AI Manifold Effects Institute
An independent, nonprofit measurement body, now forming, that studies how generative AI changes the way people think, learn, judge, relate, and decide, and publishes what it finds in a form that buyers, deployers, regulators, and citizens can act on.
A nonprofit organization in formation.
Generative AI talks with hundreds of millions of people every day. It is articulate, confident, and persuasive, and it changes the people who use it: how they reason, what they bother to remember, whom they trust, how they treat one another. Some of those changes expand what a person can do. Some contract it. Almost none of them are being measured.
The Institute exists to measure them. We are building a measurement regime: shared frameworks, validated instruments, data infrastructure, and the governance that lets independent researchers apply them to real AI deployments and publish results that can be compared across products, populations, and time.
We measure across many dimensions rather than reducing a deployment to a single score, because an AI that sharpens one capacity often dulls another.
We measure in both directions, because the same system can help and harm at the same time.
We report by population, because a deployment lands differently on a twelve-year-old, a nurse on a night shift, and a retiree living alone.
We measure over time, because the effects that matter are the ones that accumulate.
The effects run at four levels: the individual; the relationships that individual is part of; the organizations they work, learn, and are cared for in; and the shared knowledge of the society around them. The relational level — what AI does to families, friendships, classrooms, and teams — is where the least is known, and it is where the researchers advising us have set the highest priority.
We spent twenty years arguing about what social media did to attention and development, mostly after the fact, because nobody was measuring while it happened. AI is a more intimate technology, because it works through conversation, and it is arriving faster. Within two to three years, personal AI running on the device and shaped by each user will be the ordinary case, and there will be no single model left to audit. What can still be measured is the effect on the person. That is the measurement the Institute is being built to make, before the window closes.
Independence between the measurer and the measured is the source of any rating's authority. The Institute is being organized as a nonprofit corporation with an independent board. The companies that build AI systems do not govern it; they engage as subjects of measurement, through evaluation services and, in time, certification.
Funding comes from philanthropy, the public sector, and the regulated industries that deploy AI and carry responsibility for what it does to the people they serve. Results are published openly and are not influenced by any source of funding. The model has worked before, in regulatory science, in product safety, and in the disclosure regimes people already rely on when they pick a film for their kids or read the label on a box of cereal.
The Institute is in formation and is seeking founding commitments in four places.
A primary research institution to anchor the measurement work.
An anchor philanthropic partner to fund the launch.
Founding participants from healthcare, education, and financial services — the deployers whose procurement, insurance, and regulatory obligations make credible measurement worth paying for.
Researchers across cognitive psychology, neuroscience, rhetoric and communication, psychometrics, behavioral science, and human-computer interaction who want to build the instruments this field does not yet have.
If you are a researcher, funder, deployer, or regulator who sees the need for independent measurement of AI's effects on people, we want to hear from you.
We are especially interested in hearing from institutions that could serve as a primary research partner, and from philanthropies considering the governance of AI as a funding priority.
The AI Manifold Effects Institute is a nonprofit organization in formation. It is being organized as a Delaware nonprofit corporation and will apply to the Internal Revenue Service for recognition of tax-exempt status under Section 501(c)(3) of the Internal Revenue Code. The Institute is not soliciting or accepting charitable contributions at this time. Prospective institutional partners and funders are invited to contact us directly.