Designing with Intelligence
AI contributions during inquiry, creativity, evaluation and decisions: what did the system propose, and how did the designer exercise judgment?
PhygitAi | Beyond Phygital · Research orientation
A connected view of design, intelligence and responsibility.
PhygitAi is an ongoing interdisciplinary design research initiative originated by Abbas Sibai, PhD. It investigates both how people design with AI and how they design for systems whose intelligence can influence behavior in physical and digital settings.
Within the initiative, a proposed methodology has been developed, including a conceptual framework, operational procedures and educational instruments. These resources remain open to revision. Their development has drawn on conceptual synthesis of existing research, illustrative interdisciplinary applications and exploratory comparisons with established approaches. Systematic empirical studies, including course-based testing, remain planned. Practical effectiveness, educational applicability, methodological contribution and limitations are research questions, not predetermined findings.
Developed methodological proposal · Empirical testing planned
Conceptual methods and illustrative cases
No demonstrated comparative superiority
Start and orient
One inquiry, several levels of detail.
This atlas is an orientation map, not a reproduction of unpublished research manuscripts or a substitute for discipline-specific professional practice. The books develop the theory, cases and educational inquiry at greater depth; the public Wiki explains foundational terms; the platform pages describe public screening workflows.
01 / Explore concepts · Book I
Foundations before method.
Two research directions are considered in three contexts: design research, the design process, and product function and use. This is an analytical map, not a rigid sequence or a claim that all products must use AI.
Designing for Intelligence
Products and environments in which sensing, inference or generated responses may affect behavior, permission, human intervention and responsibility.
Beyond AI integration
Human intelligence and unforeseen material, cultural and technological discoveries may transform design independently of AI or in interaction with it.
Conceptual component index, C01–C15
- C01 Physical, digital and phygital foundations
- C02 Forms of computational participation
- C03 Configurations of AI participation
- C04 Designing with Intelligence
- C05 Designing for Intelligence
- C06 Three inquiry contexts
- C07 Two-by-three analytical matrix
- C08 Consequential decision episode
- C09 Reflexive intention–behavior alignment
- C10 Conformity versus adequacy
- C11 Different evidence horizons
- C12 Recognition through execution
- C13 Disagreement pathways
- C14 Moments of negotiated responsibility
- C15 Meaningful human control
The C-identifiers are research inventory references, not fifteen mandatory forms or fifteen novel claims. See terminology and usage for public-language distinctions.
02 / Use the method · Books I and II
Questions, decisions and reviewable records.
The proposed methodology connects conceptual foundations, two research directions, three inquiry contexts, operational procedures and instruments. Six revisitable inquiry gates and a seven-field Intention–Behavior Record (IBR) support its application to consequential decisions; they are not the entirety of the methodology. It is intended for proportionate inquiry, not automatic box-ticking.
Frame and authorize
G0: Intention and assumptions. G1: AI-assisted co-design. G2: Authority and delegation.
Realize, examine, revise
G3: Physical realization. G4: Disagreement assessment. G5: Accountable revision.
Preserve the decision episode
Intention · AI Input · Permission · Event · Outcome · Decision · Retest. Records must distinguish hypothetical events from observation.
Operational inventory, M01–M11
- M01 Six revisitable gates
- M02 Seven-field IBR
- M03 Methodological question reference
- M04 IBR forms and specimens
- M05 Source and assumption register
- M06 Ten-step evidence-aware case building
- M07 Twelve-step gate-to-IBR procedure
- M08 Case quality checks
- M09 Interdisciplinary comparison protocol
- M10 Recurring cross-case inquiries
- M11 Evaluation measures and coding rules
The ten-step procedure structures a bounded case; the twelve-step procedure reviews a consequential episode using gates and IBR. They serve related but different purposes. Full operational forms and unpublished implementation details require source and release review.
03 / Disciplinary transfer · Book II
Seven disciplines. Distinct evidence standards.
