A starting point for discussion
The idea
In the author’s example about tax enforcement, a human agent must request or demand that an algorithm select people or cases for a tax audit. Another system could sort data continuously and issue recommendations, arrest warrants, or court summonses for human agents to accept or reject. In this example, a system with no human involvement would leave everything to machines.
Place in the reasoning
AI assessment · Connected to reasoning
Maps command, review, and no-human-involvement cases onto a tax-enforcement example.
Computer systems can shape what people do without taking over
Concretely illustrates different degrees of human involvement in algorithmic activity, without implying that every system is fully automated.
What this means and why it matters
What this means
In the first example, a human agent must request or demand that the algorithm select people or cases for a tax audit. In the second, the system works on its own and can issue recommendations, arrest warrants, or court summonses for human agents to accept or reject. The final category leaves everything to machines.
Why it matters
The example distinguishes a required human command from a person’s later choice about an automated suggestion.
Illustrative example
A hypothetical tax office could select audit targets with an algorithm only after an agent requests or demands that selection.
Conditions & assumptions
Source coverage: Full readable text of the author manuscript from an accessible mirror; not the version formatted by the publisher. The passage maps the categories onto a tax-enforcement example and specifies what a human command, human review, and full automation could mean. Addresses: The practical meaning of the human-control categories
Questions this connects to
AI-suggested relationships. Authors have not necessarily responded to one another.