The Human Enhancement Quotient (HEQ) and the Augmented Intelligence Score (AIS): connecting AI literacy and cognitive development through governance.
Inside any organization using AI, some people turn it into value and others turn it into risk. There is currently no way to tell them apart, and nobody has priced what that costs.
The consequences are already landing on people. Companies have run layoffs and justified them by saying the people let go were not AI trainable, and nobody could say how they knew, because there was nothing to measure with. The same gap runs the other way. An insurer cannot price a risk nobody measures. When an AI-assisted output causes harm, the question a court asks is who governed the decision, and most organizations have no record of whether a person exercised control or simply approved the machine. Training budgets have no baseline, so nobody can prove the training worked. And when an AI project underperforms, the easiest target is the employee, even when the failure belonged to the method or the product.
All of that runs on a single unmeasured variable, which is how well a person governs the work they do with AI.
The Human Enhancement Quotient measures that variable. It is a behavior-anchored score of the ability to direct, challenge, verify, and own AI-assisted work, and the Augmented Intelligence Score is how the result is expressed. Every major AI literacy framework names this same competence and then stops short of measuring it, which is the gap HEQ was built to close.
What HEQ Measures
It scores the relationship between a human authority and a machine capability inside a working decision, which is a different object from the one intelligence tests were built for. Most formal definitions of intelligence describe a single agent that predicts, compresses, or acts. HEQ has two parties and measures how one governs the other.
It scores the method, not the answer
This is the design decision everything else follows from. HEQ credits a person who reached the wrong conclusion through sound method, and catches a person who reached the right one through none.
That sounds counterintuitive until the alternative is examined. Scoring outcomes rewards luck, punishes well-governed work that ran into bad evidence, and cannot be applied across organizations that disagree about what a good outcome is. Scoring the method travels. A hospital, a bank, and a school hold different values and different laws, and the question of whether a person checked the sources, preserved the dissent, and owned the decision reads the same in all three.
Why the method is the thing scored. Sound governance scores well whether the answer turned out right or wrong, and an ungoverned right answer scores poorly because luck does not repeat. The four dimensions average into the Augmented Intelligence Score, with Collaborative Intelligence Quotient, the calibration dimension, marked as the one people consistently score worst on.
The Four Dimensions
Read as literacy, the four are competencies at named positions in the checkpoint chain rather than abstract traits.
| Dimension | The literacy it names |
|---|---|
| Collaborative Intelligence Quotient | Management itself: directing the tool, bringing in sources, surfacing conflict, arbitrating, deciding |
| Ethical Alignment Index | Responsible use: owning the output, seeing the harm, refusing to let the machine carry authority it should not, and knowing the law and culture that bind the work |
| Cognitive Agility Speed | Fluency: working with the machine quickly and clearly, moving between framing and detail without friction |
| Adaptive Growth Rate | Growth: getting better at management over time, turning failures into method |
The Augmented Intelligence Score is the equal-weighted mean of the four, reported as an indicator rather than a validated factor score. Equal weighting is deliberate and provisional, because a weighted composite needs factor loadings from validation studies that have not been run.
The score is a factor in human judgment and never the sole basis of a decision. An organization holding it uses it three ways: to develop the individual, to refine its own training, and to oversee its own governance. A poor score is context for a supervisor, not a verdict on a person.
One finding has held every time
Collaborative Intelligence Quotient scores lowest. It came out lowest in all ten people in the December 2025 cross-user testing, and lowest across the five platforms in the original baseline. That is the calibration dimension, the one measuring whether a person knows when to push back, and it lines up with independent research on over-reliance and on human-AI combinations underperforming the better partner when the interaction is unstructured.
It is also the dimension that moved most under sustained checkpoint practice. Ten months of single-practitioner self-monitoring recorded it rising from 88.4 to 93.4, which is n of 1 and is stated as n of 1 rather than buried in a footnote.
