Decoding Key AI Terms: AGI, LLMs, and More

Artificial intelligence (AI) is transforming the world while inventing an entirely new language to describe its workings. Spend five minutes reading about AI, and you will encounter terms like LLMs, RAG, RLHF, and a dozen others that can make even the brightest minds in the tech industry feel uneasy. This glossary is our attempt to remedy that. We update it regularly as the field evolves, so consider it a living document, much like the AI systems it describes.

Artificial General Intelligence (AGI)

Artificial General Intelligence, or AGI, is a nebulous term. It generally refers to an AI that is more capable than the average human in many, if not most, tasks. OpenAI's CEO, Sam Altman, described AGI as the "equivalent of a median human you could hire as a colleague." Meanwhile, OpenAI's charter defines AGI as "highly autonomous systems that outperform humans at most economically valuable work." Google DeepMind's understanding differs slightly from these two definitions; the lab considers AGI to be "an AI at least as capable as humans in most cognitive tasks." Confused? No worries — the experts at the forefront of AI research are too.

AI Agent

An AI agent refers to a tool that uses AI technologies to perform a series of tasks on your behalf — beyond what a more basic AI chatbot could do — such as filing expenses, booking tickets or a table at a restaurant, or even writing and maintaining code. However, as we explained earlier, there are many moving parts in this emerging space, so "AI agent" can mean different things to different people. The infrastructure is also still being developed to realize its envisioned capabilities. But the basic concept involves an autonomous system that can rely on multiple AI systems to execute multi-step tasks.

API Endpoints

Think of API endpoints as "buttons" on the back end of software that other programs can press to make it work. Developers use these interfaces to create integrations — for example, allowing one application to retrieve data from another, or enabling an AI agent to directly control third-party services without a human manually operating each interface. Most smart devices and connected platforms have these hidden buttons, even if ordinary users never see or interact with them. As AI agents become more capable, they are increasingly able to find and use these endpoints on their own, opening up powerful — and sometimes unexpected — possibilities for automation.

Chain of Thought

Faced with a simple question, a human brain can respond without much thought — things like "which animal is bigger, a giraffe or a cat?" But in many cases, you often need a pen and paper to find the right answer because there are intermediate steps. For example, if a farmer has chickens and cows, and together they have 40 heads and 120 legs, you might need to write a simple equation to find the answer (20 chickens and 20 cows).

In an AI context, chain of thought reasoning for large language models means breaking down a problem into smaller, intermediate steps to improve the quality of the final outcome. This usually takes longer to arrive at an answer, but the answer is more likely to be correct, especially in a logical or coding context. Reasoning models are developed from traditional large language models and optimized for chain of thought reasoning through reinforcement learning.

Coding Agent

This is a more specific concept than an "AI agent," meaning a program capable of acting autonomously, step by step, to achieve a goal. A coding agent is a specialized version applied to software development. Rather than simply suggesting code to a human for them to review and paste, a coding agent can write, test, and debug code autonomously, handling the iterative and trial-and-error work that typically consumes a developer's day. These agents can operate across entire codebases, spotting bugs, running tests, and pushing fixes with minimal human supervision. Think of it as hiring a very fast intern who never sleeps and never loses focus — although, as with any intern, a human must always review the work.

Computation

Somewhat of a catch-all term, computation generally refers to the vital computing power that enables AI models to function. This type of processing fuels the AI industry, giving it the ability to train and deploy its powerful models. The term is often shorthand for the types of hardware that provide computing power — things like GPUs, CPUs, TPUs, and other forms of infrastructure that form the backbone of the modern AI industry.

Deep Learning

A subset of self-improving machine learning in which AI algorithms are designed with a multi-layer artificial neural network (ANN) structure. This allows them to make more complex correlations compared to simpler machine learning systems, such as linear models or decision trees. The structure of deep learning algorithms is inspired by the interconnected pathways of neurons in the human brain.

