Artificial Idea | AI careers · practical prompts · no hype Thursday, January 29, 2026 · Issue #51 · Prompt Tutorial
The feedback stack
How to use AI to give (and receive) better criticism
Most professional feedback is either too vague to act on or too direct to survive. These five prompts produce the kind that is neither.
Issue #50 made the case that the gap between your formal job description and your actual professional reality is a visibility problem with career consequences, and that closing it deliberately in the right direction is the mechanism by which AI capability development produces career return rather than just personal efficiency. This issue addresses the feedback infrastructure that tells you whether the closing is working.
Feedback is the most underutilised professional development resource available to most knowledge workers. Not because it is unavailable. Because the feedback that is available is almost never the feedback that is useful. The formal annual review produces carefully worded assessments calibrated to avoid conflict rather than accelerate development. The informal feedback from colleagues is filtered through relationship preservation instincts that soften every observation that might create awkwardness. The feedback from clients and senior stakeholders is either not sought at all or sought in forms so generic that the responses tell you nothing specific enough to act on.
The result is a professional operating largely on self-assessment, which the research on self-assessment accuracy consistently shows is the least reliable source of information about how your work lands with the people it is designed to land with.
AI does not fix the fundamental problem of feedback quality. It does something more modest and more immediately useful: it helps you design better feedback requests, process the feedback you receive more rigorously, deliver feedback to others more effectively, and simulate the feedback you are not receiving from the people whose assessment matters most.
Prompt 1: The feedback request designer
The problem it solves: designing feedback requests that produce specific, actionable responses rather than the generic positive assessments that most feedback requests elicit because they make it too easy for the respondent to be kind rather than useful.
Most feedback requests fail before they are sent because they ask the wrong questions in the wrong way. Questions like "what did you think of my presentation?" or "do you have any feedback on my proposal?" create an opening for a polite response rather than a useful one. The respondent fills the opening with the least uncomfortable content available, which is almost never the most useful content available.
You are helping me design a feedback request
that produces specific, honest, actionable
responses rather than polite generalities.
What I am seeking feedback on:
[describe the work, presentation, proposal,
project, or performance you want feedback on,
with enough context for someone unfamiliar
with it to understand what it was trying to achieve]
Who I am asking:
[describe their role, their relationship to
the work, what they observed or received,
and what their honest assessment would be
worth to me]
What I most need to know:
[be honest about what you are most uncertain
about or most want to improve, even if
you would not state this directly in the request]
Please design a feedback request that:
1. Opens with a specific question rather
than a general invitation, targeted at
the aspect of the work most worth
understanding better
2. Includes one question that makes it
easy to give critical feedback by
framing it as a normal part of the
evaluation rather than a departure from politeness
3. Includes one question about what
specifically worked, framed to produce
a specific answer rather than a general
positive assessment
4. Closes with a question about what
I should do differently next time,
specific enough that the respondent
cannot answer it with a generic suggestion
5. Is short enough that the respondent
can answer it in five minutes without
feeling they owe me a comprehensive review
The feedback request should make it
easier to be honest than to be kind.
Most feedback requests do the opposite
and produce the opposite result.
The instruction that the request should make honesty easier than kindness is the design principle most feedback requests violate. Kindness is the default because it is socially safer and requires less effort than specific critical observation. The only way to change the default is to design the request so that answering specifically is easier than answering generally, which means asking questions so specific that a vague positive response does not fit the question being asked.
Prompt 2: The received feedback processor
The problem it solves: extracting the maximum useful information from feedback you have already received, including feedback that was vaguely worded, diplomatically softened, or apparently contradictory.
Most professionals read feedback, form an immediate emotional response to it, and then either dismiss the parts that feel unfair or accept the parts that feel accurate without examining either reaction rigorously. Both responses leave significant useful information in the feedback unextracted.
You are helping me process professional feedback
I have received in a way that extracts
maximum useful information rather than
confirming my existing beliefs about
my own work.
