AI Use Guidelines

Related documents

Authorship and Publication of Results policy; Human Subjects Research policy; Research Data Management policy; ChatGPT and Artificial Intelligence in higher education; Recommendation on the Ethics of Artificial Intelligence; Examples of using AI to support student learning & assessment in Life Sciences

Introduction:

Artificial Intelligence (AI) is a rapidly evolving field that has the potential to transform various aspects of human life, from education and health to economy and security. As University Mohammed VI Polytechnic (UM6P) we recognize the importance of embracing AI as a strategic asset for advancing our academic and research missions. However, we also acknowledge the ethical and social implications of AI, and the need to ensure its responsible and beneficial use. These guidelines aim to establish ethical, transparent, and responsible practices that guide the integration and use of artificial intelligence in research, education, and development fostering a positive and socially conscious impact on both the university community and society at large.

These Guidelines outline the standards and expectations related to the use of AI within the (UM6P), as well as the goals and vision for the integration of AI across our curricula and research.

Q01: What is ChatGPT and what is generative artificial intelligence?

Answer: ChatGPT is an advanced language model developed by OpenAI. Short for Generative Pre-trained Transformer (GPT), ChatGPT is a tool designed to generate human-like responses to prompts. This AI system is a Large Language Model (LLM) that has been trained on a dataset to have conversational interactions with users.

While ChatGPT is well known, it is a part of a wider framework of generative AI. This type of AI creates new content based on the data it has learned during its training to perform tasks and can iteratively improve themselves based on the information they collect. The content generated can take a variety of forms: text, images, code, videos, etc. In addition to ChatGPT, there is a diverse array of other AI tools, such as Bing (Copilot AI), Bard, DeepDream and DALL-E 2.

Section 1: Education

Q02: Is the use of AI tools considered as an academic misconduct at UM6P?

Answer: While there is no university-wide ban on generative AI tools use, decisions regarding their incorporation into courses lie with individual instructors or head of department. Preserving academic integrity is paramount, and the use of AI tools must align with these principles, avoiding any facilitation of academic misconduct such as plagiarism of cheating (see plagiarism policy). Instructors are strongly encouraged to communicate their expectations early in the semester. Students are permitted to leverage AI tools for assistance in assignments, but disclosure is mandatory, and usage should comply with course guidelines and ethical standards, ensuring a transparent and ethical integration of AI technology in academic endeavors at UM6P.

Q03: Is it acceptable for students to utilize AI tools for completing assignments?

Answer: UM6P expects students to independently complete assignments. The university underscores the core principle of academic integrity, emphasizing students’ responsibility to produce their own work. Thus, students are permitted to incorporate these tools only when explicitly allowed by their instructors. Instructors are encouraged to articulate clear guidelines on the acceptable use of generative AI tools in assessments, moving beyond vague references and ensuring responsible and ethical integration within the framework of academic integrity.

Here are some examples of statements that can be found in a syllabus to clarify if the use of AI is allowed or not, and why.

Can use Generative AI tools:

– Students are encouraged to utilize technology, including tools based on generative artificial intelligence.

– Proper citation is required for any content generated mby an artificial intelligence tool.

– Students have the option to employ generative artificial intelligence tools while completing this assignment. Their use must be documented in an appendix. The documentation should encompass details such as the specific tool(s) utilized, the manner in which they were applied, and the integration of AI-derived results into the submitted work.

Can’t use Generative AI tools:

– Utilizing generative artificial intelligence tools or applications for this assignment, including tools such as ChatGPT and other AI writing or coding assistants, is not allowed.

– Presenting an idea or expression, generated by AI, as one’s own may be regarded as an academic offense.

– The intentional utilization of generative artificial intelligence tools, encompassing ChatGPT and other AI writing and coding assistants, to finalize or aid in the completion of this assignment may be deemed an academic offense in this course.

Q04: If a student is allowed to utilize Artificial Intelligence for an assignment, how should they properly cite it?

Answer: Ensuring academic integrity in the utilization of generative AI tools, UM6P mandates that students, when permitted to integrate AI into assignments, must diligently cite any content produced by the tool. To facilitate accurate citation, UM6P provide guidance on how to cite generative AI use in MLA, APA and Chicago Style.

Citing in MLA:

How do I cite generative AI in MLA style?

Citing in APA:

How to cite ChatGPT

Citing in Chicago Style:

How do you recommend citing content developed or generated by artificial intelligence, such as ChatGPT?

Please verify with your professors to ensure that the citation guidelines outlined above align with the specific requirements of their respective classes.

Q05: What should an instructor do if they suspect a student may have used unauthorized AI during the completion of an assessment?

Answer: If an instructor prohibits the use of AI tools in assessments and a student is later found to have employed such tools, UM6P’s policy mandates an initial step involving a non-adversarial conversation. This dialogue prioritizes fostering understanding and education, enabling instructors to assess the student’s comprehension of their work and providing guidance on the appropriate and inappropriate applications of AI.

After this conversation, instructors can proceed with the formal process for addressing academic misconduct concerns, ensuring a fair examination while upholding principles of education and due process (see code of conduct).

Section 2: Research:

Q06: Is Artificial Intelligence accurate and reliable?

Answer: LLM lies in their ability to predict the next word in a sentence, maintaining context over extended portions of text and generating generally coherent responses, as seen in models such as ChatGPT. However, a notable challenge arises in the accuracy of stating facts or quotations, where these models may produce claims that, while sounding authentic, can be inaccuracies to experts. The risk of perpetuating online biases remains as generative AI systems are trained on online materials. OpenAI acknowledges this concern and has undertaken significant efforts to mitigate biases, yet the potential for biased outputs persists.

