This page contains information about the chatbot and language model Copilot (previously Bing Chat Enterprise). The tool is available for students and employees at NTNU.

Copilot is in a pilot phase, and the IT Division therefore wants feedback from those of you who try the tool.

Norsk versjon - Copilot

Topic page about artificial intelligence in teaching and assessment | Pages tagged artificial intelligence

Get started

Copilot is available in your browser. If Copilot does not work properly, try switching to the Chrome or Edge browser. Press the button below to log in.

Login

About Copilot

Copilot is a chatbot and a language model that is very similar to other well-known tools such as ChatGPT. It's a tool that's built into your browser and you can ask it things. You can get help with text improvement, ask it to read through text and suggest summaries, etc.

What distinguishes Copilot from other tools is that the security level is higher than tools found on the internet.

Information you provide may be processed outside the EU/EEA. All data is deleted when you log out of the service or close the browser window. If you do not want Microsoft to process the data you enter, do not use this tool.

Read more about the information security and privacy assessments made for this service (link missing)

Frequently Asked Questions

I don't get access to Copilot. How do I solve it?

If you don't get access to Copilot, you have two options to solve it:

Option 1

Delete all browser data, and log out and log in again.

Option 2

After you have logged in you will see that it says "Copilot", as pictured below.

If you get an error message that says "Copilot unavailable with your current SafeSearch setting. Copilot is not supported when your SafeSearch setting is set to Strict", you have to change your SefeSearch settings. You do so by following the steps below:

1. Click on the three lines in the top corner of the window.

2. Click on "SafeSearch"

3. Select "Moderate" and press "Save".

What can I use Copilot for?

You can use the tool to process information that requires a higher degree of protection than open data. This includes information that should only be available to selected internal and/or external users with controlled access rights. Such information can potentially cause damage to the institution or partners if it becomes known to unauthorized parties. Examples of such types of information can be internal working documents, exam papers and help with language washing of research articles.

For example, students can use Copilot for internal working documents, assignments and help with language washing and summarizing articles.

What can't I use Copilot for?

You cannot enter information at a higher classification level than internal, and you can not enter personal information - neither your own nor that of others.

How does Copilot work?

Copilot is a language model that is integrated with Microsoft's search engine, Bing. The search engine is supplemented with artificial intelligence (GPT-4) from the company OpenAI, which is the same technology as the chatbot ChatGPT. Copilot uses the information you enter and compares it with information found on the open web. Copilot does not have access to information on your computer, but it can access tabs in an open browser or PDF document that you make available.

What should I keep in mind when using Copilot?

When using Copilot, there are several things you need to think about, including:

• You must be critical of the answers the language model gives you. You must expect that the language model generates text with incorrect information, but presents it in an authentic way.
• You must assume that the references the language model uses may be incorrect.
• You must do a manual check of the answer the language model gives you, before you pass on or use the answer it has given you in another context.
• Remember that you will probably not get the same answer twice - do not use the language model for tasks that require integrity.
• When using and passing on answers, text generated by Copilot should be marked with the fact that the text/task has been solved with the help of AI.
• Do not use the language model to process personal data about yourself or others.
• Do not use the language model to process personal data about yourself or others.

Is it safe to use Copilot?

Yes, it is. The IT Division has assessed the service and concluded that the service is safe to use for information that is classified at internal level (yellow). The solution is not recommended for confidential or strictly confidential information. It is you as the user who must assess for yourself which classification level the information you enter is on.

Levels of information security classification:

• Green: Open
• Yellow: Internal
• Red: Confidential
• Black: Strictly confidential

Read more about information security classification

Why does NTNU offer Copilot?

The aim of making Copilot available is to be able to offer a language model tool for processing yellow/internal data. By offering a place where you as a user can enter internal data, we reduce the risk that open platforms such as ChatGPT are used to process information that NTNU wishes to keep internally in the business.

We want to offer an easily accessible and safe language model where our students and employees can get to know the advantages and disadvantages of this type of tool and technology. Language models are new technology for many, and many have never tried it. By offering a safe "sandbox" for testing, we want to contribute to all employees becoming more aware of the positive and negative effects this type of tool can contribute to the way we work. What can we use this type of tool for? What can't we use it for? Can you trust the answers you get?

In addition, we want to offer a tool where we are open about how we intend to assess risk, information security and privacy. We hope that this will be able to contribute to an increased understanding of why and how we should all think about information security in everyday life.

What assessments has NTNU made?

We have made thorough risk assessments in the following areas:

1. Risk and vulnerability assessments are carried out in line with NTNU's guidelines by the Digital Security Section at the IT Division.
2. The Privacy Impact Assessment (DPIA) has been carried out in collaboration with the Privacy Cluster in the Joint Administration and the Data Protection Officer in line with NTNU's guidelines and Digdir's Guidance for the responsible development and use of artificial intelligence in the public sector.

We want to facilitate an open discussion about the assessments that have been made, the advantages and disadvantages of using artificial intelligence. The risk and vulnerability assessment itself is exempt from the public, but we have made available the information that we believe is important to be public in the privacy impact assessment itself. The privacy impact assessment is and must be a living document, and we will therefore make it openly available (but unfortunately only in Norwegian).

If you have questions, input or comments on the assessments that have been made, we encourage you to get in touch.

Learn More About Artificial Intelligence

• Read more about information on NTNU's wiki pages about artificial intelligence in teaching and assessment
• "What is artificial intelligence? What do we have and what will we get in NTNU's toolbox?" Recording of presentation from a meeting of the Communication Network 07.06.2023 (in Norwegian only) (35 minutes).

See also

• Policy for Information Security (soon available in English)
• Policy for the Processing of Personal Data (soon available in English)

Do you need help?

• Please contact us at NTNU Hjelp.