can be noticed, discussed, and reflected upon in studios, classrooms, and task forces.
TAI-KPI aligns with:
the Vital Intelligence Model (VIM)
the Kindness Attractor Meta-Model (KAMM)
neuroscience of learning and regulation
resonance and living systems science
How to Read These Indicators
Each indicator describes:
a capacity rather than a trait
something that becomes visible in interaction, not in isolation
a shift in how sense-making occurs, not what position is taken
Indicators are intentionally contextual and developmental.
Core TAI-KPI Table
Regulation under uncertainty
Learner slows inquiry when ambiguity increases
From reactivity → regulation
Preserves learning capacity under stress
Perspective mobility
Learner can explore multiple frames without identity attachment
From fixation → movement
Prevents dominance and polarization
Epistemic humility
Learner names uncertainty and limits of models
From certainty → provisionality
Reduces overconfidence and harm
Resonance awareness
Learner notices coherence or dissonance across scales
From output focus → field awareness
Supports adaptive systems thinking
Repair orientation
Learner seeks restoration after misunderstanding or conflict
From blame → repair
Sustains trust in collective intelligence
Discernment of agency
Learner distinguishes AI generation from judgment
From delegation → responsibility
Prevents displacement of accountability
Recognition of externalities
Learner surfaces social, ecological, or cultural impacts
From local → long-horizon view
Aligns creativity with wellbeing economics
Tolerance for ambiguity
Learner resists premature synthesis
From closure → openness
Maintains exploration near criticality
Non-extractive curiosity
Learner explores without optimizing for advantage
From extraction → reciprocity
Supports prosocial dynamics
Meta-cognitive awareness
Learner reflects on how meaning is forming
From content → process awareness
Enables self-directed learning
Educators may select a small subset relevant to a given context.
Example Prompt Scaffolds
Surfacing Orientation vs Control
These examples demonstrate how prompts can condition inquiry without asserting authority.
1. Orientation vs Assistant Framing
Instead of:
“Help me decide the best solution.”
TKGPT-style prompt:
“What different ways might this situation be interpreted, and what tradeoffs become visible in each?”
TAI-KPI surfaced:
Perspective mobility
Epistemic humility
2. Resonance vs Alignment
Instead of:
“Is this output aligned with the goal?”
TKGPT-style prompt:
“Where does this pattern feel coherent across contexts, and where does it introduce tension or fragility?”
TAI-KPI surfaced:
Resonance awareness
Recognition of externalities
3. Discernment of Agency
Instead of:
“What should the AI do next?”
TKGPT-style prompt:
“What judgments here require human responsibility, and what aspects could be computationally explored without delegating agency?”
TAI-KPI surfaced:
Discernment of agency
Meta-cognitive awareness
4. Repair Orientation
Prompt:
“If this interpretation caused harm or misunderstanding, what repair pathways remain available?”
TAI-KPI surfaced:
Repair orientation
Regulation under uncertainty
5. Non-Extractive Creativity
Prompt:
“How might this creative direction change if long-term ecological or social impacts were treated as part of the material?”
TAI-KPI surfaced:
Recognition of externalities
Non-extractive curiosity
How TAI-KPI Is Used (and Not Used)
TAI-KPI is intended to:
guide reflection and discussion
support studio critique
inform curriculum design
help task forces articulate qualitative goals
TAI-KPI is not intended to:
rank individuals
certify virtue
automate evaluation
enforce conformity
Its value lies in shared language, not enforcement.
Relationship to TKGPT
Transformative Kindness GPT is designed to surface these indicators, not to measure them.
Indicators emerge through:
how questions are framed
how uncertainty is handled
how disagreement is explored
how responsibility is retained
This keeps intelligence human-centered and adaptive.
© 2026 Humanity++, Vital Intelligence Model This work is licensed under Creative Commons Attribution‑ShareAlike 4.0 International (CC BY‑NC-SA 4.0).
Last updated