Q07: Can AI be used in writing and/or developing manuscripts?

Answer: As you leverage AI as a tool for writing and manuscript development, adherence to ethical guidelines is paramount. AI, particularly LLM like GPT, presents a valuable resource for predicting and generating text, maintaining context over extensive portions of content. However, it is crucial to transparently disclose the use of AI in your writing endeavors and ensure that its application aligns with the principles of research integrity. Nevertheless, be cautious in relying on these models for stating facts or quotations accurately, as they may occasionally generate claims that, while sounding authentic, can be inaccuracies to experts.

Given that generative AI systems are trained on online materials, there’s a risk of perpetuating biases present in that data. OpenAI is actively addressing this concern, but it remains a potential danger. Exercise diligence in critically assessing the outputs of generative AI, be transparent about its use, and remain vigilant to uphold ethical standards in your writing and manuscript development processes.

Q08: How should the use of AI be reported in my research?

Answer: The UM6P underscores the importance of transparency and scholarly integrity by the explicit report of AI utilization within the methodology section of research publications. This directive aligns with UM6P’s commitment to consistency in scholarly practices, requiring scholars to provide clear and detailed accounts of AI tools, algorithms, or models employed in their research.

Q09: Can AI be listed as an author?

Answer: In alignment with widely recognized standards and legal considerations and acknowledging the consensus among journals and research communities, it is affirmed that AI models, including ChatGPT, cannot meet the requirements for authorship due to their inability to take responsibility for submitted work, manage conflicts of interest, or handle copyright and license agreements. Moroccan law, as traditionally understood, designates authors as natural persons who exercise skill and judgment in the creation of a work, thereby excluding AI entities from authorship (see Authorship and Publication of Results policy).

Scholars are advised to adhere to established guidelines for citing resource types with their chosen reference style (See Q04).: If a student is permitted to use ChatGPT on an assignment, how should they cite it?

Section 3: Development:

Q10: How can AI be responsibly applied in development?

Answer: AI must be used in a manner that upholds the UM6P’s commitment to the following principles:

– Human-centricity and well-being: UM6P underscores the paramount importance of human-centricity and well-being in the development and implementation of AI. Every individual engaged in AI within the university is obligated to uphold the fundamental rights and dignity of those impacted by their work. This mandate entails a conscientious avoidance of harm to physical or moral integrity, prohibition of discriminatory or humiliating treatments, and prevention of the exploitation of vulnerabilities or states of precarity. The university places a strong emphasis on prioritizing the well-being and needs of humans throughout the AI development lifecycle (see Human Subjects Research). Additionally, our commitment extends to ensuring that AI technologies do not infringe upon internationally recognized human rights.

– Data security and privacy: UM6P highlights the critical importance of data security and privacy. Individuals involved in AI research and development within the university are expected to respect the privacy of individuals, refraining from publishing any information related to the private sphere without explicit consent. Furthermore, when handling information considered confidential by the university, individuals must strictly adhere to confidentiality standards, with any unwarranted dissemination subject to sanctions as defined in the Research Data Management policy. As a fundamental commitment, individuals engaged in AI research are required to familiarize themselves with national and international normative frameworks governing privacy, personal data usage, copyright, and related rights. This includes compliance with legal provisions such as the constitution, data protection laws, copyright laws, and archival regulations.

– Responsibility: UM6P is firmly dedicated to the responsible integration of AI into development. Within this commitment, the university underscores the critical importance of considering the ethical implications of AI applications. UM6P accentuates the accountability of AI actors, urging them to ensure the responsible functioning of AI systems and adherence to ethical principles. This commitment to responsibility extends beyond technical considerations, fostering a culture where every member of the academic community actively engages in the ethical dimensions of AI.

– Explainability: At UM6P, our commitment to the responsible integration of AI is deeply rooted in the principle of ensuring that AI outcomes are intelligible. We prioritize the development of inherently explainable algorithms. This emphasis on explainability gains particular significance when the impact on end users is substantial and enduring. Our AI actors must contribute actively to this commitment by developing algorithms that prioritize clarity in decision-making.

– Transparency: UM6P’s approach is the belief that transparency facilitates effective challenges to decisions derived from AI systems, safeguarding essential rights such as the right to a fair trial and an effective remedy. Furthermore, UM6P acknowledges that emphasizing transparency is integral not only for ethical practices but also for the effective functioning of national and international liability regimes.

– Accuracy: At UM6P, accuracy stands as a fundamental aspect in our commitment to the responsible integration of AI. A pivotal element of this commitment involves the clear recognition and communication of sources of error and uncertainty in algorithms and data. By embracing accuracy, we empower stakeholders with a comprehensive understanding of potential limitations, facilitating informed decision-making.

– Autonomy: UM6P recognizes the paramount importance of safeguarding the autonomy of developers. UM6P is dedicated to ensuring that AI technologies are deployed in a manner that enhances decision-making processes without inappropriately delegating authority in ways that could compromise autonomy. It is necessary to strike a careful balance, where AI serves as a valuable tool to support decision-making while preserving the individual agency and autonomy of stakeholders.

Section 4: Examples with ChatGPT

Example n°01: Helping to understand a topic

Use AI as a tutor to understand topics more deeply. It’s faster to get detailed answers from AI than from search engines because AI gathers information from multiple sources and can engage in conversation. This means you can ask questions in different ways until you find the answer you’re looking for.