Skip to content
Contact
☰
Color scheme: Dark.
Back to research
RESEARCH SPACE A

Human–AI Interaction & Co-Evolution

How do humans, AI systems and their relationship change when interaction continues over long periods?

What this space studies

How people and AI change each other through sustained collaboration. EIL looks at trajectories over weeks and years, because roles, trust and dependence only show over time. The object is the long-term change of a coupled human–AI system and the question whether repeated interaction produces new stable forms of collaboration.

Central questions

  • How do roles between human and AI change over weeks and years?
  • When does support turn into dependence?
  • How do human and AI repair errors and misunderstandings?

The relationship as a coupled process

The relationship between human and AI is studied as a coupled process. The open question is whether and how both sides change each other over time. A single point in time shows a state. Only the sequence of interactions shows how roles, conventions and rules arise, stabilize or disappear again.

Research programs

2 programs of its own, 1 more touch this space.

Further programs that touch this space (1)

Core phenomena

  • Longitudinal interaction
  • Co-adaptation
  • Roles and agency
  • Trust and dependency
  • Interaction repair
  • Feedback
  • Further topics

Core phenomena

Longitudinal interaction

  • sessions, weeks, months and years
  • continuity and discontinuity
  • recurrence of interaction patterns
  • change of interaction regimes
  • temporal emergence of the relationship as a coupled process

Co-adaptation

  • mutual adaptation
  • changed working practices
  • changed technical working methods
  • changed expectations
  • changed roles
  • emergence of new interaction conventions
  • change of interaction rules
  • observable adaptation versus claimed co-evolution

Roles and agency

  • role formation
  • role switching
  • role inversion
  • initiative
  • delegation
  • agency
  • shift of control
  • new roles through long-term interaction

Trust and dependency

  • trust calibration
  • over- and under-trust
  • augmentation
  • outsourcing
  • dependency
  • recovery after miscalibration

Interaction repair

  • errors and misunderstandings
  • contradiction
  • repair and repair strategies
  • restoring semantic continuity

Feedback

  • feedback loops
  • recursive interaction
  • amplification
  • attenuation
  • self-correction
  • co-adaptation
  • feedback of externalized structures on the interaction

Further topics

  • persistent human–AI interaction
  • continuity across sessions
  • augmentation versus dependence
  • interaction dynamics
  • Emergent User Class
  • Meta-Interaction

Keywords (12)

  • longitudinal human–AI interaction
  • feedback and repair
  • augmentation and dependence
  • role stabilization
  • role reversal
  • dependence vs augmentation
  • behavioral adaptation
  • semantic adaptation
  • human–AI co-evolution
  • emergence of interaction conventions
  • change of human and technical working methods
  • observable adaptation and claimed co-evolution

Further research questions of this space (3)

  • When does a single signal become a pattern, and when does a pattern become a stable structure with a functional effect?
  • Does a pattern recur across independent contexts, or does it belong to one context only?
  • When does a recurring interaction pattern become a stable structure?

8 questions of this space's programs are with the program details →

How this space is investigated

Observation and relation

The space is studied through longitudinal interaction trajectories.

Observation and relation

What is specifically observed

Behavior and role changes on both sides are recorded as observed, readings such as dependence or trust as inferred or hypothesized, and trajectories that tool drift could explain as contradicted or unknown. The emergence of new conventions and rules is followed across sequences of interactions, because they become visible only over the course of time.

Relation to the other spaces

This space supplies the interaction trajectories from which Human Cognition, Behavior & Externalization reads observable behavior and Emergence, Adaptation & Complex Dynamics derives recurring patterns. Without this space the instruments in Measurement, Evidence & Research Infrastructure have nothing to measure over time.

Behavioral & Interaction Intelligence

Using behavior and interaction as evidence about system state, constraints and decision dynamics.

Central question

What system state, relationship or mechanism best explains the observed behavior and the available evidence?

Behavioral & Interaction Intelligence

Scope
  • Behavioral patterns
  • Decision dynamics
  • Interaction effects
  • Adversarial behavioral analysis
  • Deviation patterns
  • Prediction-relevant behavior
  • Recursion depth of dependent revision cycles
  • Turn coupling & repair cycles
Approach

Collect behavioral traces and decisions over time, compare them against expected patterns, adversarially test deviations instead of explaining them away immediately, and only classify findings as prediction-relevant once they survive a counter-evidence test.

Repeated deviations are scored by the number of materially dependent revision steps rather than the sheer count of repetitions.

