The guidelines for when we use artificial intelligence and what to be aware of are set in a code of conduct that is currently being developed and will be announced once it is in place.

There will be ongoing communication about how the project is progressing, when and how one can participate, provide good suggestions, and share ideas and knowledge with colleagues. Communication occurs through various channels, including Update, department-level emails, knowledge sharing in MS Teams, and physical events.

Artificial intelligence in its latest form, Generative AI, is new for most people, and the possibilities are abundant. It requires new skills and understanding to utilize this technology. Therefore, a course will be offered to employees in administration (not specifically targeted at research and teaching). The course will be an online program in Moodle.

 An essential skill in using Generative AI is to “prompt” an AI service correctly by directing it toward the right approach for the task it needs to perform. The university library will assist with this. They will offer courses on “prompting” via www.en.aub.aau.dk during the spring.

For Everyone

AAU’s licenses provide access to Microsoft’s “Copilot Enterprise”. It is a counterpart to ChatGPT but offers better data protection. You can already use it by logging in with your AAU credentials.

Integrated into Existing Programs

Microsoft is also integrating Copilot into familiar office programs such as Outlook, Word, Excel, Teams, CRM, and more. This makes artificial intelligence an enhancement and extension of the programs we already use and know, rather than a separate program. It also means that AI draws from and learns from your own data without making it accessible to others. This significantly increases possibilities and quality.

This section, Copilot for Office 365, is being rolled out in stages. It begins with testing within IT services to ensure that functionality and security are in order. One prerequisite for Copilot for Office 365 to work is that our email and calendar systems must be moved to the cloud. So that needs to happen first.

Many are eager to get started, but it still requires a bit of patience because it cannot be rolled out to everyone simultaneously. Additionally, the value of AI in a Danish context is still being explored, but support for the Danish language will be available in a few months. Individual departments will be notified as they can be connected to the service.

It is also possible to use Generative AI embedded in a service, such as a chatbot that can provide answers and assist users or employees with information on specific topics. Here, it is crucial that the chatbot utilizes the correct data and is trained to respond as AAU desires. Other possibilities include handling sensitive data. This requires a “protected environment” and an understanding of how to handle data and use AI to support decisions and resolutions.

The project aims to create three Proof-of-Concepts (experiments) for embedded services in 2024, and candidates will be selected for this in the spring of 2024. If you have a good idea for an “embedded service” or a complex case that could benefit from AI, discuss it with your manager, who can further convey the ideas to the project (there is a template for this purpose).