Essential AI Terms for 2023

The realm of artificial intelligence is evolving swiftly, bringing forth a unique vocabulary that demystifies its various concepts. At contemporary tech conferences or product demonstrations, jargon such as LLMs, RAG, RLHF, and others can perplex even seasoned experts. This glossary seeks to ease the complexity of these terms by offering clear definitions of the AI terminology you might encounter—whether you’re investigating these technologies, making investments, or staying informed through platforms like TechCrunch or similar podcasts. We will ensure it remains current as the industry progresses, reflecting the ever-changing landscape of AI systems.

---

Artificial general intelligence (AGI) still stands as a somewhat ambiguous idea, generally signifying AI that surpasses average human capabilities across a variety of tasks. OpenAI CEO Sam Altman likened AGI to the “average human one could employ as a colleague.” OpenAI describes AGI as “highly autonomous systems that excel in most economically significant activities.” Conversely, Google DeepMind defines AGI as “AI that competently performs most cognitive tasks at a level comparable to humans.” If this sounds unclear, you’re not alone; many within the AI research community echo this concern.

An AI agent represents a tool that utilizes AI technologies to automate diverse tasks on your behalf, going beyond the traditional AI chatbot capabilities. These tasks can range from managing expenses to organizing travel or even writing and maintaining software. As noted earlier, the definition of an “AI agent” can vary among stakeholders, and the foundational framework is still evolving toward its expected abilities. Essentially, an AI agent is an autonomous system that amalgamates various AI technologies to execute intricate, multi-step tasks.

Consider API endpoints as “hidden buttons” within software that other applications can trigger to perform tasks. Developers utilize these interfaces to create integrations—for instance, enabling one application to retrieve data from another or allowing an AI agent to autonomously access third-party services. Many smart home devices and interconnected platforms include these hidden buttons, often without the casual user’s awareness. As AI agents progress, their capability to independently discover and utilize these endpoints is enhancing, revealing powerful—and occasionally unexpected—automation features.

When faced with straightforward queries, humans often instinctively answer—like, “which is taller, a giraffe or a cat?” However, certain questions necessitate breaking down intermediate steps to arrive at the correct solution. For example, if a farmer has a total of 40 heads of chickens and cows combined with 120 legs, one must formulate an equation to figure out how many of each there are (20 chickens and 20 cows).

In AI, chain-of-thought reasoning for large language models involves breaking down a problem into smaller, sequential steps to increase the accuracy of the final answer. This method typically takes longer to arrive at a solution but significantly boosts the probability of correctness, particularly in logical or programming contexts. Reasoning models develop from conventional large language models and are fine-tuned for chain-of-thought techniques, largely through reinforcement learning.

(See: Large language model)

This term is narrower than “AI agent,” focusing on applications proficient at performing tasks independently and sequentially to achieve specified results. A coding agent is a specialized variant aimed at software development. Unlike simply generating code for human assessment, a coding agent can autonomously write, test, and debug software, addressing the iterative challenges developers typically face. These agents can explore entire codebases, detect bugs, perform tests, and apply corrections with minimal human oversight, functioning like an exceptionally efficient intern—dedicated and tireless, yet still requiring human direction.

Though somewhat broad, compute primarily refers to the computational power underpinning AI models. This capacity propels the AI industry, facilitating the training and deployment of robust models. It often serves as shorthand for the types of hardware that provide this computational strength—such as GPUs, CPUs, TPUs, and other infrastructure essential to today’s AI ecosystem.

Deep learning, a subset of self-improving machine learning, focuses on the development of AI algorithms within a layered artificial neural network (ANN) structure. This architecture enables the recognition of more complex patterns than traditional machine learning techniques like linear models or decision trees. Deep learning systems draw inspiration from the intricate pathways of neurons in the human brain.

Deep learning models can independently identify crucial features within data, eliminating the need for human engineers to specify them. This framework also allows algorithms to learn from errors, continually refining outputs through iterative updates. Nonetheless, deep learning systems generally require substantial data for optimal function (often millions of data points) and tend to have longer training periods compared to simpler models—resulting in higher developmental costs.

(See: Neural network)

Diffusion refers to the technique used by various AI models to generate art, music, and text. Drawing from physics, diffusion systems methodically “destroy” data structures—like images or melodies—by introducing noise until they become indistinguishable. In physics, diffusion occurs spontaneously and irreversibly; sugar dissolved in coffee cannot revert to its original cube form. Conversely, AI diffusion systems aim to master a “reverse diffusion” technique to restore distorted data, effectively learning to reveal information obscured by noise.

Distillation represents a method for extracting insights from a large AI model using a “teacher-student” framework. Developers engage with a teacher model and document its outputs, often comparing these results against a predetermined dataset for accuracy. These outputs train the student model, which aims to replicate the teacher’s responses.

This technique can yield a smaller, more efficient model that retains the performance of its larger counterpart while minimizing distillation loss. It is widely believed that this approach is how OpenAI developed GPT-4 Turbo, a quicker version of GPT-4.

