When NOT to Use AI — Foundations Lesson 8

Scenario

You're busy. There's a stack of work on your desk. You've been using AI to speed things up, and it's been helpful. But then you catch yourself about to use AI for something that feels wrong. Firing an employee remotely with a template from ChatGPT? Handling a sensitive customer complaint with an AI-written response? Deciding whether to promote someone based on AI analysis? You stop and think: Just because I can use AI doesn't mean I should. You need a framework for saying "No, this needs a human."

---

Learning Objectives

By the end of this lesson, you will be able to:

1. Identify tasks unsuitable for AI — where human judgment is essential

2. Recognize emotional and relational red flags — situations requiring presence and empathy

3. Understand the difference between delegating to AI and avoiding responsibility

4. Apply a decision framework to know when humans must lead

5. Communicate your boundaries about AI use to colleagues

---

Think Before You Delegate

Before you hand a task to AI, ask:

Would this person feel hurt if they knew AI handled this instead of a human?

Am I avoiding responsibility by using AI?

Does this task require judgment calls only a human can make?

Is there a relationship at stake?

Could this decision affect someone's life or livelihood?

Would I be comfortable defending this choice publicly?

If you answer "yes" to any of these, don't use AI.

---

Categories of Tasks That Shouldn't Be AI-Handled

Category 1: Decisions That Affect People's Lives/Jobs

What it includes:

Hiring or firing decisions

Promotions or disciplinary action

Salary or benefit decisions

Assignment to roles

Performance evaluations

Medical or mental health recommendations

Legal advice

Career counseling

Why not AI:

These decisions require judgment about *people*, not patterns

AI can't understand context (relationship history, individual circumstances, company politics)

These decisions carry accountability — you must own them

People deserve human judgment, especially when stakes are high

Bias risk is extreme (AI perpetuates past discrimination)

What AI can do instead:

Summarize information for your review

Organize data (performance metrics, credentials, etc.)

Suggest discussion questions for the conversation

Draft the communication

Research industry standards for comparison

Example:

❌ "Use AI to screen resumes and pick the top 10 to interview"

✅ "Use AI to organize resumes by key skills. I'll review the organized list and make interview decisions."

---

Category 2: Sensitive Communications (Especially Bad News)

What it includes:

Firing someone

Delivering bad news

Addressing conflict

Expressing disappointment or criticism

Consolation or sympathy

Complex negotiations

Personal apologies

Why not AI:

These require emotional intelligence and genuine care

People can feel the difference between human and AI-generated words

Tone and empathy matter more than content

AI sounds robotic in sensitive contexts (even when polished)

Recipients deserve genuine human attention

What AI can do instead:

Help you think through what to say

Suggest communication approaches

Draft talking points you'll personalize

Help you prepare for the conversation

Review your response for tone

Example:

❌ "Use ChatGPT to draft a firing letter; send it as-is"

✅ "Use ChatGPT to organize your thoughts about why this decision makes sense. Draft your own letter. Have ChatGPT review tone before you send."

---

Category 3: Situations Requiring Empathy or Genuine Understanding

What it includes:

Customer service for frustrated/upset customers

Addressing a team member's personal problem

Mentoring or coaching

Acknowledging someone's achievement personally

Building relationships

Complex problem-solving requiring understanding

Why not AI:

AI can't truly empathize; it mimics empathy

People can feel the difference (especially in high-stakes situations)

Relationships build on genuine human attention

Understanding requires more than pattern-matching

What AI can do instead:

Suggest conversation approaches

Help you organize your thoughts

Draft templates you'll personalize

Brainstorm ideas before your conversation

Prepare talking points

Example:

❌ "A customer is upset. Have AI generate a response; send it."

✅ "A customer is upset. Understand their issue personally. Use AI to help you think through solutions, then write a genuine response."

---

Category 4: Creative or Strategic Work Requiring Your Perspective

What it includes:

Company strategy or vision

Brand voice or messaging

Organizational culture decisions

Long-term planning

Innovation or new initiatives

Setting precedent

Defining values or principles

Why not AI:

These require *your* judgment and perspective

AI remixes existing patterns; it doesn't innovate

Your strategy should reflect your values, not AI patterns

You're abdicating responsibility if you let AI decide direction

What AI can do instead:

Brainstorm options for you to evaluate

Organize pros/cons of different approaches

Research what others do in similar situations

Draft possibilities based on your direction

Help refine ideas you've generated

Example:

❌ "Use AI to develop our company's 5-year strategy"

✅ "I've decided our strategic priorities are X, Y, Z. Use AI to brainstorm how we execute each. I'll choose the approach and refine details."

---

Category 5: Situations Where Your Presence or Attention Matters

What it includes:

First-time interactions with someone

Building relationships

Showing you care about someone's development

Demonstrating commitment to a client or colleague

Being present for someone in crisis

Celebrating accomplishments

Why not AI:

The *fact that you personally handled it* sends a message

Delegating to AI says: "This isn't important enough for my time"

Relationships require your actual presence and attention

What AI can do instead:

Help you remember details about the person

Suggest personalization touches

Draft things you'll make personal

Organize information to help you be more present

Example:

❌ "Use AI to write a welcome email for new team member; send as-is"

✅ "Write a personal welcome email from you. Use AI to help organize team info you want to include, or brainstorm warm touches."

---

The "Responsibility Test"

When you're unsure, ask: "Can I defend this choice if questioned?"

Scenario 1: Using AI to draft emails

"I asked AI for initial language, then personalized it and sent my own version."

✅ Defensible. You took responsibility.

Scenario 2: Using AI to filter job applicants

"I used AI to screen resumes, and I only interviewed the top candidates it recommended."

