---
name: adopt-ai-jobs-answer
description: Prepares an honest answer to what people ask about AI and their jobs, with the questions sorted, a decided, open and unknown table, a can and cannot promise list and the named leader who gives the answer. Use for "run adopt-ai-jobs-answer", "will AI replace my job", "how do I answer the jobs question", "AI and job security answer", "what do we tell staff about AI and jobs", "FAQ on AI and jobs", "people are afraid AI will replace them", "the trainer keeps getting asked about layoffs", part of the Claude Guide for AI Adoption Pack by Polar Bear.
---

# AI and Jobs Answer

## When To Use
"Will this replace my job?" is in the room at every session, and the trainer has been left to answer it. Use this when the question keeps coming back and the people who could answer it have not. It answers: what can honestly be said today, who says it, and what happens to the next question?

## When Not To Use
Do not use it to write reassurance for a leader who has decided nothing and will not say so: the output would be an empty page with a signature. To follow single concerns and the commitments made about them, run Concern Response Log. For what people can still influence in the rollout, run Team Involvement Plan.

## Inputs
- The questions people are asking about AI and their work, as asked, without names.
- What has actually been decided, by whom and when, what is still open, and which leader has the authority to answer.
- Anything already said in public about AI and jobs, so the answer does not contradict it.
If you have none of this, I start from the questions you have heard and mark the output as a first draft.

## Approach
The method is plain and has no owner: say what is known, what is not known and what is being done about it, early, and from someone with the authority to say it. An honest "we have not decided" holds up longer than a warm sentence nobody can keep. One survey of workers in one country, fielded in 2024, found more people worried than hopeful about AI at work, so assume the question is in the room even when nobody asks it. The failure this prevents: a trainer improvises "nobody is losing their job" to get through a session, a later decision makes it untrue, and every session after that is taught by someone people no longer believe.

## Workflow
1. Ask up to three questions: what people are asking and where, what has been decided and by whom, and which leader has the authority to answer.
2. List the questions as asked, without names. Sort the questions into five groups: my job, my tasks, my skills, my pay or grading, who decides. Sort questions, never people: no note of who asked, and nobody is tagged as anxious or resistant.
3. Give each question three cells. Decided, with who decided and when. Open, with who will decide and by when. Unknown, said plainly. I fill them only from what the leader supplied and mark every gap; a gap is never closed with a guess about jobs.
4. Draft the "can and cannot promise" list for the leader to correct. Test each line: can this leader keep it, and for how long? "I cannot promise that, and here is why" is an answer.
5. Set who gives the answer, where and when: a named leader with the authority, never the trainer alone. Write the trainer's one-line handover ("that question has an owner, here is where the answer is").
6. Set the route for follow-up questions and the date of the next update, including an update that says "nothing new yet": silence between updates gets read as bad news. Mark every employment, consultation or redundancy point "check with a qualified adviser" and hand the draft to the leader, who edits it and delivers it in person.

## Output Format
```markdown
# AI and Jobs Answer
Prepared for: [leader, by role] | Draft date: [date] | Status: [draft for the leader to correct]
## The questions people are asking
| Group | Question, as asked | Where it came up |
|---|---|---|
| [my job / my tasks / my skills / my pay or grading / who decides] | [question] | [session or channel, no names] |

## Decided, open, unknown
| Question | Decided (who, when) | Open (who decides, by when) | Unknown |
|---|---|---|---|
| [question] | [decision, or nothing yet] | [owner and date] | [what nobody knows today] |

## What we can and cannot promise
| We can say | We cannot promise, and why | Adviser check |
|---|---|---|
| [statement the leader can keep] | [statement and reason] | [no, or check with a qualified adviser] |

## Who answers, and what happens to the next question
- Answer given by: [named leader], at [meeting or channel], on [date].
- Trainer's handover line: [one sentence that points to the owner and the place].
- Gaps still to fill before this is said out loud: [list].
- Follow-up questions go to [channel, with a private option]; [role] replies; next update on [date], even if there is nothing new.
## Decision
[Named leader] corrects and approves this answer by [date], and decides when and where to give it.
```

## Done When
- Every question has all three cells filled or marked as a gap for the leader.
- No sentence promises something about jobs that the organisation has not decided.
- A named leader, a place and a date are set, the trainer has a handover line, and follow-up questions have a route and a next update date.

## Quality Bar
- The questions are in people's own words and carry no names.
- "Unknown" is written as unknown, not dressed up as "under review".
- Employment, consultation and redundancy points end with "check with a qualified adviser".
- A named leader gives the answer; the document promises nothing about jobs that the organisation has not decided.

## Next
Run adopt-ai-concern-statement (AI Concern Statement) to help one person put a specific concern into words.

## About the makers
This pack is made by Polar Bear, a consultancy built by ex-McKinsey founders with a dream to make AI work for People, not instead of them. We help our clients build people systems and AI-first ways of working, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).
