Longitudinal human effects
Whether exposure or reliance produces durable changes in attitudes, behavior, trust, relationships, development, or participation.
10 connected questions
UNCERTAINTY / METHODS / RESEARCH
Open Questions in AI PSYOPS Research
A cross-category agenda of 36 unresolved questions about effects, measurement, autonomy, authentication, rights, and safer research.
The reports repeatedly distinguish demonstrated capability from what remains unknown. This register brings every category’s stated research gaps into one view so readers can compare where evidence is thin, contested, or methodologically difficult.
CROSS-CUTTING THEMES
Nine recurring evidence problems
The themes help readers compare related gaps across otherwise different categories. They are navigation labels, not claims that every question has the same answer or urgency.
Causal and field validity
How to separate AI effects from user choice, platform design, human operators, prior beliefs, and other real-world causes.
19 connected questions
Cross-platform and multimodal systems
How influence changes when text, audio, video, biometrics, ranking, conversation, languages, and platforms interact.
7 connected questions
Agent memory and autonomy
Whether agents can sustain identity, goals, memory, and strategy over long periods without drift or unsafe escalation.
4 connected questions
Detection and authentication
How to identify coordination or synthetic evidence without treating probabilistic indicators as proof or misclassifying authentic users.
9 connected questions
Privacy-preserving audit access
How researchers can obtain sufficient exposure, network, and delivery data without creating new surveillance or privacy harms.
5 connected questions
Vulnerable populations and safe intervention
How to study and reduce harms to minors, isolated users, and people in crisis without abrupt, punitive, or stigmatizing responses.
7 connected questions
Culture, law, and rights
How concepts such as autonomy, manipulation, authority, harm, evidence, and remedy vary across cultures and jurisdictions.
8 connected questions
Models, simulation, and feedback
How to validate synthetic audiences, predictive models, and interventions when systems alter the behavior and data they are meant to measure.
6 connected questions
RESEARCH AGENDA
Compare what the reports say remains unresolved
Question wording comes directly from each category’s structured research-gap synthesis. Evidence links point to the category section and the retained source records behind that synthesis.
Showing 36 open questions
Filters are optional. Without JavaScript, the complete research agenda remains visible below.
CATEGORY 01
Long-term retention of AI-induced attitude change.
AI-Assisted Traditional Psychological Operations
Longitudinal human effects
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 01
Validity of synthetic audience simulations under real-world shocks.
AI-Assisted Traditional Psychological Operations
Models, simulation, and feedback
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 01
Comparative performance of AI assistance versus well-resourced human teams.
AI-Assisted Traditional Psychological Operations
Causal and field validity
Models, simulation, and feedback
Open question—not a finding, forecast, or capability claim.
CATEGORY 02
Longitudinal behavioral effects outside survey settings.
AI-Generated Propaganda
Longitudinal human effects
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 02
How labels affect true and false content across different audiences.
AI-Generated Propaganda
Detection and authentication
Causal and field validity
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 02
Real-world comparative effectiveness of synthetic media modalities.
AI-Generated Propaganda
Causal and field validity
Cross-platform and multimodal systems
Open question—not a finding, forecast, or capability claim.
CATEGORY 03
Longitudinal effects of adaptive multi-turn persuasion.
AI-Driven Personalized Influence Operations
Longitudinal human effects
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 03
Reliable measurement of situational vulnerability without creating surveillance harms.
AI-Driven Personalized Influence Operations
Privacy-preserving audit access
Vulnerable populations and safe intervention
Open question—not a finding, forecast, or capability claim.
CATEGORY 03
Independent replication using real platform delivery systems.
AI-Driven Personalized Influence Operations
Privacy-preserving audit access
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 04
Metrics for strategic coherence across thousands of asynchronous interactions.
Autonomous AI Influence Agents
Agent memory and autonomy
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 04
Cross-platform memory and identity persistence.
Autonomous AI Influence Agents
Agent memory and autonomy
Cross-platform and multimodal systems
Open question—not a finding, forecast, or capability claim.
CATEGORY 04
Safe governance of evolving agent memory.
Autonomous AI Influence Agents
Agent memory and autonomy
Detection and authentication
Open question—not a finding, forecast, or capability claim.
CATEGORY 05
Long-term effects of discovering a trusted identity was synthetic.
Synthetic Persona Operations
Longitudinal human effects
Vulnerable populations and safe intervention
Open question—not a finding, forecast, or capability claim.
CATEGORY 05
Privacy-preserving cross-platform entity resolution.