Book II uses constructed cases to test the questions that different fields raise. An architectural authorization, cultural refusal, engineering hazard check and UX usability finding are not interchangeable.
| Field | Illustrative inquiry | Evidence boundary |
|---|---|---|
| Graphic communication | Cultural identity, authorship and creative refusal | Constructed design example |
| UX/UI | Adaptive university website, source accuracy, permissions and accessible recovery | Hypothetical episodes |
| Service design | Community mobility assistance and cross-institutional handoffs | Fictional service |
| Industrial design | Connected manufacturing workcell and qualified operator control | Hypothetical workcell |
| Intelligent product | LUMA 01 adaptive reading lamp and competing occupant needs | Concept and scripted analysis; no tested prototype |
| Interior design | STUDY COMMONS 01, conflicting comfort interests and intervention rights | Fictional interior |
| Architecture | BUILDING PATH 01, routes, emergencies and professional authority | Fictional events |
These examples demonstrate possible inquiry use, not real-world efficacy or cross-disciplinary validation. Explore the existing research and projects page for public project summaries.
04 / Learn and teach · Book III
Education should prepare independent judgment and unknown futures.
The proposed educational inquiry examines what to preserve, transform and investigate when AI becomes both a design collaborator and a feature of designed systems. It also asks what human creativity may invent beyond the technological categories we presently know.
AI Integration · Post-Integration Education · Beyond Integration
Complementary educational questions, not inevitable sequential eras: responsible AI use, reevaluation of authorship and judgment, and inquiry into the future structure of design education.
AI-free, AI-supported and AI-primary
These are choices guided by the learning objective, not a hierarchy in which more AI always means more progress.
Independent intellectual capacity
Drawing, observation, material experimentation, theory, critique and creative refusal remain vital where they enable autonomous understanding.
Educational instruments, E01–E10
- E01 Proposed 14-week teaching programme
- E02 Seven learning outcomes
- E03 Five progressive aims
- E04 Designing with Intelligence exercise
- E05 Designing for Intelligence exercise
- E06 Negotiation and Responsibility exercise
- E07 Final Research Dossier
- E08 Five-part proposed assessment
- E09 Recognition-to-action mapping exercise
- E10 Teaching glossary and reading pathway
The proposed university course offers a planned opportunity for initial empirical investigation. A pilot could examine whether students understand and apply the methodology, clarify design decisions, identify potential failures and responsibility issues, and exercise independent creative judgment. It could also assess procedural workload and comparative value against established design approaches. Course-based evidence would inform revision; it would not alone establish originality or effectiveness across settings. Interdisciplinary workshops and comparative studies could extend this investigation.
Book III develops six proposed educational hypotheses (H1–H6), concerning aided versus independent competence, future design categories, creative convergence, critical independence, guided AI participation and openness to unknown possibilities. They have not been empirically established. Instructor examples and rubrics remain proposed resources, not an accredited curriculum. Younger learners can explore separate questions in Learn with Luna.
Current and ongoing investigations
PhygitAi | Current Research Directions
Building on existing international research, PhygitAi investigates three connected questions.
The future of design
How can design research and education respond to intelligent systems that increasingly participate in creating and transforming products, environments and experiences?
The future of design education
How can design education integrate AI while developing and preserving students' independent creativity, aesthetic judgment and critical reasoning?
The practical contribution of PhygitAi
Can a connected educational and design environment help students apply established methods, investigate consequential decisions and revise their work more effectively than existing approaches?
Existing Research and Current Investigations
Selected international work relevant to these questions. These sources establish precedents and active lines of investigation; they are not evidence of PhygitAi's originality or effectiveness.
- Microsoft HAX Toolkit and Google People + AI Guidebook — human–AI interaction guidance and practical design patterns.
- Microsoft Research, Tools for Thought — understanding, protecting and augmenting human cognition in AI-supported work.
- Wharton Generative AI Studio and SuperSkillsStack (2026 preprint) — creative judgment, taste, imagination and human agency in AI education.
- Harvard GSD, AI and the Physical Imaginary — AI, physical making, embodied knowledge and human control.
- Huang, Wensveen and Funk (2023) and DRS Co-speculating with AI (2026) — speculative design, future horizons and human–AI co-creation.
- McGregor (2022), Participation Interfaces for Human-Centered AI and Curhan and Gratch (2026) — stakeholder participation, negotiation and AI roles across teaching and research; important comparisons for UAiPN.
- UNESCO's AI competency framework for students (2024) and the OECD/European Commission AI literacy framework (2026) — established educational reference points, not evidence of UAiPN's effectiveness.
- NIST AI Risk Management Framework Playbook and Singapore's agentic-AI governance framework — risk, authority, monitoring and accountable action.
