How to Get the Most Out of AI - Part 3 in the AI at Work Series
If you've ever used AI and felt underwhelmed by the results, the issue might not be the technology—it could be how you're communicating with it. AI isn’t a mind reader. It works best when given clear, structured instructions and thoughtful feedback.
Rather than treating AI like a magic box that delivers perfect answers, a better approach is to view it as a collaborator—one that improves the more you refine how you interact with it.
Why AI Needs Better Instructions Than You Think
AI models generate responses based on patterns and probabilities. If you ask a vague question, you’ll get a generic answer. If you provide detailed instructions, context, and follow-up feedback, you’ll get significantly better results.
Think of AI like an intern. If you tell a new intern, “Write a report on industry trends,” you might get something broad and unfocused. But if you say, “Summarise the top three trends in sustainable manufacturing from the past year, using examples from published reports,” you’ll get a more useful response.
The same applies to AI. The more specific and structured your prompt, the better the output.
The Role of Context: Giving AI the Right Information
One of the biggest mistakes people make when using AI is assuming it understands the full context of their request. AI doesn’t automatically know what you’re working on, your industry, or your expectations—unless you tell it.
Before asking for a response, consider:
What background does AI need? If you're working on a report, should AI be aware of previous sections or key themes?
What format do you want? A summary, a bullet-point list, a detailed analysis?
What tone or style is appropriate? Formal, conversational, analytical?
Who is the audience? A board of executives, a general reader, or a technical team?
The more of this information you include, the more relevant the response will be.
Chain of Thought: Getting AI to Think in Steps
AI doesn’t have an internal monologue—it doesn’t "think" before answering. It simply generates the most likely response based on its training. However, you can guide it to be more thoughtful by prompting it to break problems into steps.
For example, if you're using AI to help draft a research paper, instead of asking:
“Write a paper on digital transformation in finance.”
Try:
“Step 1: Outline the key sections of a paper on digital transformation in finance.
Step 2: Provide three main arguments with supporting evidence.
Step 3: Draft the introduction based on these findings.”
This forces AI to structure its response more logically, leading to better results.
Few-Shot Learning: Showing AI What Good Looks Like
AI learns from examples. One of the best ways to improve its responses is to provide a model answer.
If you need AI to write in a particular style, you can say:
“Write a summary of this report in the style of a professional business analyst. Here’s an example of the type of summary I want: [insert sample].”
This technique, known as few-shot learning, helps AI understand your expectations and tailor its output accordingly.
Iterate, Iterate, Iterate—The Back-and-Forth Process
Getting the best AI responses is rarely a one-step process. The most effective way to use AI is through a feedback loop.
If the first response isn’t quite right, don’t discard it—refine it:
“Make this more concise.”
“Use simpler language.”
“Add a real-world example.”
“Summarise this into five key takeaways.”
Each iteration sharpens the final result. AI improves with guidance, just like a human collaborator.
Giving AI a Persona: Making It Work Better for You
One unusual but effective strategy is assigning AI a role or personality. Research has shown that AI performs better when given a clear identity.
For example, instead of asking:
“Explain machine learning.”
Try:
“You are a teacher who specialises in AI in the Travel Industry. Explain machine learning in simple terms suitable for a business audience.”
This subtle shift can change the depth, tone, and focus of the response, making it more aligned with what you actually need.
Knowing AI’s Limits: When to Trust It—and When Not To
AI can be a powerful tool, but it isn’t perfect. It still makes errors, fabricates information, and struggles with certain tasks.
To use it effectively, ask yourself:
Would I trust this response if a person gave it to me?
Can I verify the information?
Does it match my own expertise in this area?
AI works best as an amplifier of human judgment, not a replacement for it.
Bringing It All Together: How to Use AI Effectively
To summarise, getting the most out of AI requires:
- Clear, structured instructions—Give it a specific task, not a vague question.
- Context and examples—Help AI understand your expectations.
- Step-by-step thinking—Guide AI to build its answer logically.
- Iteration and feedback—Refine and improve responses over time.
- A defined persona—Tell AI who it is and what role it’s playing.
The more you treat AI as a collaborator rather than a passive tool, the better the results will be.
In Part 4, we’ll explore the next stage of AI integration—how these tools are evolving and what that means for the way we work and think.