peopleanalyst

research / ai-human-interaction / pipeline

Pipeline status

What's drafted (Paper 1 in two registers, the sub-paper plan, the field-mapping roadmap, 29 literature reviews, 6 syntheses) · what's queued (long-context-emergence review, OSF preregistrations, external-operator pilot) · what's gated on the Penwright production system accumulating a useful run-in period of data.

AI–Human Interaction·Pipeline

AI–Human Interaction — pipeline status

Honest snapshot, 2026-05-03. The program is in scaffold + literature-mapping phase; empirical work runs once the Penwright production system has accumulated a useful run-in period of data.

Drafted

Paper 1 — AI-Augmented Authorship and Human Skill Development — both versions.

  • Technical literature-review draft at penwright-paper-01-technical.md. Synthesizes phenomenology of skill, attention theory, transactive memory, standpoint theory, epistemic injustice, cognitive apprenticeship, working-alliance theory, psychoanalytic theory, improv theory, translation theory, niche construction, and institutional economics into a single positioning argument for Penwright.
  • Public manifesto draft at penwright-paper-01-public.md. General-audience version of the same argument.
  • Both drafts pre-date the formal program scaffolding; structure may shift in revision but the core claim — AI writing systems should be measured on whether they make humans more capable, not on output fluency — is stable.

Penwright Sub-Paper Plan v1.0 — at penwright-sub-paper-plan.md. The publication-and-product-integration strategy for all twelve papers. Verbatim source.

The roadmap — at roadmap.md. The twelve-branch HAI field map plus the six cross-cutting research programs that define Track B. Verbatim source from 2026-05-01.

Literature reviews drafted (secondary, LLM-generated)

Twenty-nine per-topic reviews are present in sources/topic-reviews/ covering:

  • Trust calibration and reliance under varying AI reliability in software work
  • Conversational AI interaction design after 2021
  • Explainability and interpretability for AI code generation
  • Distributed cognition for AI-augmented professional work
  • Organizational and sociotechnical embedding of AI coding agents
  • Parasocial dynamics, companion AI, and emotional engagement
  • Longitudinal cognitive effects and skill change in AI-assisted programming
  • Cognitive apprenticeship, ZPD, and AI as a more-knowledgeable other
  • Feminist epistemology, standpoint theory, and AI knowledge production
  • Political philosophy and democratic theory of AI
  • Creativity, ideation, and co-creation with AI
  • Aesthetics of AI-generated art and AI-augmented practice
  • Persuasion, belief change, and AI's epistemic effects
  • Indigenous and non-Western relational ontologies in AI ethics
  • Communication accommodation theory in human-AI interaction
  • Institutional economics of AI and the reorganization of work
  • Rupture and repair in human-AI coding interaction
  • Working alliance and rupture-repair from psychotherapy to HAI
  • Companion-species and ethology of working partnerships
  • Translation and interpreting studies for AI as translator
  • Niche construction theory: feedback between organisms and environments
  • Improv theatre and ensemble performance for human-AI dialogue
  • Phenomenology of skill and AI-augmented expert work
  • Phenomenology of attention in post-2010 theory and AI-augmented work
  • Transactive memory in human-AI teams
  • Ritual studies and non-human agents for long-term human-AI use
  • Disability studies, crip theory, and AI as access infrastructure
  • Psychoanalytic theory for human-AI working alliance measurement
  • History of mediation technologies and AI
  • Memory, mourning, and digital remains

Six cross-LLM syntheses are present in sources/syntheses/ covering: stigmergic coordination, coordination-failure in multi-agent LLM systems, ironies of automation in coding-agent supervision, risk compensation outside automotive safety, transactive memory, CMU faculty-mapping (program-design adjacency).

Status of the .docx originals (2026-05-03): all 29 topic reviews have been converted to .md and live in sources/topic-reviews/. Originals archived at sources/_originals/ for the audit trail.

Queued — second pass

  • Long-context emergence — dedicated review. Highest-leverage frontier zone gap; needs its own deep-research run.
  • Calibration of personalization — dedicated review. Distinct from the trust + parasocial reviews; raises paternalism-vs-autonomy questions.
  • Developmental psychology of children with AI — review commissioning decision. Ethically and politically sensitive; warrants a program-level decision before drafting.
  • Conversation-analysis / ethnomethodology review. Roadmap §IV mentions; no review yet.

Queued — for the empirical track

  • Penwright production data accumulation. The program is gated on the Penwright system having accumulated enough session data to support claims at meaningful effect sizes. Data collection is continuous; preliminary descriptive analysis can begin once 50+ extended-session writers have logged at least 3 months of activity.
  • Paper 4 (Measurement Framework) formal write-up. Vision document exists at vela/docs/VISION-PENWRIGHT-MEASUREMENT.md; conversion to a standalone paper is the next foundational deliverable after Paper 1.
  • Paper 3 (Authorship Packet Model) formal write-up. Vision document exists at vela/docs/VISION-PENWRIGHT-AUTHORSHIP.md; same next-foundational-deliverable status.
  • Paper 5 (Dependency) preregistration. First OSF preregistration candidate. Protocol in draft.
  • Paper 7 (Genre) preregistration. Second OSF preregistration candidate. The non-collapsibility of memoir / nonfiction / fiction is load-bearing for the rest of the program.
  • External-operator pilot. 5–10 external Penwright users to mitigate the auto-ethnography threat-to-validity.

Follow-up ingestion needed

The literature reviews and syntheses surface dozens of primary sources that the empirical work will eventually need to cite, integrate, or argue against. Rather than ingest all of them upfront, the program operates on as-needed ingestion: when a sub-paper enters drafting, the relevant primary sources get pulled in, verified, and cited.

Categories of follow-up ingestion that will be queued as the program advances:

  • Articles — peer-reviewed primary studies (HCI, CSCW, empirical SE, educational psychology, psychotherapy outcome research)
  • Books — foundational monographs (Hutchins, Wegner, Stiegler, Citton, Vygotsky, Dreyfus, Bordin, Lutz, Briggs, Haraway)
  • Transcripts — therapy-transcript and case-study sources where working-alliance constructs can be operationalized
  • Effect-size tables — meta-analyses where they exist; pre-registered synthesis where they don't
  • Models — formal computational and statistical models (Bass diffusion, working-alliance models, transactive-memory operationalizations)
  • Variables and measures — instruments to be cataloged into Principia where they fit organizational measurement, and into Penwright's measurement framework where they fit writing-skill measurement

The Principia program (sibling at peopleanalyst.com/research/principia) is the canonical home for the measurement-related ingestion; Vela's emotion-corpus-expansion is the canonical home for affective-computing-adjacent ingestion. This program coordinates with both rather than duplicating.

Recently completed (2026-05-03)

  • Program scaffold landed in peopleanalyst-site at content/research/ai-human-interaction/
  • Multi-LLM literature-review pass completed for 28 topics
  • Six cross-LLM syntheses completed
  • Penwright Research Program v1.0 formalized
  • Paper 1 drafted in technical and public form
  • Cross-links from Vela's Penwright vision documents to this program established

Coming soon

  • Paper 4 (Measurement Framework) — converted from vela/docs/VISION-PENWRIGHT-MEASUREMENT.md into standalone paper form
  • Paper 3 (Authorship Packet Model) — converted from vela/docs/VISION-PENWRIGHT-AUTHORSHIP.md into standalone paper form
  • Long-context-emergence dedicated review
  • Manifest entries on the live /research index at peopleanalyst.com