Get regular updates on AI strategies that work »

Get the Most Out of ChatGPT

Ever feel like you’re not getting the most out of your AI conversations?

You’re not alone! Terms like zero-shot, one-shot, and few-shot might sound technical, but they’re key to mastering AI prompts.

Why This Matters

You might be thinking “This sounds very technical, why does this matter to me?”

Knowing these concepts helps you communicate effectively with AI tools like ChatGPT, Claude, Gemini, or Copilot. It’s all about giving the right instructions to get the best results.

In fact, I think this is one of the major reasons why so many beginners think that AI is overrated.

Zero-Shot, One-Shot, Few-Shot: What’s the Deal?

Imagine you’re an experienced barista at Starbucks and have to onboard a new staff-member how to make perfect coffee.

There are a few ways you could go about it:

  1. Zero-Shot: You simply ask them to make coffee without any help or guidance.
  2. One-Shot: You show them once how to make a coffee.
  3. Few-Shot: You demonstrate several times with different types of coffee.

That’s basically what we’re talking about with AI, but instead of coffee, it’s about how much information or examples you give the AI before asking it to perform a task.

Practical Applications

  • Drafting Emails: Instead of just saying “Write me an email,” try giving an example of the writing style, format, and content. The AI will pick up on these details, producing better work and saving you time.
  • Generating Reports: Provide a sample of a previous report you liked. The AI will follow that structure, making your job easier.
  • Brainstorming Ideas: Give the AI a couple of ideas you’ve already had. It’ll use those as a springboard, often coming up with creative suggestions you hadn’t considered.

Real Examples

Zero-Shot Learning: Zero Examples

Zero-shot learning allows the model to perform tasks without any specific examples, relying on its general knowledge.


User:

Create a follow-up email for a sales lead.


AI:

Subject: Following Up on Our Discussion

Hi [Lead Name],

I hope this message finds you well. I wanted to follow up on our recent conversation about your needs and how we can assist you. If you have any questions or need further information, please feel free to reach out.

Best regards,
[Your Name]


Model: GPT-4


One-Shot Learning: One Single Example

One-shot learning is about giving the model just one example to understand what you need. This helps the AI mimic the format and style for similar tasks.


User:

Here’s an example follow-up email. Can you create another one like this?

Subject: Checking In

Hi [Lead Name],

I wanted to follow up on our previous conversation about your needs. Let me know if you have any questions or if you’d like to schedule another meeting.

Best,
[Your Name]


AI:

Subject: Reconnecting After Our Chat

Hi [Lead Name],

I hope you’re doing well. I wanted to touch base following our discussion about your requirements. Please let me know if you have any further questions or if you’d like to arrange another meeting.

Best regards,
[Your Name]


Model: GPT-4


Few-Shot Learning: Several Examples

Few-shot learning involves providing a few examples to give the model a clearer picture of what you need.


User:

Create a follow-up email for a sales lead. Here are a couple of examples:

Example 1:
Subject: Checking In

Hi [Lead Name],

I wanted to follow up on our previous conversation about your needs. Let me know if you have any questions or if you’d like to schedule another meeting.

Best,
[Your Name]

Example 2:
Subject: Next Steps

Hi [Lead Name],

I hope you’re doing well. I’m reaching out to discuss the next steps in our process. Please let me know a convenient time for you.

Best,
[Your Name]


AI:

Subject: Follow-Up on Our Conversation

Hi [Lead Name],

I hope this email finds you well. I wanted to follow up on our recent discussion and see if you have any further questions or if we can assist you in any way. Please let me know a convenient time for a call.

Best,
[Your Name]


Model: GPT-4


More Advanced Approaches

While zero-shot, one-shot, and few-shot learning are approaches that end-users can use directly when interacting with AI models, supervised learning and fine-tuning are more advanced methods typically handled by data scientists and AI specialists. These methods involve extensive data preparation and training to customize AI models for highly specific tasks and industries, offering enhanced performance and accuracy for those applications.

  1. Supervised learning is a method where an AI model is trained on a large dataset, with each piece of data clearly labeled with the correct answer or outcome. This process can be likened to a structured educational program.
  2. Fine-tuning involves taking a model that has already been trained on a large, general dataset and further training it on a smaller, more specific dataset. This process can be compared to specialized professional development.

I’ll write more about these more advanced options in another post.

Putting It All Together

Understanding these learning methods—zero-shot, one-shot, few-shot, supervised learning, and fine-tuning—can significantly enhance how you use AI in your business. You can craft better prompts and get more effective results from AI models like ChatGPT, Claude, or Gemini.

Get regular updates on AI strategies that work.

You're almost there!

I turn AI tech & strategy into clear, actionable insights. You’ll discover how to leverage AI, how to integrate it strategically to get a competitive edge, automate tedious tasks, and improve business decision-making.

– Alastair.