05 / Examine evidence · Books I and II
A research claim needs a comparison and a test.
PhygitAi currently proposes conceptual relationships and research procedures. The following propositions and studies specify questions to examine; they do not report completed tests.
| Item | Question or proposed investigation | Status |
|---|---|---|
| P1 | Can intentions, AI contributions, permissions and behavior be examined with greater continuity? | Proposed; unvalidated |
| P2 | Can disagreements reveal behavioral failure, inadequate intention or unresolved conflict? | Proposed; unvalidated |
| P3 | Can negotiated responsibility and meaningful intervention be made more reviewable? | Proposed; unvalidated |
| P4 | Can proportionate traceability improve reconstruction without excessive design workload? | Proposed; unvalidated |
| Study A | Controlled documentation and review comparison | Not completed |
| Study B | Observation of genuine human–AI co-design and negotiation | Not completed |
| Study C | Technical and human evaluation after constructing a LUMA 01 prototype | No physical prototype or trial claimed |
| Study D | Interdisciplinary transfer against capable established methods | Not completed |
A correct completion of an IBR does not establish safety, educational improvement, methodological originality or practical superiority. Negative and null results would be meaningful research outcomes.
06 / Established practice before contribution claims
Different purposes. Intersecting functions. Open research questions.
Design methodologies evolve through prior research. We first examine each approach on its own terms, then compare the functions it performs. Similar words do not establish identical methods; different names, audiences or structures do not alone establish methodological originality.
| Comparison criterion | What the investigation asks |
|---|---|
| Purpose | What design or governance problem is the approach meant to address, and for whom? |
| Function | What does a designer or reviewer actually do with the approach? |
| Structure | How are stages, categories, iterations and dependencies related? |
| Unit of analysis | Does it examine an interaction, risk system, design rationale, engineering decision or consequential episode? |
| Instruments and outputs | What decisions, artifacts, evidence or reviewable records does it produce? |
| Evidence and burden | What has been tested, with what baseline, and at what time or cognitive cost? |
Human–AI interaction
Microsoft HAX addresses user-facing AI interaction, while Google PAIR supports human-centered AI product design. Their principal scope differs from a PhygitAi consequential decision episode, although practical work on feedback, error, authorization and revision intersects. No partnership or endorsement is implied.
Decision and system practice
Compare with QOC design rationale, NASA systems engineering, NIST governance, meaningful human control and Chinnam and colleagues’ Decision Digital Thread (2026). These approaches already preserve important relationships among evidence, reasons, authorization and revision. Their purposes and implementation levels differ from a compact PhygitAi IBR.
Learning and creativity
Compare with studio pedagogy and Huang, Wensveen and Funk (2023), whose AI design-education study also uses three future horizons. Cognitive and higher-education concerns are examined by Lee and colleagues (2025) and Simkute and colleagues (2025). None establishes PhygitAi’s effectiveness.
The emerging UAiPN educational proposal requires its own comparisons with participatory AI, negotiated human–AI collaboration, student agency and teaching frameworks. It should not be conflated with the established PhygitAi methodology inventory, or presented as a validated pedagogical intervention.
Particularly close comparisons include Contestable AI by Design, which examines challenge and intervention through development and deployment, and Hybrid Human–AI Co-Creation in Design, which examines design-stage roles and initiative. These are substantive intellectual predecessors to examine, not proof that PhygitAi is redundant.
The remaining question is whether PhygitAi's joined inquiry into creative AI input, revisable human intention, authorization, physical–digital behavior and accountable revision helps designers or learners beyond capable combinations of established methods. A fair test must compare strong baselines, reviewer reconstruction, evidence-status errors, permission mistakes and time required. Positive, null and negative findings are all possible. The cited works are independent prior research, not validators of PhygitAi.
07 / Participate and track
A developing research programme, open to criticism.
A public research identity should make source status, attribution and uncertainty visible. This page is a structured overview; full books, unpublished method assets and comparative research notes are not automatically public release materials.
Research definitions
Visit the Wiki and terminology guide for existing public explanations.
Projects and tools
See project summaries, public screening editions and their stated limitations.
Versions and contact
Check releases and updates, or use the academic contact information for scholarly inquiry.
Originator: Abbas Sibai, PhD. No research partnership, institutional approval, certification or independent validation is represented here unless separately established and documented.