The feedback I received: [paste or describe
the feedback as completely as possible,
including the exact wording where available]
The context: [what the feedback was about,
who gave it, what their relationship to
the work was, and what you know about
their standards and communication style]
My initial reaction to the feedback:
[describe honestly, including which parts
you agree with and which you find
unfair or inaccurate]
Please:
1. Identify the most specific and actionable
observation in this feedback, the one
most worth acting on regardless of
whether I agree with it
2. Identify any pattern in the feedback
that I might be minimising because
it is uncomfortable, and explain
why it might be worth taking seriously
3. Identify any part of the feedback
I might be over-weighting because
it confirms something I already believe
about myself, and whether that
self-assessment is well-founded
4. Translate any vague or diplomatically
softened observations into plain language:
what the feedback giver was most
likely trying to say without saying directly
5. Identify the single most important
change in my behaviour or output
that this feedback, taken seriously,
would produce
Do not tell me whether the feedback
is fair or accurate. Tell me what
is most useful in it regardless of
whether it is fair or accurate.
The instruction not to assess whether the feedback is fair is the constraint that makes this prompt most valuable. The fairness question is the one professionals spend the most time on and the one least relevant to professional development. Unfair feedback can contain useful signal. Fair feedback can contain nothing actionable. The prompt redirects attention from the evaluation of the feedback giver to the extraction of useful information from whatever they said.
Prompt 3: The feedback simulator
The problem it solves: generating a simulation of the honest feedback you are not receiving from the people whose assessment matters most, based on what you know about them and what they have observed of your work.
This is the prompt that most directly addresses the feedback gap described in the introduction. The senior stakeholder who has formed a view of your work but will never share it directly. The client who has decided not to renew but whose specific reasoning you will never hear. The manager who writes careful annual reviews but holds specific observations they have never delivered in a direct conversation.
You are simulating the honest professional
assessment of a specific person who has
observed my work but is unlikely to share
their genuine evaluation directly.
The person I am asking you to simulate:
[describe their role, their professional
standards, their communication style,
their relationship to my work, and
what they have directly observed]
What they have seen of my work:
[describe the specific interactions,
outputs, or performance they have
been exposed to]
What I know or suspect they think:
[describe any signals you have received,
however indirect, about their assessment]
What I most want to understand:
[what specific aspect of their assessment
would be most useful to you if you
could access it honestly]
Please simulate their honest assessment covering:
1. What they consider the strongest aspect
of my work based on what they have observed
2. What they consider the most significant
weakness or limitation based on the
same evidence
3. What they would say about me to a
colleague in a private conversation,
not in a formal feedback context
4. What would most change their assessment
in a positive direction if I were to
demonstrate it in the next ninety days
5. The question they have about me that
they have not asked directly and
that, if answered well, would most
advance their confidence in my capabilities
Flag where your simulation is speculative
rather than grounded in the information
I have provided. The useful output is
grounded simulation, not invented assessment.
The instruction to flag where the simulation is speculative is what makes this prompt honest rather than just interesting. A fully invented assessment has no value. An assessment grounded in specific observable information, with speculative elements clearly labelled, has significant value as a planning tool even when it cannot be verified. The professional who uses it knows which parts to act on as probable and which parts to hold as hypothetical.
Prompt 4: The feedback delivery designer
The problem it solves: designing feedback for someone else that is specific enough to be useful, honest enough to be worth giving, and delivered in a way that produces development rather than defensiveness.
Delivering feedback well is a capability that most professionals have never been explicitly taught and that most organisations have never systematically developed. The result is feedback that is either so softened it communicates nothing or so direct it produces defensiveness that closes the receiver to the observation being made.
You are helping me design feedback for
a specific professional that will be
useful rather than diplomatic and
honest rather than harsh.
