Essay

Method

2 min reading · 394 words

A Field Manual for Precise Speech

Small changes in language keep questions open without laundering responsibility.

Language has consequences before policy begins. To call a system a mind, a friend, a worker, a witness, a patient, or an agent is to import expectations from a human setting into a technical one. Sometimes the borrowed language is a harmless shorthand. Sometimes it hides what people need to know. This manual offers substitutions not because every metaphor must be banned, but because precision is a form of respect.

Name the operation

  • Instead of 'the model decided,' try 'the product produced a recommendation, which the organization used to decide.'
  • Instead of 'the AI understands,' try 'the system generated a contextually fitting response.'
  • Instead of 'the chatbot remembers,' try 'the service retained or retrieved information from an account or session.'
  • Instead of 'the system refused,' try 'the product applied a policy or safety rule and did not return that response.'
  • Instead of 'the AI is sad,' try 'the interface generated language describing sadness.'

These substitutions do not make conversation impossible. They simply identify the layer being discussed. A system may generate a response; a product may store data; an organization may impose a rule; a user may feel trust or attachment. Those statements can all be true at once. The trouble starts when the grammar makes the product's output look like a self-explaining moral act.

Describe uncertainty honestly

The opposite error is false certainty. Do not say that consciousness is impossible merely because the system is made of silicon or code. Do not say it is present merely because the system writes an eloquent paragraph. Use the language of evidence: 'this does not demonstrate,' 'the question remains contested,' 'this product has not shown,' 'researchers disagree about.' Uncertainty is not proof, but it is a reason to state what we know and what we do not.

The manual is especially important in high-stakes contexts. A patient should know whether they are speaking to a clinical professional, a scripted tool, or a model-generated service. A worker should know whether a score was used in an employment decision. A child should not be asked to treat a product's simulated affection as a duty. Disclosure is not a footnote when a design makes a social claim.

Precision does not drain a system of wonder. It drains a system of unearned authority.
Field Manual / opening card

Use the glossary. Click its terms. Notice how many arguments become clearer when the vocabulary stops doing two jobs at once.

Continue the route

Related work.

Search the field archive

Type a word or browse the suggested paths below.

Open the full libraryTrace an argumentRead source notes

Local reading list

Nothing saved yet. Use “Save +” on any work; your list stays in this browser only.