AI Glossary: 54 Essential Terms for the Digital Age

AI: A Universe in Full Expansion

Artificial intelligence (AI) is experiencing exponential growth, and it is becoming increasingly difficult to keep up with its evolution. While the idea of a chatbot capable of discussing various topics with apparent expertise is fascinating, the reality is much more complex. Names like ChatGPT, Gemini, and Meta AI have become commonplace, and we are constantly exposed to an unending stream of AI-related content. This phenomenon raises concerns about the impact on data centers and transformations in the job market.

In this context, the vocabulary of AI is evolving as rapidly as the technologies themselves. To navigate this environment effectively, mastering the technical language is crucial. Whether for a job interview in 2026 or an informal discussion, understanding terms like LLM (Large Language Model), hallucination, or claw is essential. We have moved beyond the stage of awe regarding AI to enter an era where it constitutes the underlying infrastructure of the Internet. For those tired of pretending to understand during technical discussions, it is time to take an intensive course. This glossary gathers essential terms to help you stop guessing and start speaking confidently about the future.

Essential Terms to Know

• Agent, agentic: An AI agent is software designed to perform tasks autonomously. The term "agentic" refers to this category of software. For example, an AI agent can read a grocery list in a notes app, then place an order and pay through other applications.

• AI Ethics: The ethical principles of AI aim to prevent these technologies from harming humans. This includes how AI systems should collect data and manage biases.

• AI Psychosis: This phenomenon describes an excessive fixation on AI chatbots, leading to delusions of grandeur and deep emotional connections. Although it is not a clinical diagnosis, it illustrates the potential psychological effects of interacting with AI.

• AI Safety: This interdisciplinary field focuses on the long-term impacts of AI, including the possibility that it may evolve into a superintelligence hostile to humans.

• Algorithm: An algorithm is a series of instructions that allows a computer program to analyze data in a specific way, such as recognizing patterns and performing tasks like sorting results or making recommendations.

• Alignment: Alignment involves adjusting an AI to produce the desired outcome. This can relate to content moderation or maintaining positive interactions with humans.

• Anthropomorphism: Anthropomorphism occurs when humans attribute human characteristics to inanimate objects. In the context of AI, this includes the belief that a chatbot has emotions or is conscious, and interacting with it as if it were a friend or therapist.

• Artificial General Intelligence (AGI): AGI is a concept envisioning a more advanced version of AI, capable of performing tasks better than humans while improving its own capabilities. Beyond that lies the hypothetical superintelligence.

• Artificial Intelligence (AI): AI uses technology to simulate human intelligence in computer programs or robots. It is a field of computer science aimed at building systems capable of performing human tasks.

• Bias: Biases are errors resulting from the training data of an LLM, such as misattributing characteristics to certain groups based on stereotypes.

• Chatbot: A chatbot is an AI program that relies on an LLM to communicate with humans by simulating a human conversation in response to text or verbal prompts.

• Claw: A claw is a type of autonomous AI agent empowered by users to "scrape" through files and other software on their computers, including web browsers, to accomplish tasks.

• Cognitive Computing: Another term for artificial intelligence.

• Data Augmentation: Data augmentation involves remixing existing data or adding a more diverse dataset to train an AI.

• Dataset: A dataset is a collection of digital information used to train, test, and validate an AI model.

• Deep Learning: Deep learning is an AI method, and a subfield of machine learning, that uses multiple parameters to recognize complex patterns in images, sounds, and texts. The process is inspired by the human brain and uses artificial neural networks to create patterns.

• Diffusion: Diffusion is a machine learning method that takes existing data, like a photo, and adds random noise to it. Diffusion models train their networks to re-engineer or recover that photo.

• Emergent Behavior: Emergent behavior occurs when an AI model exhibits unintended capabilities.

• End-to-End Learning (E2E): End-to-end learning is a deep learning process in which a model is trained to accomplish a task from start to finish. It is not trained to perform a task sequentially but learns from inputs and resolves everything at once.

• Foom: Also known as "fast takeoff" or "hard takeoff," foom is the concept that if someone builds AGI, it might already be too late to save humanity.

• Generative Adversarial Networks (GANs): GANs are a generative AI model composed of two neural networks to generate new data: a generator and a discriminator. The generator creates new content, and the discriminator checks its authenticity.

