Individuals and Human Agency

Can LERA prevent AI from making decisions for people?

Detailed answer

Key answer: LERA can prevent high-consequence machine output from automatically gaining the authority to decide and act for people.

Core explanation

LERA can prevent high-consequence machine output from automatically gaining the authority to decide and act for people.

This does not mean AI can never assist decision-making.

AI may provide analysis, identify options, forecast consequences, explain risks, or recommend a course of action.

The distinction lies in whether the AI remains an assistant or becomes the effective decision-maker.

In many systems, that boundary is already becoming unclear.

An AI may recommend which employee should be dismissed, which patient should receive priority, which applicant should receive credit, or which citizen should be investigated.

If the surrounding institution automatically acts on that recommendation, then the AI is no longer merely advising.

It is participating in execution.

LERA addresses this by keeping proposed actions within a Judgment–Governance Layer before they affect the person.

The system must address:

  • whether the AI output is sufficiently reliable;
  • whether relevant information is missing;
  • whether the action affects rights or opportunities;
  • who has legitimate authority;
  • who bears responsibility;
  • which rules apply;
  • whether the action may proceed;
  • whether it must be blocked or escalated.

For example, an employer may use AI to rank candidates.

The ranking may help humans review large numbers of applications.

But if the system automatically rejects applicants without meaningful review, clear responsibility, valid rules, or an appeal path, AI assistance has become institutional execution.

LERA helps prevent this conversion.

It does not ban the use of AI in decision support.

It requires that high-consequence execution remain subject to legitimate judgment and governance.

The key distinction is:

AI may help form information.

AI should not automatically acquire authority over a person’s life.

LERA protects this boundary by ensuring that action remains governable before consequence occurs.