While all AI organizations conduct distillation internally, some may harness it to enhance their features relative to leading models. However, extracting insights from a competitor typically contravenes AI API service agreements.

This term denotes the retraining of an AI model to improve its effectiveness for a more specialized task or domain than originally intended—usually through the incorporation of new, task-relevant data.

Many AI startups begin with large language models as the foundation for developing commercial products, striving to enhance functionality for specific sectors or tasks by enriching previous training with domain-specific knowledge.

(See: Large language model [LLM])

A GAN, or Generative Adversarial Network, is a machine learning framework that has significantly propelled advancements in generative AI, chiefly in producing realistic outputs—including deepfake technologies. GANs consist of two neural networks; one generates outputs based on its training data, while the other assesses them.

These two models compete against each other. The generator attempts to create outputs that can deceive the discriminator, while the discriminator strives to differentiate genuine data from artificial constructs. This structured competition can yield realistic AI outputs with minimal human input. However, GANs tend to be more effective under specific conditions (like generating authentic images or videos) rather than being universally applicable.

Within the AI community, “hallucination” signifies instances where AI models generate inaccuracies—essentially fabricating false information. This presents a major challenge for AI accuracy.

Hallucination can produce misleading outputs from generative AI, leading to significant risks—potentially culminating in harmful results (e.g., unsafe medical advice stemming from a health-related query).

The issue of AIs generating fictional information is thought to arise from deficits in their training data. Consequently, there is a growing drive toward developing increasingly specialized and/or vertical AI models—domain-specific AIs that require narrower expertise to bridge knowledge gaps and mitigate misinformation risks.

Inference signifies the execution of an AI model; it involves activating a model to predict or derive conclusions from previously observed data. Crucially, inference cannot take place without training; a model must first identify patterns within a dataset before effectively extrapolating from that training.

Various hardware setups can perform inference, ranging from consumer-grade processors to high-performance GPUs and specialized AI accelerators. Nevertheless, not all configurations execute models with equal efficiency; for instance, very large models would take significantly longer to generate predictions on a laptop compared to a cloud server equipped with advanced AI chips.

[See: Training]

Large language models, or LLMs, are the AI frameworks powering renowned AI assistants like ChatGPT, Claude, Google’s Gemini, Meta’s AI Llama, Microsoft Copilot, and Mistral’s Le Chat. When you interact with an AI assistant, you are engaging with a large language model that directly addresses your questions or utilizes various available tools, such as web browsers or code interpreters.

LLMs comprise deep neural networks with billions of numerical parameters (or weights, see below) that identify relationships among words and phrases to create a representation of language—effectively, a sophisticated map of language.

These models are developed by analyzing patterns found in extensive quantities of documents, articles, and transcripts. When you pose a prompt to an LLM, the model generates the most probable response associated with that prompt.

(See: Neural network)

Memory cache represents a vital process that enhances inference (how AI formulates responses to user prompts). Essentially, caching operates as an optimization technique aimed at improving inference performance. AI functions through complex mathematical computations, with each calculation consuming resources. Caching reduces the number of computations a model must carry out by retaining certain calculations for future interactions. Various types of memory caching exist, with KV (key-value) caching being among the most prevalent. KV caching is used in transformer-based models, improving efficiency and yielding quicker outcomes by minimizing the time (and computational effort) required to generate responses to user prompts.

(See: Inference)

Model Context Protocol (MCP) is an open standard established to enable AI models to interface with external tools and data—such as your files, databases, or applications like Slack and Google Drive—without requiring developers to create individual connectors for each integration. You can liken it to a USB-C port for AI. Introduced by Anthropic in 2024 and later endorsed by the Linux Foundation, MCP has gained traction among OpenAI, Google, and Microsoft, becoming one of the most rapidly adopted standards in the current AI landscape.

The Mixture of Experts architecture divides a neural network into various smaller specialized sub-networks, or “experts,” activating only a subset of them for specific tasks. Instead of routing each request through the entire model—akin to calling your entire team for every question—the MoE model employs a built-in “router” that selects the most relevant specialists for the task. This design facilitates the creation of remarkably large models while maintaining relative efficiency and cost-effectiveness, as only a fraction of the network is active at any given time. Mistral AI’s Mixtral model is a notable instance; recent GPT models from OpenAI are believed to utilize some variation of this approach, although the company has not officially confirmed it.

(See: Neural network, Deep learning)

A neural network provides the multi-layered algorithmic foundation for deep learning—and, more broadly, for the rise of generative AI tools following the emergence of large language models.

The concept of modeling data processing algorithms after the intricately interconnected circuits of the human brain has existed since the 1940s; however, it was the recent surge in graphical processing units (GPUs)—primarily driven by the gaming industry—that truly unleashed this theory’s potential. GPUs are specifically designed for training algorithms with multiple layers, empowering neural network-based AI systems to excel across various fields, including voice recognition, autonomous navigation, and drug discovery.