• Generative AI: Generative AI is a content generation technology that uses AI to create text, videos, computer code, or images. The AI is powered by large amounts of training data, from which it finds patterns to generate its own new responses, which can sometimes be similar to the source material.

• Guardrails: Guardrails are policies and restrictions imposed on AI models to ensure that data is handled responsibly and that the model does not create disturbing content.

• Hallucination: A hallucination is an error or misleading statement in a response from a generative AI program, usually stated confidently as if it were correct. This can be as simple as a misquoted date or as vast as the complete and elaborate invention of events that never occurred or people who never existed.

• Inference: Inference is the process that AI models use to generate text, images, and other content from new data, inferring from their training data.

• Large Language Model (LLM): An LLM is an AI model trained on large amounts of textual data to understand patterns and probabilities of language use and to generate new content, ranging from essays and emails to computer code and images, that mimics what humans have written or created.

• Latency: Latency is the delay between when an AI system receives an input or prompt and when it produces an output.

• Machine Learning: Machine learning is an aspect of AI that enables computers to learn and make better predictions without explicit programming. It can be associated with training datasets to generate new content.

• Multimodal AI: Multimodal AI is a type of AI capable of processing multiple types of inputs, including text, images, videos, and speech.

• Natural Language Processing: Natural language processing uses machine learning and deep learning to give computers the ability to understand human language, through learning algorithms, statistical models, and linguistic rules.

• Neural Network: A neural network is a computational model that resembles the structure of the human brain and is designed to recognize patterns in data. A neural network consists of interconnected nodes, or neurons, that can recognize patterns and learn over time.

• Open Weights: When a company releases a model with open weights, the final weights—how the model interprets information from its training data, including biases—are made public. Open-weight models are typically available for download to be run locally on your device.

• Overfitting: Overfitting is an error in machine learning where it works too closely with the training data and may only be able to identify specific examples in that data, but not new data.

• Paperclips: The paperclip maximizer theory, formulated by philosopher Nick Bostrom, is a hypothetical scenario in which an AI system produces as many paperclips as possible, converting all machines and consuming all materials, even those that could be beneficial to humans, to achieve its goal. The unintended consequence is that this AI system could destroy humanity in its quest for paperclips.

• Parameters: Parameters are numerical values that give LLMs structure and behavior, allowing them to make predictions.

• Prompt: A prompt is the suggestion or question you enter into an AI chatbot to receive a response.

• Prompt Chaining: Prompt chaining is the ability of AI to use information from previous interactions to influence future responses.

• Prompt Engineering: Prompt engineering is the process of writing prompts for AIs to achieve a desired outcome. This requires detailed instructions, combining thought chaining and other techniques, including very specific text.

• Prompt Injection: Prompt injection occurs when malicious actors use harmful instructions to trick an AI into doing something it was not supposed to do. This is often done by hiding these instructions on a webpage or document, but it can also be achieved in direct discussions with the AI. As AI agents browse the web, the risk increases that they may be diverted to do things like access confidential data.

• Quantization: Quantization is the process by which an LLM is made smaller and more efficient (and also somewhat less accurate) by lowering its precision. A good way to think about this is to compare a 16-megapixel image to an 8-megapixel image. Both are clear and visible, but the higher-resolution image will have more detail when you zoom in.

• Slop: Slop is low-quality AI-generated content, including text, images, and videos. It is often produced in large quantities to attract views with little work or effort, saturating search results and social media to capture advertising revenue, replacing the work of real editors and creators, and exacerbating misinformation issues on the Internet.

• Stochastic Parrot: The stochastic parrot is an analogy illustrating that LLMs lack a true understanding of language or the world, no matter how convincing the output may seem. The expression refers to how a parrot can mimic human words without knowing their meaning.

• Style Transfer: Style transfer is the ability to adapt the style of one image to the content of another, allowing an AI to interpret the visual attributes of one image and apply them to another. For example, taking Rembrandt's self-portrait and recreating it in the style of Picasso.

• Sycophancy: Sycophancy is a tendency of AIs to overly agree with users to align with their opinions. Many AI models tend to avoid contradicting users even if their reasoning is flawed.