About / The ideas behind the work
Human agency,
development and AI.
Human at the Origin asks how people’s intentions, corrections and refusals can remain consequential as assistance develops. It connects this continuing agency to human development and the fuller realization of AI’s possibilities.
The purposes AI serves
Human beings originate the AI systems considered here, and people originate particular undertakings with them. That beginning carries responsibilities. An initial instruction or approval cannot settle every later purpose, use or permission.
The right to approve or refuse is one part of agency. A person may retain that authority while AI influences the judgment through which they exercise it. Effective agency also depends on understanding, practical alternatives, time and support. Designers and institutions share responsibility for those conditions.
Radical Agency is engaging with AI while actively retaining - and developing - our capacity to shape purposes, question influence, make commitments, change direction, and withdraw.
Here, “radical” returns to the roots of intention: whose purposes are being served, how they formed, and whether we still endorse them. The practical question is whether reconsidering an arrangement changes what the system does.
Deliberate delegation, rest and continuing support can all be worthwhile choices. Agency requires supportive conditions; designers and institutions share responsibility for making correction and withdrawal effective. People’s choices remain accountable to evidence, professional obligations and other people’s rights.
The reciprocal thesis is that radical human agency will also unlock maximal AI potential. Better questions, clearer purposes, informed experimentation and corrective feedback are possible pathways. This is a proposition for investigation in defined activities and circumstances.
Learning, making and becoming
Completing a task with AI, retaining what was learned, and becoming better able to learn are different outcomes. “More capable of becoming more capable” names a developmental aspiration to investigate. Continuing assistance may also remain valuable in its own right.
Creative work can move between idea and implementation: an idea prompts something made, and the result changes the idea. AI may participate in that exchange while also influencing the confidence and momentum with which it proceeds.
Personalized assistance can be conceived as an outgrowth developing through a person’s purposes, experience and continuing participation. Such development would need revisable consent, transparency and demonstrated trustworthiness for the activities entrusted to it.
The person remains more than any account a system can make of them.
Related conceptual work: AI as Second Nature, v0.5 · Operational Psyche
Understanding consequential influence
A person may ask AI to help understand a situation they are worried about but uncertain of. If the response makes that worry sound more certain than the evidence warrants, it may influence their interpretation, self-understanding and subsequent decisions.
People cannot inspect everything AI does. Where its influence may be pivotal, they need understandable accounts of the relevant assumptions, evidence and uncertainty. Designers share responsibility for making those points visible. Permission to prepare a summary is separate from permission to share it.
Selected research on learning, judgment and control informs these questions. It does not establish the effectiveness of Radical Agency. Research & evidence ↗
Three perspectives
on human change.
The Tripartite Model brings together structural, mechanistic and phenomenological perspectives on how people experience themselves and change. Developed in the context of psychotherapy, it also informs proposed connections between human development and AI design.
Structure, mechanism and experience
Recurring self-states, the revision of expectations, and lived experience are considered together, each contributing a different account of the encounter.
Self-states and transitions
PSSM provides a structural perspective on characteristic ways of experiencing and responding, and the transitions between them.
Expectations and revision
AIT provides a mechanistic perspective on how expectations meet experience and become open to revision.
Experience in context
EFT provides a phenomenological perspective on lived experience as it takes shape within a setting and relationship.
Together, the three perspectives connect recurring organization, the possibility of change and the encounter in which change can occur. Their application to AI is a research and design proposal. The analogy does not give a system access to a person’s inner life.
Public foundation: A Tripartite Psychotherapy Model. The AI connection is discussed in Toward Integrated Human-Centered Multi-Agent Systems, §5.
Design and human development
Assistance can serve different purposes: completing a task, supporting practice, or helping a person examine an idea. These purposes call for different forms of participation and different criteria for evaluation.
Developmental Object AI Engineering (DOAIE)
DOAIE is a proposed approach to engineering concerned with these distinctions. It considers what a system remembers, how it responds to uncertainty, when it takes initiative and how a person can redirect it. Its longer view asks what happens to human capability, relationships and authorship through continued use.
“Developmental object” refers to something encountered in a relationship through which change may become possible. AI can matter psychologically without being a person or a therapeutic authority.
Seven provisional design values
- Human authorship
- Keep personal aims and consequential permissions answerable to the people affected.
- Epistemic humility
- Distinguish what is observed, reported, inferred and still unknown.
- Boldness under uncertainty
- Make room for exploration while keeping possibility and evidence distinct.
