Mendel University in Brno 

This article primarily focuses on the topic of academic publishing and covers only part of the issue of using AI in the university. Detailed information on the use of AI tools by students, including citation policies or an overview of AI tools for academia, is provided by the Mendel University Library.

Before any use of AI tools for academic purposes, each author should acquire an understanding of the potential risks associated with this technology, as well as the security and privacy options. This is primarily because there are concerns related to AI tools data retention policies as well as the recording of the information entered.

MENDELU is open to the new possibilities that artificial intelligence brings for educational and creative activities. Therefore, it is not the intention to prohibit artificial intelligence, including generative AI models. On the contrary, the aim is to set the basic principles for the use of AI, which simultaneously include:

1. Any uncritical direct copying of a text portion that is not created by the author is unethical, violates academic standards, and is fundamentally the same as ghostwriting.
2. AI should be used as a tool to assist authors; the replication of their outputs should not replace the author's own creative and innovative work.
3. Sections of work utilizing AI must be clearly marked, and the use of the tool acknowledged. Otherwise, it will be considered plagiarism.
4. The author assumes full responsibility for the text and data generated by AI. It is important to be aware that by using AI, the author agrees to the terms and conditions set by providers, including handling sensitive information.

The use of AI tools is also regulated in academic publishing by individual publishers, who in most cases do not allow the listing of AI tools as authors or co-authors. When publishing, it is always advisable to follow the conditions of each publisher for the use of AI tools in the text.

* Detailed terms and conditions from the publisher's website available by clicking on the links in the publisher's titles.

According to the ASC publisher, artificial intelligence (AI) tools are not eligible for authorship, so they cannot be listed as one of the authors. Authors may use AI tools on condition the generation of text or images is acknowledged in the text, in the Acknowledgement section. This section should also describe in detail how and when the AI tools were used. If AI will be used on a larger scale, the use of all tools should also be specified in detail in the Methods section, or other appropriate sections of the manuscript.

The use of artificial intelligence is possible but must be stated and explained in detail in the manuscripts, submitted to Cambridge University Press, as must the use of any other software, tools and methodologies. AI does not meet the authorship requirements concerning the requirement for responsibility for the content of the manuscript, which implies that AI tools and LLMs cannot be listed as one of the authors in any published scientific work. Furthermore, authors are responsible for the accuracy, integrity and originality of their scientific papers, including the use of AI tools. Any use of AI tools in manuscripts must not violate the Cambridge University Press policy on plagiarism.

DeGruyter Publishing does not accept manuscripts generated using artificial intelligence tools (including LLM, and MLT) due to the inability of these tools to take responsibility for the submitted work. Thus, AI tools cannot be listed as authors or co-authors. If AI tools have been used as part of the design or methodology of the research study, their use must be acknowledged and detailed in the Acknowledgement section.

For manuscripts submitted to Elsevier, it is expected that if the authors have used AI tools in the creation of the research paper, these tools have only been used to improve the readability and language. in the manuscript. AI tools must not replace key authoring tasks, such as generating scientific, medical, or pedagogical knowledge, and must not be used to draw conclusions or make recommendations. AI tools must not be listed as one of the authors or co-authors, nor be cited as an author. The use of AI tools is subject to human supervision and control of all output. Authors are responsible for the content of the manuscript and the use of AI tools must be stated in the manuscript itself.

Inderscience Publishing states that artificial intelligence tools cannot be listed as one of the authors. Authors are fully responsible for the content of submitted manuscripts, even if the manuscripts were largely created by AI tools. Thus, authors are responsible for inaccuracies, errors, or violations of publication ethics. The use of AI tools must also be acknowledged in the Acknowledgement section , including a detailed specification of the technologies used and their purposes.

According to the MDPI publishing guidelines, AI tools (including Chat GTP and LLM) cannot be listed as authors or co-authors of manuscripts as they do not meet the authorship criteria. While AI tools may be used in submitted manuscripts, authors are responsible for the originality, validity and integrity of the submitted work, and must ensure that it is produced following the MDPI ethical guidelines for publishing. The use of AI tools must be transparently stated in the Acknowledgement section, describing which tools have been used. Furthermore, a detailed description of how and when the tools were used must be provided in the Materials/Methods section.

