Working With Integrity as a CBS Student

Generative Artificial Intelligence (GenAI) is a technology that has enjoyed quite a lot of attention recently with the launch of a number of end-user services, including Bing from Microsoft, ChatGPT from OpenAI, and Bard from Google. The technology is programmed to generate new and original content, including prose and images, by pre-training patterns from existing data.

In this way, GenAI is able to author well-composed and seemingly well-written papers with only the slightest prompt from a real person, which would seem to make it the perfect companion in higher education, and indeed in education everywhere, especially during peak periods where the workload may feel overwhelming.
 

Plenty of Pitfalls

There are, however, lots of problems associated with using GenAI in academia and elsewhere.  

• One of the more serious problems with the GenAI technology, not only in a high-intensity learning environment like CBS, but also in the world as such, is its inability to distinguish between true and false. Generative AI is not in the business of providing reliable content but only of prediction and probability and for that reason it is very prone to fabrication. In fact, to be able to utilize the technology with any kind of confidence, you would need to have a robust understanding of the topic at hand, an understanding that may take many years to cultivate. 
• Generative AI can replicate or even amplify negative content and may perpetuate biases and stereotypes about people and ideas. This issue has to do with the pre-training of the software. If the topic you are exploring is contentious, you may find yourself inadvertently reproducing fake news, conspiracy theories, etc. as the software is no better, no wiser, and no more beholden to the truth than the data, from which it learned. Online trolling goes on online on an industrial scale every single day, so if you just happen to tap into a minefield of viciously competing ideas when working with these services, you could be seriously compromised in your exam submissions.
• With the relative dominance of English-language training data, there is a high likelihood that the responses you get will reflect a worldview that is specific to that part of the world. More likely than not, content generated by GenAI services is prone to represent Western perspectives and people and to drown out or misrepresent alternative, non-Western experiences, attitudes, and ideas.
• Because of the costs involved, in terms of both money and energy, the service is not retrained on an ongoing basis and as such cannot necessarily be relied on to reflect current ideas and insights. In other words, a lot of output data will quickly become outdated.
• Finally, the environmental impact of interacting with GenAI as an end-user is significant so it is always worthwhile to carefully consider if the task at hand warrants the energy consumption involved or not. 

Copyright & GDPR

To avoid violations of copyright and GDPR, you need to be very careful about the kind of information you input into the genAI solution. Content downloaded from publishers, newspaper, websites, etc. are in most cases protected by copyright and if that is the case, you are not permitted to upload it without the explicit consent of the copyright holder. This is true for all content provided by CBS Library.

In the same vein, uploading personal information collected via interviews or by any other means will most likely violate GDPR and should be avoided.    

GenAI at CBS

As a CBS student, you are allowed to use GenAI in all types of written exams in mandatory courses and mandatory courses offered as electives, as well as in final projects. Written sit-in exams in the exam hall, both with and without aids, are exempted, however.

Course coordinators and study boards have the option of prohibiting the use of GenAI in specific courses, but only if it has been explicitly stated in the course description.

However, GenAI is always permitted in final projects, i.e. bachelor's projects, diploma and executive master projects, and master theses.     

Whenever you use GenAI at CBS, it is important that you are transparent about how you have used it to make sure that your work lives up to the Rules and regulations regarding conduct and integrity at CBS. This means that in some cases you are required to leave a declaration in your assignment, much the same way that you are required to leave a citation whenever you quote somebody. This is to make sure that your assessors are able to distinguish at all times between your own intellectual and autonomous contributions and those contributions that originate in interactions with a GenAI platform.

The Guidelines

1. You may use GenAI for proofreading (e.g. checking for spelling, grammar, punctuation, and formatting issues) without declaring it in your written product. 

2. You may use GenAI as a search engine without declaring it in your written product. 

3. You may use GenAI for idea generation and conceptualization (e.g. to propose a structure or elaborate on concepts and ideas) but only if you declare this in the Methodology section of your written product.

If your product does not contain a separate Methodology section, you need to mention it in the Introduction.

4. You may use GenAI to generate text, images, and other types of output as part of a written product but only if you provide a specific reference (cf. below).

5. You may use GenAI as part of the preparation for an oral exam, including the defense of your final project. If GenAI is used as a sparring partner (e.g. for brainstorming ideas, clarifying concepts, or checking information), there is no need to declare this.

However, if you use GenAI to prepare a tangible product, such as a slide show for your oral presentation, the use of GenAI must be declared clearly, orally or in writing, within the product.

Using GenAI to generate answers during the oral exam itself is not allowed.

If you decide to use GenAI in a written assignment in ways that would require you to leave a declaration of use, this is how: 

Ideation or concept generation

If you use a GenAI tool for ideation or concept generation, you are required to mention this in the Methodology section of your paper. Alternatively, if your paper does not contain a separate Methodology section, you may also mention it in the Introduction. It is important to mention it early on, as in that case, your assessors will know what to expect.    

Example of Methodology section declaration:

In this paper, I have used Microsoft Copilot to aggregate and summarize the results of 15 interviews conducted amongst CFOs in small- and medium-sized companies and based on these interviews to make suggestions for best practice financial reporting in SMEs.

Output in the body text

Whenever you draw on GenAI output in the body text of your paper, you need to declare this by leaving in both the prompt and the relevant part of the response as well as a citation. On top of this, you need to leave the full response as an appendix for context.

Example of in-text declaration and citation (in the style of APA 7):

When prompted with "What is considered best practice financial reporting in SMEs?", Microsoft Copilot mentions a whole slew of attention points, including "adopting a financial reporting framework" and "using technology to streamline financial reporting" (Microsoft, 2023)

Citing GenAI

Whenever you leave an in-text citation in the body text of your paper, it should always be accompanied by a reference. This is no different than referencing a published work like a book or a journal article. 

In APA 7, the in-text citation and reference should look like this:

Template:

(provider name, version year)

provider name. (version year). application name (version number) [Large Language Model]. URL

Applied to Microsoft Copilot, the reference would look like this: 

(Microsoft, 2023)

Microsoft. (2023). Copilot (Feb 7 version) [Large Language Model]. https://www.bing.com/search