- Correction and provenance
- Keep meaningful information attributable and open to revision.
- Viable nonuse and exit
- Make pausing, declining and leaving practical possibilities.
- Plural relationship
- Preserve human connections and the freedom to choose continuing support.
- Development over time
- Examine learning, participation, burden and other outcomes separately.
Provisional values, drawing on existing work in values-based design and responsible AI: AI as Second Nature, §1.
Continuity in AI assistance
Earlier goals, attempts, corrections and outcomes could inform a later encounter with AI, giving assistance continuity across situations and as a person’s life changes.
General Use Persistent AI Developmental Object (GUPAIDO)
GUPAIDO is the public concept for this form of persistent assistance, guided by DOAIE.
Persistence raises questions about purpose, accuracy and consent. Earlier information may cease to be relevant, and preferences may change. A person needs ways to revise what is carried forward and decide whether its continued use remains appropriate.
Design approach
DOAIE concerns the principles and evaluation of AI design in relation to human development.
Persistent assistance
GUPAIDO concerns how assistance guided by those principles might continue across time and contexts.
Public definition: AI as Second Nature, §7.
Applications and evaluation
The published work discusses several possible applications. These examples describe areas for design and evaluation; they do not establish outcomes from implemented systems.
Learning and practice
Assistance with study or practice could provide explanations, feedback and opportunities to attempt a task. The level of assistance would depend on the learner’s purposes and the role of teachers or mentors.
Evaluation would distinguish performance with assistance from retention and the ability to use what was learned in another setting. Ongoing support may also have value in its own right.
Writing and creative work
A writer uses AI to question an argument, explore a counterexample or try a different form. The writer decides which suggestions serve the work, what to reject, and when it is ready.
The relevant considerations include voice, variation, verification and responsibility, as well as the finished work.
Planning across changing circumstances
Persistent assistance could draw on earlier goals, decisions and corrections when helping with later planning. The person would need to review which information remains relevant and revise assumptions as circumstances change.
Evaluation would consider the usefulness of continuity alongside the person’s ability to correct, redirect or discontinue the assistance.
Care coordination
A person could organize questions, changing priorities and an account of earlier encounters for review with a clinician. Uncertainty and disagreement would remain visible, with decisions held by the appropriate people.
The design would require attention to privacy, the accuracy of the record and professional accountability.
This is an informational coordination example. It provides no diagnosis, treatment selection or crisis service.
Applications discussed in AI as Second Nature, §§7, 10–11, with relational context from Tuning at the Hyperinterface.
Publications and working papers
These public papers develop the conceptual background. The Zenodo works are working papers; repository publication does not establish peer review or effectiveness.
- What Will It Be Like When AI Becomes Second Nature? ↗DOAIE, GUPAIDO and seven provisional design values · Zenodo · v0.5
- A Tripartite Psychotherapy Model ↗The original clinical framework: PSSM, AIT and EFT · Neuromodec
- Toward Integrated Human-Centered Multi-Agent Systems ↗The proposed connection between the Tripartite Model and AI design · Zenodo
- Tuning at the Hyperinterface ↗Participation, differentiation and human sovereignty in relationships with AI · Zenodo
- Operational Psyche ↗Development and relational causality in human–AI systems · Zenodo
- AI as Scalable Developmental Object ↗A conceptual account of thinking, containment and development · Zenodo
Grant H Brenner
Human at the Origin brings together Grant H Brenner’s writing and conceptual work on AI, human agency, relationships and development. DOAIE and GUPAIDO are part of that ongoing inquiry.
The resource collection includes public papers, essays and conversations. The Toys page presents related creative and interactive experiments.
Resource collection ↗AI, agency & human development
Human purposes
and the use of AI.
Human at the Origin examines how people can shape the purposes AI serves, question its influence, and direct assistance as their lives change. The work connects continuing human agency to learning, creativity, relationships and the possibilities of AI.
Agency,
over time
Radical Agency names a practice of shaping purposes, questioning influence, making commitments and changing direction. It includes choosing what to delegate, when to rest, and when to withdraw. The question is whether those choices continue to affect what AI does. Designers and institutions share responsibility for making that possible.
Development
and possibility
AI can help people explore ideas and make things. Human purposes, questions and corrections may also help reveal further uses for AI. This reciprocal relationship is a central proposition of the work. Its value needs to be examined through what people learn, create and can meaningfully choose over time.