Founding editorial
Keep the Human at the Center of Human Technology
The success of a human technology is not the elegance of the system, but the consequence for the person who encounters it.
Technology becomes human technology at the moment it begins to measure, classify, influence, or act upon a person.
From that moment, technical performance is only part of the evaluation. A system may be fast, precise, scalable, and commercially successful while failing the person it was supposedly designed to serve.
The human being is not the final component in a technological pipeline. The human being is the reason the pipeline exists.
Optimization requires a human purpose
Every technology optimizes something. It may optimize prediction, engagement, adherence, throughput, cost, sensitivity, revenue, or convenience. Those objectives are not automatically aligned with human welfare.
A metric can improve while the lived outcome deteriorates. More monitoring can create anxiety without useful action. Greater sensitivity can produce false alarms. Increased engagement can become dependence. Faster decisions can remove reflection precisely where judgment is most necessary.
The central question is not what the system can do. It is what the system does to the human being.
This question must be asked before deployment, not after harm appears. It should shape the construct being measured, the data collected, the interface presented, and the decisions the system is permitted to influence.
The person is more than the data
Human technologies necessarily reduce complex lives to observable variables. This reduction can be scientifically useful, but it is never complete. A score does not contain the person. A risk category does not describe every relevant circumstance. A behavioral trace does not reveal intention. A physiological signal does not become identity merely because it is measured continuously.
Trouble begins when the representation is treated as the human being rather than as limited evidence about one aspect of that human being. The system becomes authoritative, and the person is asked to explain any disagreement with it.
The proper relationship is the reverse. The technology must explain the evidentiary basis, uncertainty, boundaries, and consequences of its output. Human experience should not be dismissed simply because it is inconvenient for the model.
Consent is not a checkbox
A person cannot meaningfully consent to a system they cannot understand. Dense terms of service, obscure data flows, and broad secondary-use permissions may satisfy a procedural requirement while defeating the purpose of consent.
Respect requires clarity about what is measured, why it is measured, who can access it, how long it is retained, which conclusions will be drawn, and what actions may follow. It also requires a genuine ability to refuse, withdraw, correct, and seek review without punishment or loss of essential access.
The more intimate the signal and the greater the consequence of the decision, the stronger these protections must become.
Performance must include those who carry the risk
Average accuracy can hide concentrated failure. A system may work well overall while performing poorly for a smaller population, an unusual physiology, a different environment, or a person whose condition was underrepresented during development.
Those failures are not statistical debris. They occur in actual people. A responsible evaluation therefore asks not only how often the system is right, but who experiences its errors, what those errors cost, and whether the people who carry the burden receive any corresponding benefit.
A human-centered technology should demonstrate:
- Purpose: a clearly defined human benefit rather than a vague claim of innovation.
- Validity: evidence that its measurements and interpretations represent the stated physiology or condition.
- Agency: the person can understand, question, refuse, and correct the system.
- Proportionality: data collection and intervention do not exceed what the intended benefit requires.
- Equity: performance and failure are examined across the people expected to use it.
- Accountability: responsibility remains identifiable when the technology causes error or harm.
Keep judgment attached to responsibility
Automation can support human judgment, but it can also scatter responsibility. The developer points to the data, the operator points to the model, the institution points to the vendor, and the vendor points to the user agreement. The person affected by the decision is left facing a system that appears to have no accountable author.
Human oversight must mean more than placing a person somewhere in the workflow. That person must have sufficient information, authority, and time to challenge the output. Review must be possible before a consequential action becomes irreversible.
Where a system influences health, opportunity, liberty, or access, accountability cannot be delegated to an algorithm. Decisions may be technologically assisted; responsibility remains human.
Innovation deserves a higher standard
Human-centered scrutiny is sometimes treated as resistance to progress. The opposite is true. A technology that cannot explain its purpose, validate its claims, disclose its uncertainty, or identify responsibility is not mature merely because it is novel.
The strongest innovation survives these questions. It becomes more precise about what it can do, more honest about what it cannot do, and safer in the hands of the people who use it.
JHST will examine human technology at this level. We will ask whether the biological construct is sound, whether the measurement is valid, whether the system improves a meaningful outcome, and whether the person retains dignity and agency throughout the process.
The human must remain at the center—not as a source of data, a user to be retained, or a problem to be optimized, but as the purpose against which the technology is judged.