kutuphane@fbu.edu.tr
0216 910 19 07
kutuphane@fbu.edu.tr
0216 910 19 07
Generative artificial intelligence is based on advanced machine learning models, known as deep learning algorithms, that simulate the learning and decision-making processes of the human brain. These models work by identifying and encoding patterns and relationships within massive datasets, and then using this information to understand users’ natural language requests or questions and generate relevant new content in response.
Artificial intelligence has been a popular topic in technology over the past decade; however, the emergence of generative AI particularly ChatGPT in 2022 brought AI to global attention and initiated an unprecedented wave of innovation and adoption (What is generative AI?, n.d.).
What is generative AI? (n.d.). Retrieved April 8, 2026, from https://www.ibm.com/think/topics/generative-ai
Our students can access the following AI tools completely free of charge, either in their standard versions or via their university email addresses, to facilitate their education and accelerate their research processes:
AI should not be used to directly generate assignments. Instead, it should be used to:
The goal is not “to let AI do the work,” but “to help you understand.”
Copying AI-generated content directly is considered plagiarism
Our university requires clear disclosure of AI usage.
For example:
AI may sometimes provide incorrect or fabricated information.
Therefore:
AI:
Therefore, questioning outputs is essential for academic development.
Do not input:
The APA Style team is developing official guidelines for citing large language models and generative AI tools such as ChatGPT. As of February 2024, provisional guidance and examples are available on the APA Style Blog.
In-text citation:
When ChatGPT was prompted with the question, “Is the left brain–right brain division real or a metaphor?”, it responded that although the two hemispheres have some specialization, labeling people as “left-brained” or “right-brained” is an oversimplification and a popular myth (OpenAI, 2023).
Reference Format:
Author of AI tool. (Year). Name of AI tool/model [Descriptor]. URL
Example:
OpenAI. (2023). ChatGPT 3.5 (May 12 version) [Large language model]. https://chat.openai.com/chat
How do I cite generative AI in MLA style? | MLA Style Center Published: March 17, 2023
MLA style is generally more flexible than APA or Chicago style.
When citing AI-generated content in MLA:
Format:
“Prompt description.” AI tool name, version, company, date, URL
Example:
“Examples of harm reduction initiatives.” ChatGPT, March 23 version, OpenAI, March 4, 2023, chat.openai.com/chat.
In-text citation:
(“Examples of harm reduction”)
Copyright is a type of intellectual property that applies to various creative works, including articles and books, blogs and websites, images and films, music, software, and datasets. Reproduction of copyrighted materials generally requires the copyright owner’s permission.
Copyright laws define who the copyright owner is, what is protected by copyright, how long it is protected, and what a user of a work may or may not do with a copyrighted work. Copyright laws vary from country to country. For example, a work protected in one country may not be protected in another, and certain uses of a work such as for educational or research purposes may be permitted in one country but not in another.
They are advanced software systems that perceive the digital environment, make decisions, and can act autonomously (on their own) to achieve a specific goal. Unlike LLMs (Large Language Models) that merely answer questions, they perform complex, multi-step tasks on behalf of humans or in collaboration with them.
It is the ability of computers and machines to mimic human-like cognitive skills such as learning, problem-solving, decision-making, perception, and natural language processing. It operates using intelligent algorithms that analyze data to recognize patterns and improve themselves by learning from this information.
Bias in artificial intelligence refers to the phenomenon where models reflect the biases present in the data used to train them. This can inadvertently cause certain viewpoints, demographic groups, or types of data to be overrepresented compared to others, potentially affecting the fairness of the output.
Big data refers to datasets that are too large, fast-moving, and complex (structured and unstructured) to be stored or analyzed using traditional data processing methods.
A chatbot is a software program that simulates conversations with human users via text or voice, using artificial intelligence (AI) or predefined rules.
Closed Source Large Language Model such as GPT-4 or Claude, are models that are owned by specific entities and have restricted access.
Cognitive computing is an artificial intelligence approach focused on mimicking human thought processes. It encompasses capabilities such as natural language understanding, pattern recognition, and learning.
Deep learning is a subfield of machine learning that uses multi-layer neural networks and is highly effective in tasks such as image recognition and natural language processing.
Embedding is the representation of text or images as multidimensional numerical vectors. It is used in fields such as semantic search and similarity detection.
Fine-tuning is the process of customizing a pre-trained model by retraining it for specific tasks or domains.
A foundational model is an AI model trained on large datasets and adaptable for various tasks.
Generative artificial intelligence is a type of AI that can create new content such as text, images, audio, or video based on patterns it learns from large datasets.
Guardrails are rules and filters used to ensure that artificial intelligence produces safe, ethical, and appropriate outputs.
A hallucination is a situation in which large language models generate responses that are factually incorrect or illogical due to limitations in their data and architecture.