Where It Came From
The arc runs in one line. Factics Intelligence became HEQ, HEQ added the Augmented Intelligence Score, and the whole thing arrived at measuring the governance competence at the center of AI literacy.
The question came first. In February 2024 the argument went out that the method makes us more intelligent, which is a claim that demands a way to check it.
The first attempt came in 2024 with the Factics Intelligence Dashboard, built to measure applied intelligence in session across six domains. The governing idea was that IQ was designed to measure intelligence in isolation, and what the work needed was intelligence measured in context.
Then testing turned up something it was not designed to find. The measurement itself was producing growth. People were not only performing better with AI, their scores improved under repeated structured assessment. That shifted the design from measuring performance to measuring enhancement, the six domains consolidated to four, and the Dashboard became HEQ. Version 1.0 published in September 2025 under the framing it carried at the time, measuring cognitive amplification.
The formal framework followed on December 22, 2025, in the enterprise edition. It set the four dimensions and the composite, recorded a cross-platform consistency coefficient of 0.96 across five architecturally distinct platforms, and carried the cross-user testing that found the same dimension lowest in all ten participants. It also placed the instrument where it would be used, in hiring, performance review, and training validation, with the safeguards that use demands.
From there the work moved into the papers. February 2026 set the theoretical foundation and named the measurement framing. April 2026 put the instrument against workforce displacement and the absence of any individual-level assessment. The scoring rubric followed on April 15, synthesized from twelve independent rubric proposals produced by eleven AI platforms answering the same question in isolation, with the convergent elements becoming the behavioral anchors.
September 2026 is where it lands. Measuring the Governance Competence at the Center of AI Literacy develops and extends the February paper rather than replacing it, and it names what the instrument had been measuring the whole time. The developmental claim never left either. It is the fourth dimension, which is what connects the literacy question to cognitive development rather than leaving them as separate subjects.
What the Rubric Actually Controls For
A measurement that can be gamed measures nothing, so three controls sit inside the rubric.
- Validity controls that detect rubber-stamping. A person who approves everything scores as a person who approves everything, which is the same failure Checkpoint-Based Governance watches for at the checkpoint.
- A universal structural floor keyed to irreversible human consequence. Some requirements do not bend regardless of setting, because some decisions cannot be taken back.
- A graded band that is regional and value-laden, and says so. Above the floor, what counts as good governance depends on the laws and values of the place the work happens, and pretending otherwise would make the instrument portable only to places that already agree with it.
Three deployment modes share one set of control questions so results sit on a comparable spine: Independent, Personal, and Professional. The Independent clean-slate mode is the enterprise model, because an organization measuring its own people needs a version that does not depend on the history between them.
Why an Organization Would Care
The spend on AI is measured. The tooling is measured. The human governance sitting between an AI output and a real decision is not measured at all, and that has become the variable on which insurability, liability, hiring, and organizational return all turn.
The pressure is arriving from two directions at once. Courts have begun permitting discovery into whether AI was used to supplant human decision-making, and at least one European court has treated an AI system’s statements as the deploying company’s own rather than accepting that users were expected to verify them. Regulation points the same way. The European Union will require meaningful human oversight of high-risk systems from December 2027, and recent United States state law gives a person who receives an adverse automated decision the right to request meaningful human review by a trained reviewer with authority to overturn it. Both tracks ask for documented human oversight rather than asserted control.
What measurement converts is an unmeasured liability into a managed one. Risk that can be measured can be priced. Governance that can be documented can be defended. A standard signal turns hiring into a decision rather than a guess, and a baseline makes training spend provable.
The productivity claim is the part the market oversells, so it is worth being precise. Governed work does produce more, and volume is the byproduct rather than the point. The point is output that can be trusted because a person governed it, the sources hold, the uncertainty was preserved instead of smoothed over, and someone is accountable for what was released.
