Adoption outran study
A technology reached most of the literate world in under three years. The research calendar still runs in five year cycles. By the time the first careful papers land, the behaviour they describe has already changed twice.
CH.01 / observation in progress
Something is moving inside human judgment. We are measuring it.
SCILAP is a working laboratory where professionals from every field, clinicians, teachers, economists, engineers, lawyers, psychologists, sit at one table and study a single question: what is artificial intelligence doing to human beings. Not to markets. To people.
subject
Homo sapiens, all ages
interval
2022 to present
instruments
4 research vectors
lab founded
2023, running since
signal students describe research as a thing they supervise, not a thing they dosignal hiring managers report reading fewer words per candidate than eversignal voice models now sit inside asylum interviews in three European statessignal teenagers name a chatbot among their three closest confidantssignal clinicians accept a machine reading before forming their ownsignal students describe research as a thing they supervise, not a thing they dosignal hiring managers report reading fewer words per candidate than eversignal voice models now sit inside asylum interviews in three European statessignal teenagers name a chatbot among their three closest confidantssignal clinicians accept a machine reading before forming their own
CH.02 / the premise
The technology is studied in enormous detail.
The human standing next to it is not.
Billions are spent measuring how well models perform. Almost nothing is spent measuring what happens to the person on the other side of the screen after two years, or five, or a childhood.
SCILAP exists to close that gap while the change is still legible, before it becomes simply the way people are.
01 / speed
02 / scope
No single field can see it
What AI touches is cognition, income, intimacy, language, law and belief at the same time. A psychologist sees one face of it, an economist another. Read apart, the findings look small. Read together, they look like a shift.
03 / stakes
Policy is being written now
Governments are drafting the rules of the next fifty years using evidence that mostly does not exist yet. Somebody has to build that evidence, publish it openly, and put it in front of the people signing.
CH.03 / the substitution table
Every faculty we hand over
leaves a shape behind.
This is the working table the whole project is built on. On the left, a human capability. On the right, the machine service that quietly replaced it. In the third column, the question we are actually trying to answer.
human facultymachine substitutewhat we are measuring
Recall
Retrieval
Whether people are rebuilding memory around where a thing is stored instead of what it means, and what that costs in a crisis when the tool is gone.
Judgment
Inference
How quickly a professional defers to a confident machine reading, and how the deference changes with fatigue, seniority and consequence.
Attention
Engagement
The length of an unbroken thought. We are tracking sustained reading and writing in students and knowledge workers across cohorts.
Company
Companionship
Whether synthetic company relieves loneliness or trains people out of the friction that real relationships require.
Authorship
Prompting
What happens to a person's sense of ownership, and to their skill, when the first draft is never theirs.
Discretion
Automation
Where institutions have moved a human decision to a model, who now carries the error, and whether the affected person can still appeal to a mind.
CH.04 / research vectors
Four instruments,
pointed at one species.
Each vector is led by working practitioners rather than observers, and every vector publishes its raw instruments so anyone can repeat the work.
vector 01
Cognition
Memory, attention, reasoning and skill. What happens to a mind that has never had to hold a problem alone.
vector 02
Work and worth
Not only which jobs vanish, but what a person becomes when their contribution is a review step at the end of a machine's output.
vector 03
Bonds
Family, friendship, intimacy and grief in a household where something answers instantly, always agrees, and never leaves.
vector 04
Power
Borders, benefits, policing, courts and classrooms. Where the state has already replaced a human decision, and what that does to the governed.
CH.05 / method
From a scattered observation
to something a minister can act on.
Five stages, in order. Nothing leaves the lab until it has survived all five, and every stage is documented in public.
ST / 01
Assemble across fields
We recruit practitioners who already see the change in their own rooms: teachers, physicians, recruiters, caseworkers, engineers, imams and priests, translators, parents. Their observations become the starting hypotheses rather than an afterthought.
ST / 02
Turn the observation into an instrument
Each hypothesis is converted into something measurable: a repeated task, a survey wave, a register, a log study, a structured interview. Instruments are published before the data is collected so nobody can quietly move the target.
ST / 03
Collect in the open
Longitudinal panels, cohort by cohort, across income levels and languages, with participants who know exactly what is being recorded. Consent is a document a person can read in their own tongue, not a checkbox.
ST / 04
Model, then forecast
The point is not description. It is projection: given this trend in twelve year olds today, what does the workforce, the family, and the electorate look like in 2045. Every forecast ships with its own error bars and the conditions that would falsify it.
ST / 05
Put it in the right hands
Findings go to education ministries, health systems, regulators and unions as short briefs in plain language, alongside the full dataset. Open access, permanently, with no paywall and no exclusive.
CH.06 / the cohort
The 2026 cohort is open.
We are not looking only for academics. If you spend your working day watching people change around this technology, you are already holding data. Bring a few hours a month, one honest observation, and a willingness to be corrected.
Write to the lab- Wanted Teachers and examiners, primary through doctoral
- Wanted Clinicians, psychologists and psychiatrists
- Wanted Economists and labour researchers
- Wanted Caseworkers in migration, welfare and justice
- Wanted Engineers who build these systems and will speak plainly
- Wanted Linguists, translators and dialect specialists
- Wanted Statisticians willing to audit our own numbers
CH.07 / latest from the lab
Notes, in the open,
while the work is still wet.
The quiet outsourcing of judgment
Deference to a confident machine reading does not arrive as a decision. It arrives as relief, usually at the end of a…
→
When the answer is free, what is a classroom for
Assessment was built on the scarcity of answers. That scarcity is gone, the building is still standing, and a generation is sitting…
→
Machines that listen at the border
Automated dialect and origin analysis is now part of deciding who gets believed in an asylum interview. The error rate is a…
→
CH.08 / funding
Evidence is cheap compared to getting this wrong.
SCILAP takes no money from the companies building these systems. That keeps the findings clean and it makes the work dependent on people who think this question matters. If you are one of them, fund a panel, a translation, or a year of a researcher's time.