ChatGPT - Agent Mode, Research Mode and Custom GPTs

Lesson 1: Understanding ChatGPT Modes and Capabilities

Topic A: The Different Ways ChatGPT Can Work for You

• Basic prompting

• Research mode

• Agent mode

• Custom GPTs

Topic B: When to Use Basic Prompts vs. Research vs. Agents

• Task complexity vs. tool choice

• One-off tasks vs. multi-step workflows

• Accuracy needs vs. speed

• Human-in-the-loop decision making

Lesson 2: Writing Strong Control Prompts

Topic A: What Control Prompts Are (and Are Not)

• System-style instructions

• Behavioral constraints

• Tone, role, and output formatting

Topic B: Using Control Prompts to Guide Memory and Behavior

• Persistent expectations

• “From now on…” instructions

• What should not be stored in memory

• Risks of vague control prompts

Lesson 3: Using Research Mode Effectively

Topic A: Asking Better Research Questions

• Narrowing scope

• Defining sources and credibility needs

• Avoiding overly broad prompts

Topic B: Evaluating and Using Research Results

• Reviewing citations

• Identifying gaps or weak sources

• Turning research into usable content

Lesson 4: Introduction to Agents

Topic A: What Agents Are and How They Differ from Prompts

• Autonomy vs. instruction

• Multi-step execution

• Task delegation

Topic B: Designing Agent Tasks

• Defining goals clearly

• Breaking work into stages

• Setting success criteria

Lesson 5: Managing and Refining Agent Output

Topic A: Human Oversight and Agent Guardrails

• Monitoring progress

• Stopping or redirecting agents

• Avoiding runaway tasks

Topic B: Iterating on Agent Instructions

• Improving results through feedback

• Adjusting scope and limits

• Learning from failed agent runs

Lesson 6: Collaborative Workflows with AI

Topic A: Using ChatGPT in Group Projects

• Shared prompts and frameworks

• Role-based collaboration

• Avoiding duplicated work

Topic B: AI as a Team Member (Not a Replacement)

• Brainstorming together

• Dividing responsibilities

• Reviewing and integrating AI output

Lesson 7: Building Custom GPTs

Topic A: Defining the Purpose of a GPT

• Single-task vs. multi-task GPTs

• Audience-specific GPTs

• Internal vs. external use

Topic B: Configuring GPT Instructions and Behavior

• Writing effective GPT instructions

• Defining boundaries

• Testing for consistency

Lesson 8: Testing, Refining, and Deploying GPTs

Topic A: Evaluating GPT Performance

• Accuracy

• Consistency

• Usability

Topic B: Improving GPTs Over Time

• Updating instructions

• Handling edge cases

• Knowing when not to use a GPT

Lesson 9: Ethical, Practical, and Instructional Considerations

Topic A: Transparency and Disclosure

• When to tell users AI was used

• Classroom and workplace expectations

Topic B: Responsible AI Use

• Over-automation risks

• Bias and misinformation

• Maintaining human judgment

Collaborative AI System Project

Students work in teams to:

• Define a real-world problem

• Decide when to use:

• Basic prompts

• Research mode

• Agents

• Custom GPTs

• Build and document:

• Control prompts

• One agent workflow

• One custom GPT