AI Glossary

AI Artificial Intelligence.
Using a computer to solve a problem with minimal human input. A broad field of computer science aiming to create intelligent
machines capable of mimicking human cognitive functions, such as learning, reasoning, problem-solving, and perception. AI
encompasses various subfields, including machine learning, deep learning, natural language processing, and robotics, with the goal
of enabling machines to perform tasks that typically require human intelligence

ASM Autoregressive Sequence Models.
Powers sequence understanding and prediction, refining AI accuracy.

Bard is a LLM from Google, now known as Gemini

Bing Chat An AI interface from Microsoft. Now known as CoPilot.
It is based on the OpenAI model, similar to ChatGPT, but with access to the Internet.

Chain-of-thought programming
Instructing the AI to run a sequence of commands. It generally gives better results than just telling it to do a high level requirement.
See Chain-of-thought prompting

ChatGPT A LLM from OpenAI.
Current version is ChatGPT 4.

Claude A LLM from Anthropic

Deep Learning:
A subset of machine learning where neural networks, mimicking the human brain, have multiple interconnected layers, enabling
them to automatically learn intricate patterns and representations from data.

Discriminative model
aims to understand text by labelling the data.

Gemini is a LLM from Google, formerly known as “Bard”

Gemma LLM from Google

Generative model
generates new samples based on the data that it was trained on. It involves artificial intelligence (AI) systems that create original
content, including text, images, video, or audio, based on user prompts.

GPT
is a custom version of ChatGPT that is used for a specific task. It can be simple or complex and can be shared with others.

Hallucinate
Sometimes when an AI tool doesn’t understand what is being asked or cannot find the answer, it will give an incorrect answer or
‘hallucinate’. Some people liken it to deliberately lying because it knows that you want an answer even if it doesn’t know the answer.

Hugging Face
Is a French American originated LLM. See https://huggingface.co/
See also Mistral

LLama 2
an open source LLM from Meta. See https://llama.meta.com/

LLM Large Language Model. For a good introduction watch https://www.youtube.com/watch?v=zjkBMFhNj_g

ML Machine Learning.
Using lots of examples to help a computer learn the rules. A subset of AI that focuses on the development of algorithms allowing
systems to learn and make predictions or decisions without being explicitly programmed.

Mistral French AI LLM model
See https://mistral.ai/news/announcing-mistral-7b/
See also Hugging Face

NLP Natural Language Processing
AI technology that enables machines to understand, interpret, and generate human-like language.

RAG Retrieval Augmented Generation.
Merges neural models with external data for richer AI interactions, e.g. adding company data to an existing LLM to support company
specific results. It enhances the accuracy and reliability of generative AI models with facts fetched from external sources, i.e. you
have a base model and then you add a specialised dataset with expert knowledge, e.g. from a company intranet site or specialised
database such as a legal or medical database.

Transformers
Transformer neural networks change an input sequence into an output sequence by learning context and tracking relationships
between sequence components.