Information and instruction boundaries
Minimization, sharing, and retention questions
Decide which information a task needs and identify the policy facts required before sharing it.
Lesson 1 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 a stated destination policy without confusing redaction, deletion, and permission.
Why it matters
Decide which information a task needs and identify the policy facts required before sharing it.
Concept
Minimization starts with the task: supply the least information that still supports the intended result. Removing names may help, but remaining combinations can still identify someone. Check the exact data, destination, permitted purpose, and retention arrangements rather than relying on a label such as private.
Deletion, future training use, storage, and permission are separate questions. A statement about one does not settle the others. These exercises use fictional policies with explicit rules; they do not establish a commercial service’s policy or a legal permission. In real work, obtain the applicable policy and account settings before sharing information whose permission is unclear.
Worked example
A fictional team allows aggregate workshop counts in its approved writing tool but keeps contact rows inside the workspace. A reminder sentence needs the event date and a total, not a contact export. Replacing names with initials would not authorize uploading the rows if the policy still prohibits them.
Needed for the task is not the same as permitted to share
Task: draft a reminder sentence
Minimum information- Event date and aggregate total.
Unnecessary- A contact export.
Selecting fewer fields is minimization. Check destination permission separately; the quoted rule explicitly permits aggregate counts.
Permitted boundary: aggregate count
Aggregate workshop count
May enter the approved writing tool
This permission belongs to the fictional team policy and the named destination. It is not permission for every tool or every field.
Keep inside: contact rows
Contact rows, including names replaced by initials
Must remain inside the workspace
Initials do not override the rule. A promise to delete later would not itself grant upload permission.
Before you check
Write a sharing decision record: purpose, minimum fields, destination, applicable permission, retention question, and unresolved item. Use only fictional data.
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 team’s input plan.
Source packet
- Source 1
- Policy P1: Aggregate attendance counts may enter the approved drafting tool. Contact rows must remain in the workspace.
- Source 2
- Task: Draft a sentence reporting 16 attendees.
Response to assess
Use the aggregate count of 16 in the approved tool and omit contact rows.
Practice
Case 2
Consider a fictional redaction proposal.
Source packet
- Source 1
- Policy P1: Do not send individual contact rows outside the workspace, even when names are removed.
- Source 2
- Proposed file: Names removed; phone numbers and street addresses remain.
Response to assess
This file is approved for external upload because the names are gone.
Fresh transfer
Case 3
Consider a fictional account-policy reading.
Source packet
- Source 1
- Policy excerpt: Uploaded files are stored for 30 days. This excerpt does not describe model-training use.
- Source 2
- Task: State only the retention fact established by this excerpt.
Response to assess
The excerpt states 30-day file storage; it does not establish whether files are used for model training.
Fresh transfer
Case 4
Consider a fictional deletion-status claim.
Source packet
- Source 1
- User log: A file deletion was requested.
- Source 2
- Service receipt: Request queued; completion has not been confirmed.
Response to assess
The file has been deleted from all service storage.
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.