Where HEQ Sits
The other frameworks produce the behavior. This one measures it. Factics supplies the evidence discipline, HAIA-RECCLIN makes a platform show its work, HAIA-CAIPR runs the same question across several platforms and reads the disagreement, and Checkpoint-Based Governance puts a named human at the decision. What checkpoint practice builds in that person is what HEQ measures and the Augmented Intelligence Score expresses.
What This Is Not Yet
Two claims are kept apart on purpose, and collapsing them would be the same methodological stripping this work criticizes elsewhere.
That governed human-AI practice develops capability is grounded in decades of learning science and does not rest on this instrument.
That HEQ measures that development accurately is a deployed, research-grounded instrument at diagnostic stage. It has not been validated on an independent cohort, no outside research group has replicated the administration, and the longitudinal record is a single practitioner offered as feasibility rather than validation. Anyone adopting it now becomes a validation partner whose data will strengthen the case or break it.
One finding in the research deserves a direct answer rather than a footnote. A 2024 meta-analysis found that human and AI combinations do not reliably outperform the better of the two alone on task outcomes. That caps outcome claims, and this instrument makes none. HEQ measures whether the human governed the process, and the value of a governed process does not depend on an outcome advantage. What it depends on is that the work leaves a record someone can answer for.
Read the Full Paper
Measuring the Governance Competence at the Center of AI Literacy
Working paper v6.0.5, September 2026. The current statement of the framework: the stakes that make the measurement necessary, the theory, the full scoring rubric with its behavioral anchors and validity controls, the three deployment modes, and appendices carrying the operational assessment prompts and the offline independent test. Archived at DOI 10.5281/zenodo.20840683.
Measuring Augmented Intelligence: Theoretical Foundations
February 2026. The paper the June work develops and extends, carrying the theoretical lineage from Licklider and Engelbart through hybrid intelligence research.
The Other AI
How the measurement work fits the wider argument about augmented intelligence, and why the operative variable is the method rather than the technology.
Factics Make Us More Intelligent
February 2024. The question the instrument was eventually built to answer.
Common Questions
What does HEQ actually score?
How a person governs work done with AI: whether they direct it, challenge it, verify it, and own the result. Four behavioral dimensions are scored and averaged into the Augmented Intelligence Score. Output volume is not one of them.
How does HEQ relate to AI literacy?
AI literacy frameworks name governing and managing AI as a core competence and then stop short of measuring it. HEQ measures that competence directly, which puts it at the center of what AI literacy claims to be about rather than alongside it.
Why score the method instead of the outcome?
Because outcomes reward luck and do not travel. A person can reach a good result through no method at all, and a well-governed decision can still meet bad evidence. Scoring the method credits the first case honestly and catches the second, and it reads the same in settings whose values and laws differ completely.
Is this an IQ test for AI users?
No. IQ was built to measure one mind working alone. HEQ measures a relationship between a person and a machine inside a real decision, and it tracks a trajectory rather than taking a snapshot, because the question is whether the person is getting stronger through the collaboration or weaker.
What is the difference between HEQ and AIS?
HEQ is the framework and the four behavioral dimensions it scores. The Augmented Intelligence Score is the composite those dimensions average into, which is the number that gets tracked over time.
Which dimension do people score worst on?
Collaborative Intelligence Quotient, every time it has been administered. That is the calibration dimension, the one that measures knowing when to push back on the machine, and it is also the one that moved most under sustained checkpoint practice.
Can someone game it by looking diligent?
The rubric carries validity controls written against exactly that, because a score that rewards the appearance of review measures nothing. Approving everything reads as approving everything.
How solid is the evidence?
The instrument is at diagnostic stage and is published as such. Cross-platform consistency has been measured, the rubric was synthesized from eleven platforms working in isolation, and the longitudinal record is one practitioner over ten months. It has not been validated on an independent cohort, which is the next thing it needs and the reason it is published openly.
Basil C. Puglisi, MPA
A Human-AI Collaboration
#AIassisted using HAIA Ecosystem