Understanding AI: What is (Generative) AI?
Artificial intelligence (AI) refers to systems that mimic human thinking processes such as learning, problem-solving and decision-making. This is achieved using algorithms that identify statistical patterns in large amounts of data and generate outputs based on these patterns. The term AI encompasses various technologies, such as neural networks, deep learning and large language models (LLMs).
Neural networks form the foundation of this: broadly modelled on the human brain, they consist of many interconnected processing units that learn together to recognise patterns in data. Deep learning builds on this and works with many successive layers, each recognising more complex patterns than the previous one – ranging from simple structures to entire faces, or the meaning of sentences. Large Language Models (LLMs), finally, are deep learning models specialised in language, which have been trained on enormous amounts of text and are capable of composing texts, answering questions, or explaining contexts.
Despite their different approaches, all these systems have one thing in common: they mimic human cognition without actually thinking or understanding. Their behaviour is based on statistical correlations in the training data, which means they are fundamentally dependent on the quality, scope, and potential biases of that data.
Generative AI is a subfield of AI that focuses on generating new content such as text, images, audio or video, and is primarily based on deep learning approaches. Well-known commercial examples include ChatGPT from OpenAI, Claude from Anthropic, and Gemini from Google.
The University of Greifswald’s University Computer Centre provides students and staff with access to generative AI systems via its AppHub. These are hosted on the university’s own servers. Data is therefore not passed on to external companies. This provides a secure framework for using generative AI in studies, teaching, research, and administration.
For students
Generative AI already plays an important role in studies. It helps with literature research, writing papers, and creating flashcards. In order to use generative AI effectively and to be able to assess the risks, users should have a basic understanding of how AI works, what it can do, and, above all, what it cannot yet do.
Generative AI creates human-like content based on probabilities. It was trained using huge amounts of data. Based on these vast amounts of data, AI can recognise and replicate patterns. However, AI does not yet have a genuine understanding of the content. The content AI produces is always based on the training data. Consequently, AI may reproduce stereotypes, and content may be heavily mainstream-oriented. Furthermore, AI can produce ‘hallucinations’: the AI generates text and states facts that are not actually true.
For researchers and lecturers
Generative AI also plays an important role in research. It is used, for example, to devise experimental designs, carry out studies, and gain an overview of the relevant literature. In order to use generative AI effectively and to be able to assess the risks, users should have a basic understanding of how AI works, what it can do, and, above all, what it cannot yet do.
Generative AI creates human-like content based on probabilities. It was trained using huge amounts of data. Based on these vast amounts of data, AI can recognise and replicate patterns. However, AI does not yet have a genuine understanding of the content. The content AI produces is always based on the training data. Consequently, AI may reproduce stereotypes, and content may be heavily mainstream-oriented. Furthermore, AI can produce ‘hallucinations’: the AI generates text and states facts that are not actually true.
For the administration
Generative AI can support administrative processes at the university, for example when drafting and revising texts, organising information, preparing emails, reports or proposals, and researching and summarising content. In order to be able to use generative AI effectively and assess the risks, users require a basic understanding: how does AI work, what can it do, and what are its limitations?
Generative AI creates content based on statistical probabilities. It can recognise patterns, suggest wording, and present information clearly, but it does not understand the content. The results must therefore be reviewed carefully. In the administration, members of staff must ensure, in particular, that personal and confidential data is handled with care, legal requirements are observed, and potential errors, distortions, or so-called ‘hallucinations’ are thoroughly reviewed. Please refer to the Guidelines on the Responsible Use of Artificial Intelligence (AI) at the University of Greifswald.