MIT Press does not allow AI tools (including ChatGTP and LLM) to be listed as authors or co-authors due to their inability to meet the requirements for authors. Authors of submitted manuscripts using AI tools to create text, figures/graphics, or data collection are required to inform the editors of this use and transparently describe the use of AI tools, in the submitted manuscript so that the role of these tools in the creation of the manuscript is clear. The authors are responsible for the content of the manuscripts, including the AI-generated parts, and are liable for any ethical violations.

AI tools such as GTP chat cannot be considered authors or co-authors , according to Oxford University Press, and their use will be checked in accepted manuscripts. The use of AI, including code generation, data analysis, or content generation, must be acknowledged in the cover letter to the editors, as well as in the Methods or Acknowledgement sectionsof the submitted manuscript.

All manuscripts using artificial intelligence for study or content submitted to PLOS must include a statement in the text of the manuscript, in the Methods section, or an Acknowledgement for manuscripts that do not have a Methods section. The statement should include all names of the AI tools used, including a description of the use of the tools, an assessment of the validity of the tools' outputs, and a clear statement of which aspects of the study, paper content, data, or supporting files were influenced/generated by the use of the AI tools.

In cases where LLM AI tools contribute to the generation of text submitted to PLOS, authors are responsible for ensuring the accuracy and validity of the content, the manuscript's compliance with PLOSE's anti-plagiarism policy, and the citation of all relevant sources. In addition, the author must ensure that all statements in the manuscript stating the hypotheses, interpretations, results, conclusions, limitations, and implications of the study represent the authors' ideas.

AI tools cannot be listed as authors or co-authors on submitted manuscripts. Authors are required to clearly indicate in the manuscript, which language models were used, including which model was used and for what purpose. Place this notice in the Methods or Acknowledgement section. of the manuscript as appropriate. In addition, authors must verify the accuracy, validity, and appropriateness of the content and all citations generated by the language models and correct any errors or inconsistencies.

It is also the author's responsibility to list the sources used to generate content and citations, including sources generated by language models. Review the citations to ensure that the authors are accurate and that they are properly referenced. Authors must also look for the possibility of plagiarism in cases where the LLM reproduces substantial portions of text from other sources, and then check the original sources to ensure that authors are not plagiarizing someone else's work. The potential limitations of language models should be noted in the manuscript, including the possibility of bias, errors, and gaps in knowledge.

According to the guidelines of the American Association for the Advancement of Science (AAAS), AI tools, including ChatGTP, LLM, and tools used to generate figures, are not qualified for authorship and should not be credited as authors or co-authors of the manuscript, nor cited. Authors who use AI tools in their work are required to inform the editors in the cover letter and also in the Acknowledgement section of the manuscript. Furthermore, detailed information about which tools were used, including their version, must be provided in the methods section.

Authors are responsible for the accuracy of the text and the compliance of the manuscript with ethical principles, including the rules for assessing plagiarism. If AI tools are used, authors are required to double-check the sections of the manuscript to ensure that they are correct. The use of images generated using AI tools is strictly prohibited in Science (AAAS) publications without the explicit permission of the editors.

AI tools (including LLM models and ChatGTP) are not qualified for authorship and cannot be credited as authors or co-authorsof the work in submitted manuscripts. If AI tools are used during manuscript development, their use must be properly documented in the Methods section or other appropriate sections of the manuscript. Furthermore, Springer Nature does not allow the publication of manuscripts containing AI-generated images except in articles and materials that directly reference these materials.

According to the Taylor & Francis publishing policy, AI tools (including LLM, and ChatGTP) may be used in submitted manuscripts provided they comply with the publisher's policies on authorship and publishing ethics. Authors assume responsibility for the content and its accuracy when using AI tools. AI tools may not be credited as authors or co-authors, but all use of AI in manuscript development must be acknowledged and properly documented.

The Royal Society accepts manuscripts using AI tools under the condition that this is required to be stated directly in the text of the manuscript. When a manuscript is published, an author's statement on the use of AI will also be required , which should include details of the tools used and how they were used, including defining which parts of the text were created with or by AI tools. It is the responsibility of the editors and reviewers to assess the appropriateness of the use of AI tools.

If AI tools were used in the process of writing the manuscript, it should only be to improve the readability and language of the work. The use of AI to edit the manuscript must be stated in the paper and must not replace the key tasks of the researcher, such as generating scientific knowledge, analysing and interpreting data, or drawing scientific conclusions.

