Microsoft Copilot. 7 estrategias para una Interacción Efectiva con la IA Generativa

Generative AI: 7 Strategies for Interacting with Copilot

When asking Copilot to do something for you, think of it as a person at the beginning of their career and consider yourself their supervisor: the more details, context, and resources you provide in the prompts, the better the response will be. This is key to unlocking the power and potential of generative AI.

1. Optimize interaction with generative ai by clearly defining your goals

In the prompt, it’s important to express the outcome you want Copilot to return to you.

Do you want it to return a list or a text? Do you need an email or the script for a brainstorming session?

One of the strengths of generative AI is imitation. Large language models (LLMs), are artificial intelligence systems designed to process and generate text in natural language. These models use deep learning techniques and are trained on large amounts of linguistic data to understand and produce text similar to what a human speaker would use. So, try to give examples of how you would like the outcome to be.

“Write this thank-you email in an informal and friendly style” or “Rewrite this as a highly structured summary with long and complex paragraphs.”

2. Provide the necessary context for effective and understandable communication

Indicate the context or relevant details about the outcome Copilot is working on.

If you’re preparing for a meeting with a client or trying to write a speech for your team, communicate it to Copilot. If you’re working for a particular client, which one? And what part of their campaign are you thinking about?

It’s also helpful to give Copilot a perspective from which to respond. That usually involves some explanation of who you are and what you’re trying to achieve so the AI can think from your point of view: for example, you’re a social media manager writing LinkedIn copy, you’re a product marketer working on a new campaign, you’re a programming tutor who’s great at explaining Python to students.

“With traditional search, people have gotten used to being very concise”, says Tara Roth, Corporate Vice President of Microsoft 365 Customer Success Engineering.

“You use just a few keywords and get a lot of links, and then you do a lot of processing on your own”. But with Copilot, she says, being more descriptive gets you better responses.


3. Specify relevant details for generative AI and get better responses

It’s not a strict requirement, but specifying the sources of information Copilot should use will yield better results.

You can introduce a broad prompt or label specific files. Specify which date ranges Copilot should focus on or which meetings the information should be extracted from.

Jaime Teevan, Chief Scientist and Technical Fellow at Microsoft, likes to ask Copilot to explain how it arrived at an answer. “It’s similar to the way math teachers ask students to show their work: they get better answers from kids”.

4. Adjust your tone according to the situation and context

Indicate your expectations for how Copilot should give its response. Who is the audience? What kind of tone are you thinking for this task?

This usually involves providing a brief explanation of your identity and goals. For example: “Explain this to me as if I were a novice programmer” or “I’m a busy vice president: give me just the top-line conclusions”.

Additionally, you can request a response in the style of a specific person or approach. For example, ask it to solve a problem with the expertise of a Stanford business professor or explain a concept in a way that’s understandable to someone without technical knowledge.

For instance, consider this prompt:

“Generate 3 to 5 points of a list to prepare for a meeting with Client X to discuss their ‘Phase 3’ marketing campaign. Focus on emails and Teams chat since June and use simple language for quick understanding, please.”

  • Specify the desired outcome.
  • Provide relevant context about the meeting with Client X and the marketing campaign.
  • Define the sources of information and parameters Copilot should use.
  • Indicate tone expectations and the type of response.

5. Be polite and respectful in your interaction with generative AI

It’s crucial to maintain a polite and respectful tone when interacting with generative AI. Kurtis Beavers, Director of the Copilot Design Team at Microsoft, emphasizes this. Using polite language not only sets the tone of the response but also contributes to generating collaborative and respectful results.

Generative AI reflects the levels of professionalism, clarity, and detail of the contributions provided. Maintaining a polite approach is not only a courtesy practice but can also influence the quality of the generated responses.

6. Be curious and ask relevant questions to improve the quality of responses

To facilitate a conversation that generates valuable responses, it’s important to have a curious and proactive attitude. As Sam Schillace, Vice President of Marketing and Chief Technology Officer at Microsoft, points out, “Knowing what you don’t know is essential”. In the context of interacting with generative AI, this growth mindset becomes even more relevant.

Asking preliminary questions to gather information can be crucial for obtaining more accurate and useful results. Encouraging curiosity not only expands knowledge about the subject in question but can also influence the quality and relevance of the responses provided by generative AI.


7. Make iterations to improve your interaction and get optimal results

Copilot doesn’t get discouraged easily, and neither should you. The initial question you pose is just the starting point. According to Jon Friedman, Executive Vice President of Design and Research at Microsoft, “It’s like the first turn of a crank”. In this collaborative process with generative AI, you can request adjustments, edits, or additions. It’s an opportunity to improve what was initially generated.

Collaboration involves a constant flow of information: inputting data, receiving responses, asking questions, and mutually learning. Working with generative AI is not just about giving orders and getting responses. It’s a dynamic and interactive process. You and Copilot co-create content, solve complex problems, and acquire new knowledge in the process.

Conclusion

In summary, effective interaction with Copilot and other generative AI is crucial to unlocking its full potential. By following these strategies, you can optimize your communications, obtain more accurate and relevant responses, and collaborate more efficiently with AI. Remember that working with generative AI is a dynamic and collaborative process, where curiosity, courtesy, and willingness to iterate are key.

Ready to make the most of generative AI in your daily work? Contact us to discover how we can help you optimize your communication processes and achieve optimal results with Copilot and other artificial intelligence tools!

Marina Albacar Subirats – Product Owner at Itequia