The person I am giving feedback to:
[their role, their seniority relative to mine,
our working relationship, and their
likely response to direct criticism
based on what you know about them]
What I want to give feedback on:
[describe the work, behaviour, or
performance specifically, including
what happened, what the impact was,
and what you believe should change]
What I have tried before if anything:
[describe any previous feedback attempts
and their results]
What I most want to achieve:
[a specific behaviour change, a development
conversation, a performance correction,
or something else specific]
Please design feedback that:
1. Opens with the specific observation
rather than a general assessment,
describing what happened rather than
what it says about the person
2. Connects the observation to its impact
in terms the receiver cares about,
not just in terms of what matters to me
3. Makes the desired change specific
enough that the receiver knows exactly
what doing it differently would look like
4. Creates space for the receiver's
perspective without making the space
so large that it becomes a negotiation
about whether the feedback is valid
5. Closes with a forward-looking commitment
rather than a backward-looking verdict
Draft the opening three sentences of
this feedback conversation. The opening
determines whether the rest of it
lands as development or as judgment.
The instruction to draft the opening three sentences rather than the entire feedback script is deliberate. The opening is where most feedback conversations are won or lost, and it is the part most worth investing in. A feedback conversation that begins well has a significantly higher probability of producing the development outcome it was designed to produce than one that begins badly regardless of the quality of the content that follows.
Prompt 5: The 360 synthesiser
The problem it solves: synthesising feedback from multiple sources into a coherent developmental picture rather than treating each piece of feedback as an isolated data point that can be accepted, rejected, or averaged without examining what the pattern across all of it is saying.
You are helping me synthesise feedback
from multiple sources into a coherent
developmental picture.
Feedback sources and their content:
Source 1: [describe who and what they said]
Source 2: [describe who and what they said]
Source 3: [describe who and what they said]
Add additional sources as available.
My role and the professional context
in which this feedback was given: [describe]
My own assessment of my performance
in this period: [describe honestly]
Please:
1. Identify the themes that appear across
multiple sources, distinguishing between
themes that represent consistent signal
and those that may reflect a shared
blind spot in the feedback sources
2. Identify any significant divergence
between sources and what might explain it:
different exposure to different aspects
of my work, different standards,
or genuine disagreement about what
good looks like in my role
3. Identify the gap between my self-assessment
and the external assessments, specifically
where I rate myself higher than others do
and where I rate myself lower
4. Produce a single developmental priority
from this synthesis: the one change
that, if made, would most improve
my performance across the full range
of contexts the feedback covers
5. Identify the feedback source whose
assessment I am most tempted to dismiss
and make the case for why their
perspective might be the most valuable
one in this synthesis
The synthesis should produce one clear
priority rather than a list of everything
that could be improved. A developmental
list is not a developmental plan.
The closing instruction, that the synthesis should produce one priority rather than a list, is the constraint that makes the output actionable rather than overwhelming. Most 360 feedback processes produce a comprehensive report that identifies everything worth developing, which is a different document from one that tells you what to work on next. Development that is spread across ten priorities simultaneously is development that advances on none of them.
The feedback practice
These five prompts are most valuable when used as a recurring practice rather than a one-time exercise. The professionals who reach the career inflection point described in Issue #47 most quickly are not those who develop the fastest in isolation. They are those who have the most accurate and current information about how their work is landing with the people who matter most, because that information directs their development toward the gaps that are actually limiting their career rather than the gaps they have assumed are limiting it.
The feedback practice does not need to be elaborate. Running Prompt 2 after every significant piece of feedback received, running Prompt 3 quarterly for the two or three stakeholders whose assessment matters most, and running Prompt 5 before every performance conversation is a practice that takes less than two hours per quarter and produces a consistently more accurate picture of how your professional work is perceived than the informal, filtered, diplomatically softened feedback most professionals are currently working from.
More accurate information produces better development decisions. Better development decisions produce faster progress toward the career inflection point. The practice is the infrastructure that connects those two things.
Monday we are examining the Block layoffs in detail: what four thousand eliminated roles actually had in common, what was kept and why, and what the decision criteria that produced that specific pattern reveal about how AI-driven restructuring actually works inside organisations doing it seriously rather than using AI as a cover for cost cutting they wanted to do anyway.
The headline number is the least interesting thing about that announcement. The pattern underneath it is the most instructive data point on AI-driven restructuring published so far.
— Team Artificial Idea