❌ Not defensible. AI made the decision, not you.

Scenario 3: Using AI to handle a customer complaint

"I read the complaint, understood the issue, used AI to help with tone, then sent my own response."

✅ Defensible. You handled it; AI assisted.

Scenario 4: Using AI to write a performance review

"I asked AI to draft a performance review based on metrics I provided."

❌ Not defensible. You're not making the judgment.

Rule: If you can't defend the choice with "I made the decision and AI helped," then you're probably using AI wrong.

---

Red Flags: Stop and Don't Use AI

These are immediate red flags. If you see any of these, stop.

🚩 **"This will save me time"** — if that's the *only* reason you're using AI, reconsider. Efficiency isn't always good.

🚩 **"I don't have to think about this anymore"** — if using AI means you're not engaging with something important, don't.

🚩 **"The person won't know the difference"** — if you're hiding that AI helped, that's usually a sign you shouldn't be using it.

🚩 **"This is just a template, no one will notice"** — templates for important communications should be personalized, not sent as-is.

🚩 **"AI is better at this than I am"** — for high-stakes judgments, you need to lead, not AI.

🚩 **"I'm too busy to handle this personally"** — "too busy" isn't an excuse to offload responsibility to AI.

🚩 **"This impacts people I don't know well"** — the less you know someone, the more personal attention they deserve, not less.

---

Decision Framework: Should I Use AI?

Step 1: What's the task?

[Describe it]

Step 2: Who's affected?

[List people who'll be impacted]

Step 3: Is this a judgment call or a routine task?

If judgment → Continue to Step 4

If routine → Probably OK to use AI

Step 4: Would I be comfortable explaining my AI use to affected people?

If yes → Continue

If no → Don't use AI

Step 5: Is AI replacing my judgment or supporting it?

If replacing → Don't use AI

If supporting → Continue

Step 6: Could I have done this better without AI?

If yes → Don't use AI

If no → Probably OK to use AI

Step 7: Am I avoiding responsibility?

If yes → Don't use AI

If no → Proceed

If you pass all seven steps, using AI is probably responsible.

---

When to Over-Communicate Your Human Involvement

Some tasks don't need AI. But if you're tempted to use it, you might need to show that you're handling it yourself.

Example situations:

Situation 1: First client meeting

Don't use AI-written agenda or notes. Doing it yourself signals: "I'm invested in this relationship."

Situation 2: Responding to criticism

Don't use AI-written response. Writing it yourself shows: "I care enough to address this personally."

Situation 3: Major life decision for someone (promotion, feedback)

Don't use AI drafts. Writing personally shows: "This matters to me enough to give it real thought."

The pattern: When you want to show someone they matter, handle it yourself.

---

Mission: Identify Your AI Boundaries

The Task

1. List 5-10 tasks you currently do (or might delegate to AI)

2. For each task, assess: Should humans lead or can AI support?

3. Document your boundaries — What are you not comfortable delegating to AI?

Example tasks:

Writing emails

Handling customer feedback

Making hiring decisions

Strategic planning

Performance reviews

Onboarding new employees

Addressing conflict

Brainstorming

Summarizing documents

Organizing data

---

Deliverables

AI Boundaries Document:

`

Task: [Task name]

Human-led or AI-supported? [Primarily human OR Supported by AI]

Why? [Your reasoning]

What does AI do? [How AI helps, if at all]

What do YOU do? [Your responsibility]

Example:

Task: Performance review

Human-led or AI-supported? Primarily human

Why? This is about judgment and relationship

What does AI do? Organize metrics; suggest discussion topics

What do YOU do? Make the judgment; write the review; deliver personally

`

Document 5-10 tasks. This is your personal AI boundaries framework.

---

AI Coach: Efficiency Isn't Everything

Here's the cultural pressure you'll face: "We need to be more efficient. Use AI for everything."

But here's the uncomfortable truth: Some things shouldn't be efficient.

Firing someone shouldn't be optimized. Mentoring shouldn't be scaled. Relationships shouldn't be automated.

The professionals who stay in demand aren't the ones with the best automation. They're the ones who know when to not automate. They're the ones who show up personally when it matters.

Key insight: Using AI for everything signals "I'm busy and everything's equally urgent." Choosing to handle some things personally signals "I know what matters most."

---

Reflection: What's Your Professional Line?

Every person has tasks they won't delegate, even to AI. These often reveal what you value.

Reflect:

What tasks feel important enough to handle yourself?

What would feel wrong if an AI handled it?

What's the difference between "I could use AI" and "I should use AI"?

Write 3-4 sentences about what tasks feel too important to automate.

---

Portfolio Check

Portfolio principle: Your portfolio should show thoughtful AI use, not maximum automation.

This means:

Some pieces show AI support (draft + your refinement)

Some pieces show pure human work (no AI)

You can explain the difference

You chose carefully for each piece

Building this habit now positions you as someone who uses AI well, not just frequently.

---

Bonus Challenge

Observation: Watch someone you respect at work. How do they use (or not use) technology to handle important things? What's their pattern? What can you learn from how they prioritize human vs. automated work?

---

Key Takeaways

Don't use AI for: life-changing decisions, sensitive communications, empathy-requiring situations, strategic/creative work, situations requiring your presence

Red flags: saving time is the only reason, hiding AI use, not thinking about it, impact on people you don't know

Decision framework: Who's affected? Judgment call? Would you explain? Replacing or supporting? Could do better without? Avoiding responsibility?

Efficiency isn't everything; some work should stay human

Your boundaries reveal what you value

Next: Lesson 9 explores Context & Conversation Design — using conversation history and iteration to build better AI workflows.