This list provides a comprehensive overview of crucial AI concepts, ensuring clarity and understanding of each AI term with Definitions and Explanations.
1. Machine Learning (ML)
A subset of AI where systems learn from data to make predictions or decisions: It involves training algorithms on datasets, enabling them to improve over time.
2. Artificial General Intelligence (AGI):
AI possesses cognitive abilities equal to or greater than those of humans. AGI is a significant focus for research, raising both exciting possibilities and ethical concerns.
3. Generative AI:
A type of AI capable of creating new content, such as text, images, or code. Examples include tools like ChatGPT and Google’s Gemini, which are trained on extensive datasets.
4. Hallucinations:
Are errors in generative AI responses where the system confidently produces incorrect or nonsensical information due to limitations in its training data?
5. Bias:
Systematic prejudices in AI outputs, arising from the data on which the AI is trained, reflect societal biases and lead to inaccurate or unfair results.
6. Vector:
A vector is a mathematical object that encodes a length, direction, position, or even a change in some mathematical framework or space. A vector space is a collection of objects closed under addition and scalar multiplication rules. A simplistic representation of a vector might be an arrow in a vector space with an origin, direction, and magnitude (length).
7. AI Model:
A computational system trained on data to perform specific tasks or make decisions autonomously.
8. Large Language Models (LLMs):
A class of AI models designed to process and generate human-like text, exemplified by models such as OpenAI’s Claude.
9. Diffusion Models:
AI models that generate images from text prompts by learning to reverse the process of adding noise to images.
10. Foundation Models:
Versatile generative AI models trained on extensive datasets can support multiple applications without task-specific training.
11. Frontier Models:
Next-generation AI models, which are still in development, promise significantly enhanced capabilities compared to current models but raise potential risks.
12. Training:
The process through which AI models learn from data, refining their ability to recognise patterns and make predictions.
13. Parameters:
The internal variables of an AI model that influence how it converts inputs into outputs are crucial for its predictive accuracy.
14. Natural Language Processing (NLP):
The branch of AI enables machines to understand and generate human language, facilitating interactions like those with ChatGPT.
15. Inference:
The act of generating outputs from an AI model, such as producing a response to a query.
16. Tokens:
AI models analyse and generate language using segments of text (words, parts of words, or characters).
17. Neural Network:
A computer architecture inspired by the human brain enables machines to learn complex patterns from data.
18. Transformer:
A neural network architecture that uses attention mechanisms to understand the relationships between sequence elements is pivotal in generative AI.
19. RAG (Retrieval-Augmented Generation):
A method that enhances AI outputs by allowing models to pull in external information, improving the accuracy of responses.
AI Hardware
20. Nvidia’s H100 Chip:
A leading GPU designed for AI training, known for its efficiency in handling complex AI workloads.
21. Neural Processing Units (NPUs):
Specialised processors that optimise AI tasks on devices, enhancing performance for features like speech recognition.
22. TOPS (Trillion Operations Per Second):
A benchmark indicates AI chips’ processing power for performing AI inference tasks.
AI Applications
23. OpenAI / ChatGPT:
A popular AI chatbot recognised for its conversational abilities and wide-ranging applications.
24. Microsoft / Copilot:
An AI assistant integrated into Microsoft products, leveraging OpenAI’s GPT models.
25. Google / Gemini:
Google’s AI assistant encompasses various AI models across its services.
26. Meta / Llama:
An open-source large language model developed by Meta.
27. Apple / Apple Intelligence:
AI features are incorporated into Apple products, enhancing user experience through tools like ChatGPT in Siri.
28. Anthropic / Claude:
An AI company is creating models like Claude, with significant backing from major investors.
29. xAI / Grok:
An AI venture founded by Elon Musk focuses on large language models.
30. Perplexity:
An AI-powered search engine known for its innovative but controversial data practices.
31. Hugging Face:
A platform provides developers and researchers access to various AI models and datasets.