News and Insights

Generative AI Glossary

January 18, 2024

What follows is a list of terms related to Generative AI. Intended to serve as a reference guide that defines and explains the ‘language of of Generative AI’, the purpose of this document is to help:

ELIMINATE CONFUSION and ensure a common understanding of the language of Generative AI for FINN Partners employees, be they beginners or experts.
EDUCATE AND INFORM all FINN Partners employees by enabling them to familiarize themselves with the specialized language and concepts related to Generative AI.
FACILITATE COMMUNICATION across the entire FINN Partners network by providing a standardized vocabulary and language specific to Generative AI.
SUPPORT THE CREATION OF DOCUMENTATION AND RESEARCH by ensuring consistency and clarity in the usage of terms and concepts.
EASE ONBOARDING AND TRAINING for FINN Partners and its teams as we explore and implement Generative AI tools and work to bring training sessions to our colleagues and clients.
ASSIST IN OUR COMPREHENSION OF TECHNICAL CONTENT as we access technical articles, papers, or manuals related to Generative AI.
PROMOTE BEST PRACTICES by defining and standardizing terms and fostering a shared knowledge base.
KEEP US CURRENT AND FLUENT by incorporating new terms, advancements, or changes in the field of Generative AI.

(I.) CORE TERMS

ALGORITHM

A series of instructions that allows a computer program to learn and analyze data in a particular way, such as recognizing patterns, to then learn from it and accomplish tasks on its own.

ALIGNMENT

The process of tweaking an AI to better produce the desired outcome.

ARTIFICIAL INTELLIGENCE (AI)
Computer systems’ ability to perform tasks characteristic of human intelligence, like reasoning, learning, and creativity.

CHATBOT
A program that communicates with humans through text that simulates human language.

COGNITIVE COMPUTING
Another term for artificial intelligence.

COMPUTER VISION

AI’s capability to interpret visual information from images and videos, used for object detection, image segmentation, and more.

DATA AUGMENTATION

Inspired by the human brain and using artificial neural networks to create patterns, this subfield of machine learning leverages multiple parameters to recognize complex patterns in pictures, sound, and text.

DEEP LEARNING

Advanced neural networks with multiple layers, learning complex patterns in large data sets. Widely used in image and speech recognition, and natural language processing.

DIFFUSION

A method of machine learning that takes an existing piece of data, like a photo, and adds random noise (i.e., re-engineering or recovering that photo).

END-TO-END LEARNING
A deep learning process in which a model is instructed to perform a task from start to finish rather than accomplish it sequentially.

EXPLAINABLE AI BY DESIGN
Incorporating transparency and understanding into AI systems from the development stage.

FEDERATED LEARNING

Collaborative machine learning across decentralized data sources, preserving privacy.

GENERATIVE AI
A content-generating technology that uses AI to create text, video, computer code or images. The AI is ‘fed’ (or trained on) vast amounts of data and finds patterns to generate its own novel responses.

HALLUCINATION
A response from Generative AI to a prompt that produces an answer that while incorrect is stated with confidence.

MACHINE LEARNING
A subfield of AI where algorithms learn and improve from data over time, without explicit programming for each task.

MULTIMODAL AI

AI systems capable of processing and generating diverse data types — like text, images, voice, and video — often simultaneously, to provide a more comprehensive output or interaction.

NATURAL LANGUAGE PROCESSING

AI’s ability to understand and process human language, encompassing tasks like translation, sentiment analysis, and text generation.

NEURAL NETWORKS
Models composed of interconnected nodes that process data, inspired by the human brain’s neurons.

NEURO-SYMBOLIC AI

Combining symbolic reasoning with deep learning for more interpretable and reliable AI models.

REINFORCEMENT LEARNING

Training AI systems through trial and error, using rewards to guide them towards desired outcomes. Applied in robotics, game playing, and dynamic decision-making.

TRAINING DATA
The datasets used to help AI models learn, including text, images, code or data.

TRANSFORMER MODEL
A neural network architecture and deep learning model that learns context by tracking relationships in data, like in sentences or parts of images. Instead of analyzing a sentence one word at a time, it can look at the whole sentence and understand the context.