Deep learning AI models are capable of identifying important features in data themselves, rather than requiring human engineers to define those features. The structure also supports algorithms capable of learning from their mistakes and, through a process of repetition and adjustment, improving their own outcomes. However, deep learning systems require a lot of data points to produce good results (millions or more). They also generally take longer to train compared to simpler machine learning algorithms — thus, development costs tend to be higher.

Diffusion

Diffusion is the technology at the heart of many AI models generating art, music, and text. Inspired by physics, diffusion systems "slowly destroy" the structure of data — for example, photos, songs, etc. — by adding noise until nothing remains. In physics, diffusion is spontaneous and irreversible — sugar dissolved in coffee cannot be restored to cube form. But AI diffusion systems aim to learn a "reverse diffusion" process to restore destroyed data, acquiring the ability to recover data from noise.

Distillation

Distillation is a technique used to extract knowledge from a large AI model using a "teacher-student" model. Developers send queries to a teacher model and record the outputs. The responses are sometimes compared to a dataset to verify their accuracy. These outputs are then used to train the student model, which is trained to approximate the teacher's behavior.

Distillation can be used to create a smaller, more efficient model based on a larger model with minimal distillation loss. This is likely how OpenAI developed GPT-4 Turbo, a faster version of GPT-4.

While all AI companies use distillation internally, it may also have been used by some AI companies to catch up with leading models. Distillation from a competitor generally violates the terms of service of AI APIs and chat assistants.

Fine-tuning

This refers to the additional training of an AI model to optimize performance for a task or domain more specific than what was previously the focus of its training — typically by feeding in new specialized data (i.e., task-oriented).

Many AI startups take large language models as a starting point to build a commercial product but seek to enhance utility for a target sector or task by complementing previous training cycles with fine-tuning based on their own expertise and domain-specific knowledge.

Generative Adversarial Network (GAN)

A GAN, or Generative Adversarial Network, is a type of machine learning framework that underpins some significant developments in generative AI regarding the production of realistic data — including (but not limited to) deepfake tools. GANs involve the use of a pair of neural networks, one of which relies on its training data to generate an output that is passed to the other model for evaluation.

The two models are essentially programmed to try to outdo each other. The generator tries to get its output past the discriminator, while the discriminator works to spot artificially generated data. This structured competition can optimize AI outputs to make them more realistic without requiring additional human intervention. While GANs work best for narrower applications (like producing realistic photos or videos), they are not designed for general-purpose AI.

Hallucination

Hallucination is the industry term for AI models that invent information — literally generating incorrect information. Obviously, this is a huge problem for AI quality.

Hallucinations produce GenAI outputs that can be misleading and could even pose real-world risks — with potentially dangerous consequences (think of a health query that returns harmful medical advice).

The issue of AIs fabricating information is seen as a consequence of gaps in training data. Hallucinations contribute to a push towards increasingly specialized and/or vertical AI models — that is, domain-specific AIs that require narrower expertise — as a means to reduce the likelihood of knowledge gaps and decrease the risks of misinformation.

Inference

Inference is the process of executing an AI model. It is allowing a model to make predictions or draw conclusions from previously seen data. To be clear, inference cannot occur without training; a model must learn patterns in a dataset before it can effectively extrapolate from that training data.

Many types of hardware can perform inferences, ranging from smartphone processors to powerful GPUs to custom-designed AI accelerators. But not all can execute models equally. Very large models would take ages to make predictions on, say, a laptop compared to a cloud server with high-end AI chips.

Large Language Model (LLM)

Large language models, or LLMs, are the AI models used by popular AI assistants, such as ChatGPT, Claude, Google’s Gemini, Meta’s Llama, Microsoft’s Copilot, or Mistral’s Le Chat. When you chat with an AI assistant, you are interacting with a large language model that processes your request directly or with the help of various available tools, such as web browsing or code interpreters.

LLMs are deep neural networks.