Question: Create a glossary of terms that come up when using AI applications such as ChatGPT. The individual terms should be explained for people who are just starting to use AI applications but are familiar with computers and the internet. The explanations of the individual terms should not exceed 100 words and should be sorted alphabetically. Please explain words such as prompt, hyperparameter, bot, etc. in English. Please limit yourself to a maximum of 30 to 40 keywords.

ChatGPT: Here is a glossary of basic terms that frequently appear in connection with AI applications such as ChatGPT, tailored to users who are gaining their first experience with AI:

1. anomaly detectionThe process of identifying unusual patterns or data points that deviate from the norm, which may indicate potential problems or interesting anomalies.
2. API (Application Programming Interface)An interface that enables different software applications to interact with each other and exchange functions or data.
3. automationThe use of technology to perform tasks that would otherwise require human intervention, often using AI.
4. Autonomous systemsSystems that can operate without human intervention, such as self-driving cars.
5. Bias (distortion): Systematic and unfair distortions in data or AI models that can lead to erroneous or discriminatory results.
6. bot: Short form for robot. In AI, this often refers to programmes that perform automated tasks, particularly in the context of communication, such as answering questions.
7. chatbotsProgrammes that can conduct text or voice interactions with people, often used in customer service systems.
8. cloud computingThe provision of computer services (including AI capabilities) via the internet, allowing users to access powerful computing resources without having to directly own or manage them.
9. cluster analysisA technique in machine learning for organising data points into groups so that points in the same group have more similarities than points in different groups.
10. Computer visionA field of AI that focuses on enabling computers to see and interpret visual information.
11. data recordA collection of data used to train AI models. This can include text, images, videos or other formats.
12. deep learningA special method of machine learning that uses deep neural networks to recognise and interpret complex patterns in large amounts of data.
13. dialogue systemA computer programme that communicates with users via natural language, often in the form of conversations or dialogues.
14. edge computingA method in which data processing is brought closer to the location where the data is generated (e.g. on a mobile device instead of in a remote data centre) in order to improve response times and save bandwidth.
15. Ensemble learningThe approach of combining multiple models or algorithms to achieve better predictions than would be possible with individual models.
16. decision treeA modelling approach in which data is divided through a series of decisions, similar to the branches of a tree, in order to make predictions or classifications.
17. Ethics in AI: Considerations and guidelines designed to ensure that AI technologies are used responsibly and fairly.
18. Explainable AI (XAI)An approach in AI that aims to make the decision-making of AI systems comprehensible and transparent.
19. Fine-tuningA process in which a pre-trained model is further adapted to specific tasks or data sets in order to improve performance.
20. Generative modelAn AI model designed to generate content such as text, images or music based on what it has learned during training.
21. hyperparametersVariables that control the structure of an AI model (such as the number of layers in a neural network) and the learning process (such as the learning rate), which must be set before training.
22. interpretabilityThe ability to understand and explain how an AI model makes decisions or processes data.
23. AI (Artificial Intelligence): A field of computer science that deals with enabling machines to perform tasks that normally require human intelligence, such as understanding language or recognising patterns.
24. Artificial neural networks (ANN): Model structures in AI that are inspired by the functioning of the human brain and form the basis of many modern AI systems.
25. LLM (Large Language Model)Large language models are advanced AI systems trained to understand and generate human language. They are trained by analysing large amounts of text data and can perform a variety of tasks, including answering questions, translating between languages, summarising texts, and much more. A well-known example of a large language model is OpenAI's GPT (Generative Pre-trained Transformer).
26. Machine learningA subfield of AI that uses algorithms to learn from data and make decisions or predictions without explicit programming.
27. modelIn AI, this refers to the specific structure consisting of an algorithm and trained data to perform specific tasks.
28. Natural Language Generation (NLG)The process by which computers automatically generate text that is natural and understandable to humans.
29. Natural Language Understanding (NLU)The field of AI that deals not only with processing human language, but also with understanding its meaning and purpose.
30. Neural networkA computer-based model inspired by the human brain, consisting of interconnected units (neurons) that process data to recognise patterns and make decisions.
31. NLP (Natural Language Processing): An area of AI that deals with how computers can understand and generate human language.
32. prompt: Input text that a user sends to an AI system. The prompt serves as a trigger or question to which the AI responds or performs a task based on.
33. Reinforcement learning: An area of machine learning in which models learn to develop optimal strategies or behaviours through rewards for their actions.
34. Semantic WebAn extension of the current web in which information is structured in a way that enables computers to better understand and respond to the content of the data.
35. Speech RecognitionThe technology that enables computers to convert spoken words into text.
36. tokenIn AI, and specifically in language processing, a token often refers to the smallest unit of text that is processed during the training of a model. A token can be a word, part of a word, or even a single character, depending on the tokenisation process used. Tokenisation is an important step in converting text into a form that can be used effectively by AI models.
37. tokenisationThe process of breaking text down into smaller units, such as words or sentences, to make it easier to process.
38. transfer learningThe process of adapting a model that has been trained on one task to solve another, but related, task. This is particularly efficient for saving time and resources when training AI models.
39. Supervised learningAn approach in machine learning in which the model is trained using sample data and known answers to it.
40. X-Shot promptingThis term refers to a method in which a language model is given examples (known as „shots“) to learn how to respond to a new task or request. „X“ stands for the number of examples. Single-shot or few-shot are common formats. This enables the model to respond to new situations without extensive retraining or further training by generalising patterns from a few examples and applying them to new problems.

These terms can help to develop a broader understanding of various aspects and techniques of artificial intelligence and its application.

___________

This glossary may also be of interest outside of the series of articles, as a reference work or dictionary. That is why we have dedicated a separate article to the glossary.

Read the next article – Artificial Intelligence 5 – to find out how you can optimise your questions and tasks for the chatbot.

Here you can download the glossary.