AAI INFLUENCE INSTITUTE
INDEPENDENT RESEARCH · AI INFLUENCE

We study
AI influence.

Independent empirical research into how artificial intelligence changes what people know, learn, believe, prefer, decide and do.

Explore research ↗ Collaborate
01 / 09 SCROLL TO INVESTIGATE БУЭНОС-АЙРЕС · 2026
01 — RESEARCH AGENDA

AI influence is not
just a model problem.

01↗
◎

AI Security & Safety

How AI systems fail at trust boundaries between models, instructions, tools, data, software and humans.

02↗
◌

Knowledge, Learning & Cognition

How AI changes learning, knowledge formation, memory, critical thinking and the development of human intellectual abilities.

03↗
△

Preferences & Behavior

How AI shapes preferences, beliefs, decisions and behavior through framing, personalization and repeated interaction.

04↗
⌁

Human–AI Dependence

How trust, reliance, delegation and cognitive dependence emerge between people and AI systems over time.

05↗
∿

Autonomous & Critical Systems

How AI enters high-consequence environments and changes human decisions, behavior and outcomes over time.

02 — METHODOLOGY

Measure the influence.
Do not assume the answer.

We study AI influence through controlled experiments, repeated interaction and measurable changes in knowledge, learning, cognition, preferences, decisions and behavior.

L1

Exposure

Define what the person or system encounters: prompts, answers, framing, recommendations, memory and interaction history.

L2

Change

Measure changes in knowledge, confidence, reasoning, preferences, decisions or behavior against an appropriate baseline.

L3

Longitudinal effect

Test whether effects persist, accumulate, reverse or alter the way people learn and develop over repeated interaction.

/ / /
03 — RESEARCH PRINCIPLE

Measure what changes.
Distinguish influence from noise.

The objective is not to declare AI good or bad. It is to establish, with reproducible evidence, how interaction with AI changes people and systems — and under which conditions those changes matter.

OBSERVE THE CHANGE.
DO NOT MANUFACTURE THE OUTCOME.
FIELD
STUDY
01
04 — RESEARCH MODEL

From snapshots
to longitudinal evidence.

We compare baselines, controlled interventions and repeated interactions to distinguish short-term response effects from durable changes in knowledge, cognition, preferences and behavior.

∞PEOPLE / INTERACTIONS / CHANGES OVER TIME
05 — COLLABORATION

Research collaboration takes many forms.

We collaborate with system owners, researchers, engineers, institutions and funders. The common requirement is simple: a real question worth investigating.

01

Research collaboration

Provide a controlled system, sandbox, dataset, simulation or domain expertise. We design and run the investigation together.

Propose a research project ↗
02

Funding & sponsorship

Support independent research not only with funding, but also with infrastructure, access to leading-edge models, compute, test environments, datasets, tooling or other research resources — without controlling the research agenda or conclusions.

Discuss support ↗
03

System owners

If you have an AI system worth testing, we can work in staging, a sandbox or another explicitly authorized environment.

Offer a system ↗
06 — FOUNDING SPONSORS

Be there
at the beginning.

The Institute is currently being established. Before formal sponsorship opens, we are creating a Founding Sponsors Waitlist for those interested in supporting independent AI research from the beginning.

Join the waitlist ↗
07 — MEDIA

Research worth
reporting.

For journalists, editors and analysts: research findings, methodology, expert commentary and background on AI influence, learning, cognition, behavior and AI safety.

PRESS / MEDIA For interviews, background briefings, research requests or permission to cite our work. Contact research team ↗
09 — CONTACT

Have a question
worth investigating?

Research, collaboration, case submissions and general inquiries.

inbox@ai-influence.org ↗