Blog: A look at Copilot for M365
A look at Copilot for Microsoft 365

AI in the modern workplace, part 2

Author: Hasmik Najaryan
Vendor Relationship Manager, FSi Strategies

AI in the modern workplace, part 2

We’ve all marveled at the rapid content generation capabilities of AI, whether it’s drafting text on various topics, writing code, or generating photos, among other things. These tasks, which typically demanded a significant amount of time from humans, can now be accomplished in a much faster manner thanks to AI.

While we’ve shared a few chuckles over the occasional inaccuracies of AI responses, many of us have also experienced the advantages of the positive feedback and capabilities it offers. It’s natural for us to aspire to perfection, but can we honestly think of any software or solution that is flawless?

AI isn’t designed to be perfect, and it isn’t meant to replace anyone’s job. Instead, its aim is to serve as an assistant, aiming to expedite our tasks. It relies on the data it’s provided and operates within the boundaries of that information. However, this reliance on data can occasionally lead to limitations and interpretations that may not align perfectly with our expectations. Consequently, the ultimate responsibility lies with us, the humans, to review and refine the AI-generated output as needed.

Introducing Copilot

Copilot for Microsoft 365 serves as a model of AI-powered tool that seamlessly integrates into the Microsoft 365 ecosystem and our everyday applications like Word, Excel, Outlook, PowerPoint, Teams and more. It tailors AI-powered capabilities to suit your unique environment, tasks, and requirements, offering intelligent assistance, very much like a qualified copilot, who supports and alleviates the pilot’s workload but does not assume control.

You need a quick outline of your Teams meeting including the agenda items discussed and follow-up items highlighted? Not a problem! It can be ready to go immediately after the meeting—all you need to do is review it and send it via email to the intended recipients! No longer is there a need to find time in your busy schedule for additional post-meeting follow-up actions. Furthermore, it can assist in addressing specific questions you might have overlooked. Additionally, it offers customization options for structuring the summary according to your preferences, such as creating task lists, highlighting key decisions, or a bullet list. This makes it possible to tailor it exactly to your needs.

It streamlines the creation of PowerPoint presentations from your existing documents, effortlessly tailoring them to meet your specific parameters such as presentation length and format. Similar capabilities extend to Word, enabling it to supplement content, offer draft suggestions, provide text summaries, and revise sections or entire documents to enhance conciseness and alignment with your requirements. Additionally, it can function as your editor, offering suggestions to strengthen your arguments and address any inconsistencies. The extensive list of capabilities spans across all applications within Microsoft 365.

The potential of Copilot has already been seen in the short period of time it has been in use. According to Microsoft’s Work Trend Index, 70% of early Copilot users say that they were more productive while using the AI assistant.

How can you get the most out of Copilot for Microsoft 365?

The key to successful implementation lies in grounding it with your data. Grounding is the process of connecting or anchoring it to real-world data or context. This helps an AI system understand and respond to your queries or tasks based on the specific information, data, or context it gets access to, enabling it to provide more meaningful and context-aware interactions. This may involve access to your documents, emails, calendar, chats, meetings, contacts, or other relevant data sources. Naturally, this process requires meticulous design to uphold security, compliance, and privacy.

By design, Copilot is anchored on several parameters to this end, out of which the following must be highlighted:

  • It does not utilize your data for training the foundational Large Language Models that drive its AI capabilities.
  • It exclusively presents organizational data to individual users who have, at the very least, viewing permissions for that data.
  • It restricts its search for information to the user’s specific tenant and does not include data from other tenants to which the user might have access or is a guest of.

This underscores the importance of the user’s permissions to access the data and the implementation of robust information protection measures including compliance requirements across your Office 365 environment. The responsibility for this lies with you, the client. Prior to implementing Microsoft Copilot, it is imperative to conduct a thorough review and adjustment (if needed) of user permissions, covering internal, external, and guest users. Furthermore, this process should not be considered a one-time task but rather an ongoing practice requiring periodic reviews and adjustments as and if needed.

Furthermore, it’s imperative to conduct a thorough review of your data landscape, identifying and categorizing sensitive data. Proper data labeling should be implemented ensuring that sensitive data is appropriately secured, encrypted, and sharing permissions and restrictions applied. If your organization needs to adhere to any compliance standards such as GDPR, it is vital to scrutinize and implement all related requirements such as limiting the use of personally identifiable information (PII), which may entail excluding such data from Copilot’s scope. Additionally, comprehensive Data Loss Prevention policies should be consistently enforced across all data to mitigate potential oversights or errors that could result in the accidental leakage of sensitive information.

It’s crucial to bear in mind that, while Copilot may excel at automating various complex tasks, it does not have the same sensitivity filters humans have and apply intuitively. Therefore, any restrictions on what it can and cannot do, must be explicitly configured. A significant portion is contingent on the effectiveness of your overall permissions and data management implementation.

On top of that, it is important to draft policies that will dictate the scope of how Microsoft Copilot may be used by your organization. For example, AI can assist in creative tasks like content generation and design but should not be a substitute for human creativity and innovation in a art, music, or content creation requiring originality. Another company policy may dictate that any content generated by Microsoft Copilot should be subject to thorough review ensuring accuracy, fairness, and ethics before it is distributed internally and externally.

Responsible utilization of AI capabilities and the restrictions thereof should be approached comprehensively. This entails the parameters enabled by Microsoft by design, the data access controls enforced by your organization, and internal policies that dictate the permissible use of these capabilities with maturity. Key considerations should revolve around principles of safety, security, privacy, fairness, reliability, transparency, accountability, and most importantly retaining human control. This multi-layered approach can serve as the foundation for developing our models of responsible AI utilization which will inevitably lead to considerable boosts in productivity and efficiency.

At FSi Strategies, our team of experts focus on helping you utilize the latest Modern Work developments to achieve your mission. Contact us to learn more about how you can be ready to leverage Copilot for Microsoft 365 to its fullest potential.