Artificial Intelligence

Artificial intelligence is a set of technologies capable of performing tasks that mimic human thinking – for example, writing text, creating images, analysing data, or suggesting decisions. In reality, it generates responses by recognising patterns and on a probabilistic basis.

Three main types of AI are worth distinguishing:

Data-driven AI system: A business analytics solution built on computer models that uses AI technologies to process large volumes of data, recognise patterns, make predictions, optimise, and support decision-making. An example is a timetable-optimising algorithm that compiles the most efficient schedule based on available data.

Generative AI system (GenAI): A machine-based system, tool, or service designed to produce output content of a quality comparable to human problem-solving, taking into account the user’s questions, requests, and other input data (text, image, video). This includes large language models (LLMs) and standalone systems built on them (ChatGPT, Copilot, Gemini, or Claude), AI-based image generators (e.g. DALL·E, Midjourney), and solutions embedded in other software (e.g. GitHub Copilot).

Agent-based AI system: Agent-based artificial intelligence refers to AI systems capable of autonomously setting goals, making plans, and executing them. These digital agents do not merely follow programmed instructions but dynamically adapt to changing environments. An example is an AI agent that independently breaks down tasks – gathers information, checks the calendar, and prepares an email.

1. Machine Learning (ML): A subfield of AI where algorithms ‘learn’ from data without being explicitly programmed.
2. Deep Learning (DL): A subfield of machine learning that uses multi-layered neural networks to recognise complex patterns.
3. LLM (Large Language Model): An AI system trained on vast amounts of text, capable of generating natural-language text (e.g. ChatGPT, Claude, Gemini).
4. GPT (Generative Pre-Trained Transformer): A specific type of LLM on which ChatGPT and other tools are built.
5. Prompt: The instruction or question given to the AI by the user.
6. Token: A linguistic unit processed by the LLM (approximately the size of one syllable). Service providers often price usage based on tokens.
7. Hallucination: When the AI confidently states something that is actually incorrect or non-existent (e.g. citing fabricated sources or generating false data).

Corvinus University supports the responsible use of artificial intelligence in accordance with the principles set out in Section 5 of Executive Board Resolution 12/2025:

• Openness – The University supports the conscious, responsible, ethical, sustainable, and transparent use of AI tools.
• Initiative – The University prepares its students for the changing technological environment.
• Professional and scientific integrity – Protecting the integrity of intellectual work is a shared responsibility.
• Human oversight and responsibility – Responsibility must always be assigned to natural or legal persons.
• Autonomy – Members of the community may freely decide on the use of AI tools within the framework of the rules.
• Transparency – The use of AI must be clearly indicated.
• Equal opportunity – AI tools must be accessible to everyone under the same conditions.
• Data protection – The use of AI must comply with data protection regulations.
• Responsibility – AI tools must not be used in a misleading or manipulative manner.
• Dialogue – Questions and concerns may be raised at any time with the course leader or programme directors, and if necessary with the AI Dean.
• Sustainability – AI tools must be operated in an environmentally, economically, and socially sustainable manner.

According to Section 3(2) of Executive Board Resolution 12/2025, the personal scope of the university regulation extends to the University’s employees, contractors, scholarship holders, and students. This FAQ was primarily prepared for students, but the principles apply to everyone.

Yes, but only with the permission of the subject leader and in compliance with the applicable rules. For each course, the course leader determines how and to what extent AI use is permitted, while adhering to university guidelines. You can find the rules in the course syllabus or in the instructor’s briefing during the first week of the course. The general university rules apply even if the course syllabus does not contain any restrictions! If there is no explicit prohibition, generative AI may be used, but you must declare its use and comply with the relevant rules. If in doubt, ask the instructor in writing!

It is prohibited if:

• the instructor or the task description explicitly prohibits the use of AI;
• you submit AI-generated text as if you had written it yourself;
• you fail to indicate that you used AI for any purpose or task during your work;
• you use AI during exams or closed assessments without permission.

These constitute violations of the plagiarism policy and may result in disciplinary consequences. During closed assessments (e.g. written exams, in-class tests, Moodle tests), the use of AI is fundamentally prohibited unless the instructor explicitly permits it.

According to Section 17(1) of the Vice-Rector for Education’s Resolution 6/2025: in the absence of an explicit prohibition declared by the course instructor, generative AI may be used for completing tasks requiring independent student work, within the framework of the regulations. In case of doubt, pursuant to Section 17(2), the student shall request written clarification from the course leader or instructor.

Important: Unauthorised AI use entails the legal consequences defined in the Study and Examination Regulations and the Plagiarism Policy.

In all cases, clearly and accurately indicate in the submitted work:

• which AI tool you used (e.g. ChatGPT 5, Claude 4.5);
• what you used it for (e.g. brainstorming, drafting an outline, literature research, programming, text editing, translation);
• to what extent it contributed to producing the result.

The Corvinus University AI regulation also provides a declaration template, which may be required by the instructor or other regulations to accompany the submitted material. For example, it is mandatory for theses.

• Submitting AI-generated text, images, video, or program code as your own work.
• Incorporating AI-generated material without citing the source.
• Misleading the instructor, reviewer, or fellow students about the origin of the work.
• Sharing the work, teaching materials, or personal data of other students or instructors with AI.
• Using AI for plagiarism, cheating, or facilitating such activities.
• Using AI for any other unethical activity.

You can use process-recording tools (e.g. DraftBack in Google Docs, track changes in Word), version control for coding (e.g. Git), or simply save multiple versions of your work to document your progress. These can help prove that the work was genuinely your own creation. But remember: work done with AI can also be an independent achievement if you properly document what you did.

Today’s generative AI is based on probability distributions generated from training data. The generative process consists of producing content that is highly likely to be coherent with the user input (prompt). It is important to know that current generative AI systems do not understand the words they generate. Image generators do not understand the physical structures they depict either.

The statistical foundation provides far-reaching coherence, but reliance on statistics sets limits on the reliability of generative AI. Seemingly plausible but incorrect output (so-called ‘hallucination’) occurs frequently. Therefore, always verify AI-generated information from reliable sources!