Large language model is deep learning-based AI systems trained on vast amounts of text data that are capable of understanding, summarizing, translating, predicting, and generating content in human language. Typically utilizing the Transformer architecture, these models generate fluent responses in natural language by understanding the complex relationships between words.
The latent space is the space in machine learning and deep learning where high-dimensional complex data (images, texts, sounds) are transformed into a lower-dimensional, compressed, and abstract representation while preserving their essential features.
Machine learning is a subset of artificial intelligence (AI) that enables computers to learn, make decisions, and make predictions through experience without being explicitly programmed by using patterns in data and statistical methods. Algorithms optimize themselves using training data and improve their accuracy when presented with new data.
A multimodal model is an advanced artificial intelligence model capable of simultaneously processing, understanding, and establishing relationships among multiple different types of data (modalities) such as text, images, audio, video, and sensor data in the field of artificial intelligence.
Natural language processing is a field of artificial intelligence and computational linguistics that enables computers to understand, interpret, analyze, and generate human language (written or spoken). This technology, which makes human-computer interaction more natural, uses machine learning and deep learning techniques to extract meaningful insights from large text and audio datasets.
Artificial neural networks are a machine learning and deep learning technology developed based on the principles of the human brain, enabling computers to recognize and learn patterns in data and solve complex problems. As one of the fundamental building blocks of artificial intelligence, these networks consist of interconnected nodes called “neurons.”
Open-source large language models are models that can be freely used and modified, such as Mistral or LLaMA. Open-source language models.
Parameters are numerical values and internal configuration variables that a model learns from data during the training process and uses to transform inputs into outputs. These values function as a “repository of knowledge” that enables the model to recognize patterns in data and make predictions.
Pattern Recognition is the discipline of automatically detecting, classifying, and interpreting regularities, recurring structures, and similar features in raw data using artificial intelligence and machine learning algorithms.
Predictive analytics is a method of analyzing historical data using statistical modeling, data mining, and machine learning techniques to predict future outcomes, trends, and behaviors.
Prescriptive analytics is an advanced type of data analytics that not only predicts the future using data but also recommends the “best course of action” to achieve desired outcomes. By using artificial intelligence, machine learning, and mathematical modeling, it provides answers to the questions “What could happen?” and “What should I do to make it happen?”
Pre-training is the process by which an artificial intelligence or machine learning model is trained on very large and general datasets before being used for a specific task, allowing it to acquire fundamental knowledge, language structure, or feature extraction capabilities.
A prompt is a text-based command, question, or instruction that tells generative artificial intelligence models (such as ChatGPT, Midjourney, etc.) what to do. It acts as a “compass” used to obtain the most accurate and high-quality output from AI, consisting of detailed descriptions or inputs.
Prompt engineering is the discipline of optimizing, designing, and managing prompts in order to obtain the most accurate, creative, and high-quality outputs from artificial intelligence (AI) models (such as ChatGPT, Midjourney, etc.). It acts as a bridge between humans and AI, minimizing incorrect or irrelevant results and enabling the generation of targeted, specific responses or visuals.
Prompt tuning is a method of improving performance by optimizing prompts instead of retraining the model.
Quantum computing is a revolutionary computational model that solves complex problems difficult for classical computers by using the principles of quantum physics (superposition and entanglement). Instead of traditional bits (0 or 1), it uses “qubits,” which can exist as both 0 and 1 simultaneously, exponentially increasing processing speed and capacity.
Sentiment analysis is a natural language processing (NLP) technology that automatically determines the emotional tone of people in text-based data (such as reviews, social media, and emails) regarding products, services, or topics as positive, negative, or neutral.
Structured data refers to data that is organized according to a specific order, model, or schema and can be easily understood by machines. Examples include research datasets, metadata, and formats such as JSON-LD.
Synthetic data is “artificial” data generated digitally using artificial intelligence, algorithms, or simulations instead of being obtained from real-world events. It mimics the statistical properties and patterns of real-world data. Its main purpose is to protect data privacy, address data scarcity, and reduce costs while providing high-quality data, especially for training AI and machine learning models.
A transformer model is a neural network architecture that forms the foundation of modern artificial intelligence systems. It is a type of AI model that learns to understand and generate human-like text by analyzing patterns in large amounts of textual data.
The Turing Test is an artificial intelligence evaluation criterion developed by Alan Turing in 1950 to measure a machine’s ability to exhibit intelligent behavior equivalent to or indistinguishable from that of a human.
This document was translated by the DeepL translation tool. The content has been carefully reviewed and checked in terms of accuracy, terminology, and contextual consistency, and has been evaluated in line with critical thinking criteria. The author assumes full responsibility for the final content.
Atatürk mah. Ataşehir Bulvarı Metropol İstanbul, 34758 İstanbul
0216 910 19 07