1.4.3 Microsoft Copilot
1.4.3 Microsoft Copilot
With Copilot, AI is making its way directly into the Office environment, and this is changing the way we work.
Microsoft Copilot is a suite of generative AI features integrated directly into the applications of the Microsoft 365 product range – including Word, Excel, Outlook, PowerPoint and Teams. It is based on Large Language Models (e.g. GPT-4), which enable typical office and administrative tasks to be carried out more efficiently and intuitively.
The aim is also to make work processes more efficient, to foster creativity and to facilitate interaction with complex data.
Copilot Chat is in use at HAW Hamburg.
Typical features of Microsoft Copilot
In Word
- Automatically generate draft text
- Suggest outlines
- Rewrite or summarise text
In Excel
- Analyse data sets
- Suggest formulas
- Recognise and visualise patterns
In PowerPoint
- Create presentations from bullet points
- Design suggestions for layouts and graphics
In Outlook
- E-mail suggestions based on previous correspondence
- Summaries of long email threads
In Teams
- Automatically generate meeting notes
- Extract tasks and to-dos from meetings
Benefits in a higher ed context
Administration
- More efficient creation of reports, minutes and e-mails
- Faster processing of internal communication
Teaching
- Drafting and structuring course materials or presentations
- Support with formatting teaching documents
Research
- Structuring of interim results
- Data analysis in Excel
- Drafting of third-party funding applications or project texts
Studying
- Outlining and revising assignments
- Creation of learning materials
- Formatting assistance and translation suggestions
Notes on usage
- Directly integrated: Copilot is not a standalone programme – its functions are embedded within familiar Microsoft applications.
- Prompt quality matters: The clearer and more precise the input, the more helpful the output.
- Check data protection: Even though Microsoft has security measures in place, sensitive or personal data should only be used with caution.
Limitations and challenges: Misleading formulas, generic slides, incomplete e-mails
- Prone to errors in complex tasks Example: In Excel, automatically generated formulas can be misinterpreted – for instance, in nested calculations or when certain cell formats are not taken into account.
- Superficial or generic content Example: PowerPoint slides sometimes appear interchangeable, as the content is based on standard phrasing and does not take subject-specific requirements into account.
- Gaps or inaccuracies in text suggestions Example: Outlook may suggest e-mails that do not correctly include important attachments, contacts or contextual information from previous mails.
- Technical limitations depending on the licence model Not all Copilot features are automatically available to all university users – particularly where GPT-4 support is lacking or system integration is limited.
Teaching tip: AI literacy through application
Copilot can significantly reduce the workload for lecturers when creating materials and for researchers when documenting their work. Students benefit from structural and linguistic assistance in Office documents. However, as with all generative AI tools, the following applies: only those who use Copilot consciously and reflect on it critically will truly reap the added value.
Information
Information on Copilot prompts
Further information can be found in the HOOU course (in German) KI-Tools kurz erklärt! So verwendest du ChatGPT, Leonardo.AI & Co.
💡 Learning Summary Chapter 1.4.3: Microsoft Copilot
- Microsoft Copilot supports directly within M365 apps: Copilot is integrated into Word, Excel, PowerPoint, Outlook and Teams and helps with creating, structuring, visualising and summarising content.
- Wide range of applications in a higher education context: Copilot can support lecturers, researchers, administrative staff and students in creating materials, analysing data and communicating.
- Conscious and competent use required: Copilot is no substitute for critical thinking. Prompts must be clearly formulated, sensitive data entered with care and suggestions checked for accuracy.