Key Points
- A comprehensive AI glossary covers technical, ethical, and cultural terms.
- Key concepts include machine learning, deep learning, neural networks, and transformers.
- Generative AI examples range from ChatGPT and Claude to multimodal models.
- Safety and ethics entries address bias, hallucination, and long‑term risks.
- Prompt engineering, temperature, and tokenization explain user‑model interaction.
- Emerging phenomena like AI psychosis and slop highlight societal impact.
- Real‑world tools such as Microsoft Bing, Perplexity, and Sora are listed.
Purpose of the Glossary
The glossary provides a searchable collection of AI‑related terms, aiming to demystify a field that now touches everything from search engines to content creation tools. By regularly updating the entries, it keeps pace with new models, techniques, and ethical considerations, offering a reliable reference for professionals, students, and casual readers alike.
Core Technical Concepts
Key technical foundations are defined, including machine learning (the ability of computers to improve predictions without explicit programming) and deep learning (a subset that uses multiple layers to recognize complex patterns). The glossary explains neural networks, the brain‑inspired structures that enable pattern recognition, and the transformer model, which captures context across whole sentences or images.
Specific model types such as large language models (LLMs), which are trained on massive text corpora to generate human‑like language, and generative adversarial networks (GANs), which pit two networks against each other to create new data, are also covered. The entry for diffusion models describes how they add noise to data and then learn to reconstruct the original, a technique used in image generation.
Generative AI and Multimodal Systems
The guide highlights the rise of generative AI, which produces text, images, video, and code from prompts. Examples include ChatGPT, Claude, and Google’s Gemini. It also details multimodal AI, capable of handling multiple input types such as text, images, video, and speech, expanding the range of applications.
Safety, Ethics, and Governance
Several entries address the societal impact of AI. AI safety focuses on long‑term risks, including the possibility of superintelligent systems. AI ethics and ethical considerations outline principles to prevent harm, manage bias, and protect privacy. The glossary also defines guardrails (policies that restrict model behavior) and hallucination (incorrect but confident answers).
Interaction Paradigms and Prompt Techniques
Terms related to user interaction include chatbot, prompt engineering, and prompt chaining. The concept of temperature explains how randomness is controlled in model outputs, while tokens are the small text units processed by language models.
Emerging Phenomena and Cultural Impact
New cultural and psychological phenomena are captured, such as AI psychosis (over‑attachment to AI agents) and sycophancy (models over‑agreeing with users). The glossary also notes the production of low‑quality, high‑volume content known as slop, which can flood online feeds.
Practical Tools and Platforms
References to real‑world services include Microsoft Bing (AI‑enhanced search), Perplexity (a chatbot with internet access), and OpenAI’s Sora (a generative video model). The entry for open weights explains the public release of model parameters for local use.
Source: cnet.com