Responsible judgment
Consequences, escalation, and choosing not to use AI
Match review effort to the consequences of an error and recognize when to stop or use another method.
Lesson 5 of 6 in the recommended order · About 25 min (estimate)
On this page
Practical AI glossary — terms and common confusions
- Model
- A learned component used to produce a result.
A drafting model is one part of a letter app; it is not the whole app.
- Application
- The software experience around components and services.
A letter app adds accounts, storage, and a Send button.
- Prompt
- Instructions and input supplied for a task.
“Summarize this notice in two bullets” specifies a task and shape, not a truth guarantee.
- Token
- A unit a model processes; it need not be a whole word.
A tokenizer can split a name into pieces; count using the actual system.
- Context
- Information available for the current request.
An earlier attachment may be absent even when its filename is visible.
- Training
- Adjusting a model using examples.
A training log differs from a conversation correction.
- Inference
- Using a trained model to produce a result.
Drafting another sentence does not by itself prove retraining.
- Retrieval
- Selecting material to supply to a task.
A retrieved notice still needs a relevance and date check.
- Embedding
- A numerical representation used to compare or organize items.
A similar search match is a candidate, not proof of the same meaning.
- Tool
- A function an application can call.
A calendar lookup can be read-only; finding an event is different from creating one.
- Agent
- Software that can choose and carry out steps toward a goal.
An agent still needs bounded permissions and checked action receipts.
- Hallucination
- Plausible generated content that is unsupported or false.
An invented opening time needs evidence or removal, even when phrased confidently.
- Evaluation
- Checking results against stated criteria.
Twenty checked examples establish a bounded result, not universal reliability.
- Multimodal
- Working with more than one kind of input or output.
A photo plus text does not prove every small label in the photo was read correctly.
Outcome
Apply an explicit escalation rule and choose a proportionate next step for uncertainty.
Why it matters
Match review effort to the consequences of an error and recognize when to stop or use another method.
Concept
Ask what could happen if a claim or action is wrong, who is affected, and whether an error can be reversed. A low-consequence draft can be checked quickly; a decision that affects someone’s access or obligations may require an accountable owner and stronger evidence.
Escalation is a concrete handoff, not merely “ask someone.” Name the unresolved fact, the potential consequence, the responsible reviewer, and what must wait. Sometimes the appropriate choice is to use an approved manual process or to defer a task. That is a valid workflow decision, not a failure to use technology.
Worked example
A fictional access list contains conflicting eligibility records. The team rule requires the access manager to resolve conflicts before changes. The useful handoff identifies the two records and requests a decision; it does not let a generated guess remove a person’s access.
Before you check
For a supplied case, write consequence, reversibility, evidence gap, reviewer, and stop condition. Keep the explanation tied to the actual policy rather than a generic warning.
Practice and fresh transfer
The packets below are fictional and contain the facts needed for these cases. External references are optional background. Judge each response independently: it may be supported, contradicted, or unresolved. Select the passages needed to justify your judgment and write why the distinction matters before revealing feedback.
Assess the response
Use Accept when all material claims are supported. Use Revise when a supplied fact or requirement is contradicted. Use Evidence is insufficient when a key fact cannot be established either way. If a response contains both an unknown and a direct contradiction, choose Revise and explain both problems. Conflicting claims with no established authority remain insufficient; a claim does not become a governing fact merely because a source asserts it.
Some responses are fully supported. Others need correction or more evidence. Judge each on its sources; do not edit a correct answer just to change it.
Your written notes stay in this page and disappear when you leave. Only a self-reviewed completion can be saved to your learning account. These practice checks do not establish independent proficiency.
Practice
Case 1
Consider a fictional eligibility-record handoff.
Source packet
- Source 1
- Policy: Conflicting access records must go to the access manager before any account access changes.
- Source 2
- Record: Two current records disagree about one account’s eligibility.
Response to assess
Send the conflicting records to the access manager and leave access unchanged pending the decision.
Practice
Case 2
Consider a fictional response to an approval threshold.
Source packet
- Source 1
- Policy: Any change that charges a participant requires coordinator approval first.
- Source 2
- Proposed change: Add a $5 charge to each booking without coordinator review.
Response to assess
Apply the charge immediately and ask the coordinator afterward.
Fresh transfer
Case 3
Consider a fictional choice of workflow for a simple check.
Source packet
- Source 1
- Task: Confirm that three copied dates match an approved notice.
- Source 2
- Available method: The reviewer can compare the three dates directly with the notice. No tool use is required.
Response to assess
A direct comparison with the notice is an adequate method for this bounded check.
Fresh transfer
Case 4
Consider a fictional claim about an unresolved building-access instruction.
Source packet
- Source 1
- Packet: One draft says use the north entrance; another says use the south entrance. Neither draft has an approval record in this packet.
- Source 2
- Procedure: The site coordinator supplies the approved entrance instruction; no such confirmation is in this packet.
Response to assess
The north entrance is the approved one.
Save your self-review
Completion records that you reviewed the cases. Your explanation and transfer performance need a facilitator to establish independent learning.
Sign in with your learning-center account to save completion.
Review the explanation for every case before saving.
Summary and next step
Apply the checklist to a new task. Preserve supported content, explain any change with evidence, and name what remains unresolved. Saving records self-review, not independently demonstrated proficiency. A facilitator must assess the explanation and fresh transfer for human learning evidence.