Read a sample: 1. When a Decision Feels Automatic
Maya submits a claim after a storm damages her roof. A short time later, she receives a message saying that the claim needs additional review. The message offers no explanation beyond a request for more information. Maya wonders whether a computer has flagged her as suspicious.
Across town, Daniel applies for a job he believes he is qualified to do. His application is acknowledged, then declined the next morning. He has not spoken with anyone at the company. He wonders whether an algorithm rejected him before a person saw his résumé.
Priya applies for a small loan to replace a failing car. The lender declines her application and sends a notice referring to information considered in the decision. She is surprised because she has been careful about paying her bills. She wonders whether the lender used inaccurate information—or whether a model drew an unfair conclusion from accurate information.
Meanwhile, Carlos tries to regain access to an online account after a routine sign-in fails. A chatbot repeatedly directs him to the same reset page, which he has already tried. He wonders whether the company has locked his account for a reason no employee can see.
These situations have something in common: a consequential outcome arrived through a process the person experiencing it could not see. They also have an important difference. Maya, Daniel, Priya, and Carlos do not yet know precisely what happened behind the scenes.
Maya’s claim may have been selected by an automated rule, sent for a routine document check, or referred by an employee. Daniel’s application may have been screened by software, reviewed quickly by a recruiter, or affected by a hiring change. Priya’s lender may have used a combination of application details, outside information, business rules, and predictive tools. Carlos may be dealing with an automated security step, a technical error, or a service process that gives the chatbot too few ways to help.
The feeling that a decision was automatic is worth paying attention to. It can prompt you to preserve a notice, check the facts, and ask for an explanation. But the feeling is not, by itself, proof of how the decision was made or whether it was improper.
This chapter begins with a simple distinction: what happened to you is a fact you can describe; why it happened may still be a question. Keeping those two things separate gives you a stronger starting point.
We often picture an automated decision as a single machine making a single choice: approve or deny, hire or reject, allow access or lock the account. Real processes can be less tidy.
A system might sort applications before a recruiter reviews them. It might suggest that an insurance claim needs a closer look, leaving an adjuster to decide what to do next. A lending tool might produce an assessment that is combined with other information. A customer-service chatbot might not decide anything about an account, but it might control which questions a customer can ask and where the request goes.
Automation can therefore affect an experience even when a person is involved somewhere. The reverse is also true: a fast, standardized, or frustrating response does not establish that artificial intelligence was used. A human can follow a fixed policy, send a template, or make a mistake just as a software-driven process can.
For now, you do not need to identify the technology. You need a clear account of the outcome and a useful next question.
Consider the difference between these two messages to an insurer:
> Your AI falsely accused me of fraud. Reverse its decision immediately.
> My claim was referred for additional review, and I was asked to provide more information. Please explain what is being reviewed, what information you need from me, and whether there is a deadline for providing it.
The first message asserts facts Maya does not know. The second describes what she received and asks for information she can act on. If the insurer later identifies an error, Maya can address it directly. If the review turns out to be routine, she can decide how to respond without having started from an accusation.
A focused question does not mean accepting an unfair outcome. It means giving yourself the best chance of finding out what needs to be challenged.
The next steps depend on the kind of decision and what the notice actually says. The four consumers’ situations show why it helps to pause before choosing a response.
Maya’s insurer has asked for additional information. That may delay a decision, but the message does not say her claim has been denied. She starts by saving the message and checking the claim number, the requested documents, and any stated date for responding.
She also writes down what she does not understand: Is the insurer asking for photographs, repair estimates, or something else? Is a particular part of the claim under review? How should she submit the material securely?
Maya can ask:
> Is my claim still being considered? Please tell me what information you need, why it is relevant if you can share that, and when I should expect an update.
If the company later denies some or all of the claim, she can request clarification about that outcome and any available review process. At this stage, her immediate goal is narrower: understand the request and avoid missing a stated deadline while she gathers information.
Daniel’s…