AI Tools should not be cited as an author or co-author of the publication and should not be cited as such.. It is the author's responsibility to verify originality, but also to check that the use of AI tools does not violate the publisher's ethical guidelines and that the content of the paper itself is correct.

Artificial Intelligence AI Tools, in the opinion of the publisher Wiley, does not meet the requirements for authorship and therefore cannot be cited as an author or co-author. of the article. When using these tools in the creation of a manuscript, the author must detail what kind of tool was used, for what purpose, and in what part of the text. This notice should be included in the Methods and Acknowledgementssections of the manuscript..

The authors take full responsibility for the accuracy of all information provided by the tool and for the correctness of the references to the supporting works on which this information is based. Artificial intelligence tools used to improve the quality of spelling, grammar or general manuscript editing are not included in these guidelines..

If you use generative AI to create your manuscript, you should indicate in the relevant sections of the text, following the instructions of the publisher, explaining how you used the tool, even if you do not include any AI-generated content in your work. We recommend that you include a description of the AI tool you used, how you generated the information, including the prompts you used.

Example 1:

Please note the use of [name of AI tool or technology] and [link] to create [fill in; preferably include specific parts of the manuscript]. I have entered the following prompt/commands [fill in]. I have used the output to [fill in]. I have edited the output of [fill in].

Example 2:

I certify that I have used [name of AI tool or technology] in the development of this assessment to create materials for basic research and self-study.

Example 3:

In this section, I would like to point out that I have used ChatGPT ChatGPT [https://chat.openai.com/] in developing this assignment to get ideas and input for basic research and project planning. The following commands/prompts were entered into ChatGPT on March 30, 2023 "There is evidence of water on planets outside the solar system".

Example 4:

Note the use of [name of AI tool or technology] to structure or plan this evaluation.

Example 5:

I certify the use of [name of AI tool or technology] to create the materials that have been included in modified form in my final evaluation.

Example 6:

I certify the use of [name of AI tool or technology] to create the materials included in my final evaluation unchanged.*

*An example might be a quote, or an image created by AI output.

Generative AI tools carry great potential and risks, and it is important to be aware of the potential for misuse of these tools in academic settings, for example, passing off AI text output as one's work. Therefore, it is good to be aware of the possibilities of identifying misuse of AI technologies.

There are many tools available for free or for a fee on the Internet to identify AI-generated text. Their major drawback is low accuracy and reliability, but often also the lack of information about how the provided text is handled and to whom the text is subsequently available. Available identifying tools also often cannot keep up with the rapid development of new AI technologies, for which AI detection of use may not be possible at all. MENDELU cautions against the use of these tools and recommends that authors use them only to check their texts, with caution against the risks involved.

In addition to the tools for identifying AI-generated text, there are also some general characteristics that AI texts may, but will not always, exhibit. These characteristics may not be present in all AI-generated texts or data sets but serve as basic references. According to the established Principles of Artificial Intelligence Utilization at Mendel University in Brno issued as of 11.5.2023, these include:

• The text contains generally known information formulated without adherence to terminology and contextual relevance to the thematic focus of the work.
• Sentences with the same idea are often repeated. AI-generated text includes repetitive phrases, utilizing identical structures and words repeatedly. Human-written text tends to have a more natural and varied style with a wide range of sentence structures and word choices.
• The text includes contradictory statements (even within the same clause or paragraph).
• Truth distortion occurs in the text by claiming that there is only one answer to a particular question.
• Arguments are used that are nonsensical, contradictory, or unrelated to the topic.
• The text exhibits a low level of emotional expression. If the AI receives a question without any prompts regarding tone or other attributes, it generates factual text without personal insights or emotions.
• The text contains statements based on biases and stereotypes and is biased towards a Western understanding of the world.
• The text references sources dated only up to 2021 (the current model is trained on a temporally limited dataset) or includes citations to non-existent sources (the AI fabricates them).
• The text lacks errors (grammatical, linguistic, typos) despite the fact that making mistakes is a human trait.
• The text includes long sentences – AI tends to produce redundant words, especially when given a broad question without specific instructions.
• The text extensively employs words such as "it," "this," "these," "those." Humans typically strive to minimize their frequent repetition to enhance content readability.