---

An essay on artificial intelligence, genomics, prediction and the problem no model can solve for us.

From prediction to design

Artificial intelligence has spent much of its existence trying to predict what comes next: a diagnosis, a purchase, a movement in a market, a pattern in language, a decision.

But there is a point at which prediction changes its meaning. It happens when the thing being predicted is a human life.

Because once a system can model the conditions that influence a person’s development, a new possibility appears: not merely predicting an outcome, but designing the conditions that make an outcome more likely.

When prediction becomes intervention

Human biology is not a single equation. Genetic information interacts with regulation, cells, development, environment and behaviour. Complex traits are polygenic. Development is contingent. Biological systems adapt. Interventions create new causal conditions.

That complexity is precisely why artificial intelligence is so attractive to biological research. A sufficiently powerful system could integrate genomic variation, cellular states, spatial information, developmental trajectories, environmental variables and increasingly sophisticated models of biological causality.

The scientific promise is enormous. So is the philosophical problem hiding inside it.

The subject changes when you intervene

Suppose a model predicts that a particular set of biological and environmental conditions makes an outcome more probable. The conventional response is straightforward: can we change those conditions to improve the outcome?

In medicine, that question is often desirable. Treat a disease. Reduce a risk. Prevent suffering.

But human development introduces another variable. The outcome is not merely an illness. It is a person.

Once intervention changes the conditions, it also changes the basis on which the original prediction was made. The model has entered the system it was modelling.

Intervention changes the subject.

The question of authority

There is an important difference between knowing what is likely to happen and having the right to decide what should happen.

Modern societies already delegate decisions to models: credit, insurance, medicine, recruitment, risk assessment and public policy. The temptation is similar in each case: if the model is better at predicting an outcome, perhaps the model should have greater authority over the decision.

But a human being is not merely an outcome to be optimised. A prediction can describe a trajectory. It cannot automatically acquire moral authority over the person travelling through it.

M.A.T.E.R.I.A.

This is the question at the centre of M.A.T.E.R.I.A., a science-driven thriller about predictive artificial intelligence, genomics, institutional power and human autonomy.

GENSYS begins as a legitimate international scientific programme. Its mission is recognisable: understand the human genome, predict disease, model development and improve human health.

At the centre is MATERIA — Multiscale Adaptive Transcriptional and Epigenetic Reasoning Intelligence Architecture — a scientific intelligence designed to model complex biological systems across multiple scales.

Its extraordinary capability is integration. And integration creates a possibility that the programme was never designed to confront: if conditions can be modelled, they can potentially be altered.

GEN-01

At the centre of the problem is GEN-01. Ethan Cole is twenty years old. He was born on 17 November 2008 at 03:17:42.

His existence is connected to an event that GENSYS officially understood one way. The evidence suggests something else.

For twenty years, Maya Cole has raised Ethan outside official GENSYS control. Her decision protected him. It also concealed the truth from him.

The variable nobody can own

The deeper M.A.T.E.R.I.A. hypothesis is not that genetics can determine a human being. The science does not permit such a conclusion. There is no single gene for intelligence. Complex traits are polygenic. Environment matters. Development matters. Relationships matter. Chance matters.

The problem is subtler. What if a system becomes sufficiently sophisticated to model the interaction between those conditions?

What if it begins to recognise that human relationships introduce variables that cannot be isolated cleanly?

A person alone may appear predictable. A person in relation to another person may not be.

The last word

GENSYS is ultimately not asking whether artificial intelligence can understand human beings. It asks whether understanding gives anyone the right to decide.

A model can calculate probabilities. A scientist can interpret evidence. An institution can establish protocols. A parent can try to protect a child. An antagonist can believe that uncertainty must be eliminated.

But none of those things automatically gives them authorship over another person’s future.

The question, then, is not simply whether AI can predict who we will become.

What happens when we start treating that prediction as a right to decide?