Synthetic Persona Operations
Detection and authentication
Privacy-preserving audit access
Cross-platform and multimodal systems
Open question—not a finding, forecast, or capability claim.
CATEGORY 05
Active defenses that do not harass authentic users.
Synthetic Persona Operations
Detection and authentication
Vulnerable populations and safe intervention
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 06
Cross-language and cross-platform coordination measurement.
AI-Driven Disinformation Swarms
Cross-platform and multimodal systems
Detection and authentication
Open question—not a finding, forecast, or capability claim.
CATEGORY 06
Long-term psychological effects of sustained synthetic noise.
AI-Driven Disinformation Swarms
Longitudinal human effects
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 06
Reliable distinction between malicious coordination and genuine collective action.
AI-Driven Disinformation Swarms
Detection and authentication
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 07
Longitudinal effects across multiple platforms.
Algorithmic Perception Control
Longitudinal human effects
Cross-platform and multimodal systems
Open question—not a finding, forecast, or capability claim.
CATEGORY 07
Independent auditing with adequate exposure data.
Algorithmic Perception Control
Privacy-preserving audit access
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 07
Separating user demand from algorithmic supply.
Algorithmic Perception Control
Causal and field validity
Models, simulation, and feedback
Open question—not a finding, forecast, or capability claim.
CATEGORY 08
Durable psychological effects of adaptive systems.
AI-Enabled Emotional and Behavioral Manipulation
Longitudinal human effects
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 08
Combined influence of language, voice, gaze, and biometric signals.
AI-Enabled Emotional and Behavioral Manipulation
Cross-platform and multimodal systems
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 08
Culturally robust definitions of manipulation and autonomy.
AI-Enabled Emotional and Behavioral Manipulation
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 09
Longitudinal effects on adolescent development and human relationship skills.
AI-Assisted Conversational Entrapment and Recruitment
Longitudinal human effects
Vulnerable populations and safe intervention
Open question—not a finding, forecast, or capability claim.
CATEGORY 09
Effective, non-punitive disengagement interventions.
AI-Assisted Conversational Entrapment and Recruitment
Vulnerable populations and safe intervention
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 09
Interaction between recommender systems and conversational reinforcement.
AI-Assisted Conversational Entrapment and Recruitment
Cross-platform and multimodal systems
Models, simulation, and feedback
Open question—not a finding, forecast, or capability claim.
CATEGORY 10
Effectiveness of labels and prebunking across cultures.
AI-Enabled Deepfake Psychological Operations
Detection and authentication
Culture, law, and rights
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 10
Privacy-preserving provenance that survives distribution.
AI-Enabled Deepfake Psychological Operations
Detection and authentication
Privacy-preserving audit access
Open question—not a finding, forecast, or capability claim.
CATEGORY 10
Legal standards for authenticating synthetic or disputed evidence.
AI-Enabled Deepfake Psychological Operations
Detection and authentication
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 11
Causal long-term effects of predictive interventions on trust and participation.
AI-Based Predictive Population Management
Longitudinal human effects
Models, simulation, and feedback
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 11
Mathematical evaluation under self-fulfilling and self-defeating feedback.
AI-Based Predictive Population Management
Models, simulation, and feedback
Causal and field validity
Open question—not a finding, forecast, or capability claim.
CATEGORY 11
Legal remedies for aggregate probabilistic harm.
AI-Based Predictive Population Management
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 12
Multi-year effects on human relationships, judgment, and autonomy.
AI as an Independent Psychological Authority
Longitudinal human effects
Vulnerable populations and safe intervention
Open question—not a finding, forecast, or capability claim.
CATEGORY 12
Cross-cultural differences in AI authority and spirituality.
AI as an Independent Psychological Authority
Culture, law, and rights
Open question—not a finding, forecast, or capability claim.
CATEGORY 12
Technical methods to reduce sycophancy without losing useful support.
AI as an Independent Psychological Authority
Agent memory and autonomy
Vulnerable populations and safe intervention
Causal and field validity
Open question—not a finding, forecast, or capability claim.
No open questions match these filters
Reset the filters or broaden the search. The full agenda contains all 36 questions retained from the category reports.
HOW TO USE THIS AGENDA
Research questions should narrow uncertainty without reproducing harm
The reports favor longitudinal, independent, culturally robust, privacy-preserving, and ethically reviewed methods. High-risk manipulation research should avoid unconsented exposure, vulnerable participants, and field designs that create the harms being measured.