TURING TEST
Named after famed mathematician and computer scientist Alan Turing, it tests a machine’s ability to behave like a human. The machine passes if a human can’t distinguish the machine’s response from another human.

WEAK AI
AI that’s focused on a particular task and can’t learn beyond its skill set. Most of today’s AI can be categorized as weak AI.

(II.) TRENDING TERMS

AGENTS
Autonomous programs in AI designed to perform tasks and make decisions on behalf of users, often learning and adapting from interactions to improve over time.

ARTIFICIAL GENERAL INTELLIGENCE (AGI)

A concept that suggests a more advanced version of AI than we know today, one that can perform tasks better than humans while also teaching and advancing its own capabilities.

EFFECTIVE ACCELERATION
promotes unbridled technological advancement, especially in AI, free from regulatory or ethical constraints.

EFFECTIVE ALTRUISM
Using AI with evidence-based approaches to maximize societal benefits.

EMERGENT BEHAVIOR
When an AI model exhibits unintended abilities.

FOOM
Also known as ‘fast takeoff’ (or ‘hard takeoff’), the concept that if someone builds an AGI that it might already be too late to save humanity.

RETRIEVAL AUGMENTED GENERATION
Where language models enhance their responses by accessing and incorporating information from external data sources, such as the web, for more accurate and up-to-date outputs.

SUPERINTELLIGENCE
AI with intelligence far beyond human capability, capable of self-improvement and potentially unpredictable outcomes.

TECHNOLOGICAL SINGULARITY
A hypothetical point where AI could surpass human intelligence, leading to significant societal changes.

(III.) CREATIVITY TERMS

AI-ASSISTED CREATIVITY
A Collaborative approach where AI and humans work together to push the boundaries of creative expression.

AI COPILOT
A conversational interface that uses large language models to support users in various tasks and decision-making processes across multiple domains within an enterprise environment (i.e., Microsoft 365).

BARD
Sometimes referred to as ‘Google Bard’, this AI chatbot functions similarly to ChatGPT but pulls information from the current web, whereas ChatGPT is limited to data published up until 2021 and isn’t ‘connected’ to the Internet (note: this is only true of the free version of ChatGPT).

CHATGPT
Created by OpenAI, ChatGPT is an AI chatbot that uses large language model technology to generate content based on patterns it has learned.

GENERATIVE ADVERSARIAL NETWORKS
A model duo where one generates data and the other discerns it from real data, enhancing the realism in generated content.

LARGE LANGUAGE MODELS

AI systems trained on vast text datasets, capable of high-quality text generation and translation.

(IV.) ETHICS AND EXPLAINABILITY TERMS

AI ETHICS
Principles aimed at preventing AI from harming humans, achieved through means like determining how AI systems should collect data or deal with bias.

ALGORITHMIC JUSTICE

Efforts to address biases and inequalities in AI algorithms, promoting fairness for all.

BIAS

With regard to large language models, errors resulting from training data that can result in falsely attributing certain characteristics to certain races or groups based on stereotypes.

ETHICAL CONSIDERATIONS
An awareness of the ethical implications of AI and issues related to privacy, data usage, fairness, misuse and other safety issues.

EXPLAINABLE AI

Techniques that make AI decision-making processes transparent and understandable.

HUMAN-IN-THE-LOOP

Integrating human judgment with AI decision-making, ensuring optimal outcomes and alignment with human values.

RESPONSIBLE AI

Designing, developing, deploying, and using AI in a way that is ethical, transparent, accountable, and aligned with human values.

(IV.) SAFETY AND SECURITY TERMS

AI SAFETY
An interdisciplinary field that’s concerned with the long-term impacts of AI and how it could progress suddenly to a super intelligence that could be hostile to humans.

ALIGNMENT PROBLEM

Keeping AI systems within ethical boundaries and aligned with human values.

EXISTENTIAL RISKS

The long-term dangers posed by advanced AI, like superintelligence, requiring proactive risk management

ROBUSTNESS AND ADVERSARIAL AI

Strengthening AI systems against attacks designed to produce incorrect outputs.