Every observed behavioral change is checked against plausible alternative explanations (a model change, a shifted task mix, growing user experience) before it is attributed to a cause.

At the interaction level, this tracks how strongly consecutive turns are coupled in content, how terms get reused across turns, and how repair and clarification cycles develop into stable conventions or recurring disagreement.

Relationship

Supplies the behavioral evidence that Reconstruction, Perspective & World Models and Measurement, Evidence & Research Infrastructure build on, and feeds the Observatory's drift and role-stability signals.

Human–AI Systems

Studying how human and AI behavior co-evolves across long-running interaction.

Central question

How do roles, agency and trust between human and AI change across sustained interaction, and when does adaptation tip into dependency?

Human–AI Systems

Scope
  • Interaction trajectories
  • Adaptation
  • Agency
  • Dependency
  • Semantic integrity
  • HMI / UX
  • Recovery
  • Longitudinal evaluation
  • Offloading vs. active appropriation
  • Role stabilization and inversion
  • Reciprocal adaptation
  • Neuroadaptive extension (planned)
Approach

Capture interaction trajectories longitudinally rather than session by session, compare agency and semantic integrity over time, separate interface-induced information loss from actual behavioral change, and explicitly test recovery after misadaptation.

With delegation to an AI system, the distinction is whether a result is merely accepted or actively checked, corrected and reused, to separate agency from plain dependency.

In parallel, it tracks whether roles between human and system stabilize or invert over time.

Repeated interaction can change behavior and the architecture through which later interaction takes place at all; reciprocal adaptation is therefore tracked longitudinally through a dedicated co-evolution index.

Neuroadaptive extensions, such as attentional or physiological signals as an additional evidence source, are a planned, not yet implemented research direction.

What remains open is which signal types (such as EEG, other BCI signals or wearable data) could meaningfully be linked to interaction trajectories, which consent and privacy questions would need to be settled before any collection, and how such a signal would be calibrated against established behavioral measures. These signal types are not yet an existing capability.

Until the prior questions are answered, the space stays with longitudinal behavioral and interaction analysis alone, without added physiological data.

Relationship

Connects the HMI perspective with Behavioral Analysis and supplies the longitudinal baseline the Observatory measures drift and role stability against.

Methods, systems and axes

Methods (20)

All methods of this space

Methods and analysis procedures (9)

Research artifacts that emerged from interaction (11)

Analysis framework (1)
Architectures, protocols and benchmarks (3)
Measurement systems and research instruments (4)
Metrics (3)

See all methods →

Systems (8)

All systems of this space

Research systems (2)

Multi-agent systems (1)

Shared technical foundation (2)

System state, memory, security & integrity (3)

See all systems →

Cross-cutting research axes (6)

How the axes act in this space

Cross-cutting condition Ethics / Human Agency / Human Rights →

Evidence & limits

Adaptation and co-evolution

What can be observed first is that the behavior of one side follows the other. Co-evolution is spoken of only when both sides change each other so that the change of one alters the conditions of the other. That an observed adaptation belongs to such an interaction remains a hypothesis and is tested against explanations such as tool drift.

Boundary of this space

Observed changes in human to AI relationships cannot automatically be attributed causally to the AI interaction. Other social, biographical and technical influences have to be considered separately.

How to read status and origin

Every statement carries its status (observed, inferred, hypothesis, planned or unknown), no program is externally validated, and origin statements about systems are candidates.

Evidence register → Origin of the systems →

Topics and publications

working formulation Formulated from the research program, not yet a fixed wording. Exception: 01 Emergent Intelligence.

Topics with this space as main space (2)

Further topics that touch this space (5)

Guiding questions of the topics

  • 02 Human–AI Co-Evolution

    How does repeated interaction change the human's use of the system and the structure of the interaction itself?

  • 09 Interaction Dynamics

    How do interaction patterns evolve across extended periods, beyond individual conversations?

  • 01 Emergent Intelligence

    When does interaction stop being merely interaction and begin producing persistent intelligent structure?

  • 04 Neuroadaptive Human–AI Systems

    How can systems adapt to changing cognitive states while keeping measured and inferred states apart?

  • 06 Memory, State and Identity

    What should a system keep over time, and when does memory produce false continuity?

  • 11 Adaptive Systems

    How does adaptive behavior stay inspectable and traceable?

  • 12 Human–Machine Interfaces

    How do interfaces shape cognition, trust and system understanding, and what will an interface need to show?

Publications (10)

Publications (10)