CALE-AI 2027 — Evaluating AI Interventions: Causal and Longitudinal Human Effects

CALE-AI 2027 is a one-day workshop at the 41st AAAI Conference on Artificial Intelligence (AAAI-27) focused on the rigorous evaluation of how AI systems affect people over time.

AI systems increasingly operate not only as predictors but also as interventions that can influence human decisions, preferences, trust, behavior, autonomy, and long-term outcomes. Recommendations, explanations, nudges, warnings, conversational responses, and actions taken by AI agents can all change the information, options, timing, or decision environment presented to users.

CALE-AI brings together researchers from artificial intelligence, machine learning, recommender systems, information retrieval, human–computer interaction, causal inference, behavioral science, digital health, computational social science, and responsible AI. The workshop focuses particularly on causal and longitudinal evaluation under repeated, adaptive, and personalized AI exposure, including intended benefits, unintended effects, human–AI co-adaptation, and feedback loops.

The workshop aims to move AI evaluation beyond predictive accuracy, offline performance, short-term engagement, and one-time user studies toward methods that can establish whether, how, for whom, and over what time horizon AI interventions affect human outcomes.

Submission Guidelines

We welcome original contributions that address the causal, longitudinal, behavioral, and human-centered evaluation of AI interventions. Submissions should clearly specify, where relevant, the AI intervention or exposure, comparison condition or alternative policy, target population and context, human outcome(s), evaluation methodology, relevant time horizon, adaptation mechanism, and possible unintended effects.

Submission Types

We welcome the following types of submissions:

• Full Research Papers — up to 9 pages, excluding references. These papers should present mature and original research, including a well-motivated research question, appropriate methodology, and validated empirical or theoretical results.

• Short Papers / Work-in-Progress — up to 6 pages, excluding references. These submissions may present early-stage research, preliminary findings, prototypes, methodological developments, ongoing studies, or promising research directions.

• Vision / Position Papers — 4–6 pages, excluding references. These papers should present well-argued perspectives, conceptual frameworks, provocative ideas, open challenges, future research directions, or positions relevant to the evaluation and responsible design of AI interventions.

• Demo Papers — up to 4 pages, excluding references. These submissions should describe system prototypes, software tools, interactive platforms, evaluation environments, or other technical artifacts relevant to the workshop. Authors should explain the purpose, functionality, and potential contribution of the demonstrated system.

• Dataset / Benchmark Papers — up to 4 pages, excluding references. These submissions may introduce datasets, benchmarks, simulation environments, evaluation protocols, annotation resources, or other infrastructure supporting causal, longitudinal, behavioral, or human-centered evaluation of AI systems. Authors should describe the resource, its intended use, construction or collection process, and its relevance to the workshop themes.

• Extended Abstracts — up to 2 pages, excluding references. These submissions may describe preliminary results, novel concepts, experience reports, ongoing projects, interdisciplinary perspectives, or previously presented work that would benefit from discussion within the CALE-AI community. Prior or concurrently submitted work is permitted under this category. Extended abstracts will be treated as non-archival.

All submissions should follow the AAAI formatting style.

Each submission will receive at least two reviews. Accepted contributions may be presented as oral presentations, lightning talks, posters, or demonstrations depending on the submission type and program structure.

Accepted workshop contributions will be made available on the workshop website, subject to author permission, and will be treated as non-archival workshop publications.

Submission deadline: November 20, 2026
Author notification: December 2, 2026

Submission site:
https://easychair.org/my/conference?conf=caleai2027

List of Topics

Topics of interest include, but are not limited to:

• Causal evaluation of AI interventions

• Longitudinal and dynamic effects of AI systems

• Repeated and adaptive AI interventions

• Randomized experiments and A/B testing

• Micro-randomized and sequential trials

• Quasi-experimental and counterfactual evaluation

• Causal machine learning and heterogeneous treatment effects

• Off-policy evaluation and sequential decision making

• Time-varying confounding and non-stationarity

• Human–AI co-adaptation and feedback loops

• Habituation, reactance, and preference formation

• Recommender systems and personalized AI

• LLM-based agents and conversational AI

• Digital health and behavior-change technologies

• AI-supported decision making

• Trust and appropriate reliance

• User autonomy and informed consent

• Transparency, fairness, privacy, and contestability

• Non-manipulation and responsible AI interventions

• Intended and unintended behavioral effects

• Human-centered evaluation metrics

• Health, sustainability, well-being, and user agency

• Long-term and distributional outcomes

• Evaluation datasets and benchmarks

• Simulation environments

• Reporting frameworks and evaluation standards

• Reproducible evaluation protocols

Organizing Committee

Mehrdad Rostami — University of Oulu, Finland
Mehrdad.Rostami@oulu.fi

Christoph Trattner — University of Bergen, Norway
Christoph.Trattner@uib.no

Giuseppe Alessandro Veltri — National University of Singapore, Singapore
gaveltri@nus.edu.sg

Mourad Oussalah — University of Oulu, Finland
Mourad.Oussalah@oulu.fi

The organizing team brings complementary expertise in responsible AI, recommender systems, personalization, behavioral science, causal evaluation, longitudinal user behavior, and behavioral analytics.

Venue

CALE-AI 2027 will be held in person in Montréal, Canada, as part of the AAAI-27 Workshop Program.

AAAI-27 takes place from February 16–23, 2027, with workshops scheduled for February 22–23, 2027. CALE-AI is planned as a one-day workshop; the exact workshop date will be announced once the final AAAI-27 workshop schedule is confirmed.

Contact

All questions about submissions and the workshop should be emailed to:

Mehrdad Rostami
University of Oulu, Finland
Mehrdad.Rostami@oulu.fi

Workshop website:
https://caleai.github.io/