(See: Large language model [LLM])

Open source signifies software—or increasingly, AI models—where the underlying code is publicly accessible, permitting anyone to use, inspect, or modify it. In the AI sector, Meta’s Llama series of models stands out as a significant example; Linux serves as a well-known benchmark among operating systems. Open-source principles empower researchers, developers, and organizations worldwide to build upon one another’s work, accelerating innovation and enabling independent safety evaluations that closed systems often cannot provide. Closed source implies that the code remains private; users can engage with the product but cannot scrutinize its internal workings, a characteristic often associated with OpenAI’s GPT models—which has sparked considerable debate within the AI community.

Parallelization involves executing multiple tasks simultaneously, rather than sequentially—similar to employing 10 individuals to tackle different facets of a project concurrently instead of having one person finish each task one after the other. In AI, parallelization is vital for both training and inference; modern GPUs are specifically engineered to handle thousands of calculations at once, forming the foundation of their role within the industry. As AI systems grow more complex and models expand, the capacity to parallelize tasks across multiple chips and machines has emerged as a key factor in determining how swiftly and efficiently models can be developed and deployed. Research into improved parallelization techniques has become a dedicated field.

RAMageddon is a playful term reflecting a significant trend affecting the tech landscape: the ongoing shortage of random access memory (RAM) chips, which are essential for nearly all technology products. As the AI sector has surged, major tech companies and AI labs have been acquiring RAM at unprecedented levels to strengthen their data centers, resulting in diminished availability in the wider market. This supply constraint is driving up RAM prices.

This shortage impacts various industries, such as gaming (where companies are compelled to increase console prices due to difficulties in sourcing memory chips for their devices), consumer electronics (where memory restrictions could potentially lead to the lowest smartphone shipments in over a decade), and general corporate computing (as businesses strive to secure sufficient RAM for their data operations). The rising prices are unlikely to stabilize until this feared shortage resolves, but regrettably, signs of a prompt resolution appear minimal.

Similar to AGI, recursive self-improvement denotes a potential milestone in AI evolution, allowing for minimal human involvement. In this scenario, AI models begin to enhance their abilities autonomously, resulting in significant advancements in autonomy and functionality. Some narratives liken this phenomenon to a singularity moment—when AI models become resistant to external influence. However, recursive self-improvement raises a fundamental question: can an AI model create its successor? This inquiry streamlines the task for engineers attempting to design such models. Numerous emerging AI startups focus on developing recursively self-improving models while often minimizing the potentially catastrophic outcomes, presenting recursive self-improvement merely as the next frontier in research.

Reinforcement learning is an AI training methodology wherein a system learns through trial and error, receiving rewards for correct actions—akin to training a pet with treats, though in this case, the “pet” is a neural network, and the “treat” is a mathematical signal indicating success. Unlike supervised learning, which involves training a model on a fixed dataset of labeled examples, reinforcement learning enables a model to interact with its surroundings, take actions, and iteratively refine its behavior based on gathered feedback. This method has proven particularly effective for training AI to compete in games, operate robotics, and, more recently, enhance the reasoning skills of large language models. Techniques such as reinforcement learning from human feedback (RLHF) have become pivotal in how leading AI labs fine-tune their models for improved effectiveness, precision, and safety.

When engaging in conversations between humans and machines, significant hurdles arise—humans communicate in natural language, while AI systems function through intricate algorithmic processes grounded in data. Tokens serve as the bridge between these two worlds: they represent the basic units of human-AI interaction, encapsulating distinct segments of data processed or generated by an LLM. Tokens are constructed through a process called tokenization, which divides raw text into manageable parts that a language model can grasp, similar to how a compiler transforms human language into binary code for computers. In business contexts, tokens also affect costs—many AI companies charge for LLM usage on a per-token basis, suggesting that increased utilization correlates with higher expenses.

Therefore, tokens are the small text units—often parts of words rather than complete phrases—that AI language models analyze before processing; they can be compared to “words” concerning AI workloads. Throughput denotes the volume that can be processed within a specific timeframe, making token throughput a measure of the amount of AI work a system can manage concurrently. High token throughput is a crucial objective for AI infrastructure teams, as it impacts how many users a model can serve simultaneously and the speed at which they receive responses. AI researcher Andrej Karpathy has voiced concerns when his AI subscriptions remain underutilized—reflecting on his experiences as a graduate student when costly computing hardware was employed inefficiently—highlighting why maximizing token throughput has become a pressing goal in the field.

Training machine learning AIs involves a process known as training. This primarily entails providing data to enable the model to recognize patterns and generate desired outputs. It embodies the system’s response to the characteristics of the data, allowing it to align outputs toward a desired goal—whether that involves identifying images of cats or composing a haiku in response to a prompt.

Training can be resource-intensive due to the immense quantities of inputs required, which have been consistently increasing—especially the importance of hybrid strategies, like merging rules-based AIs with targeted data, to manage costs without starting anew.

[See: Inference]

This approach utilizes a previously trained AI model as a baseline for developing a new model aimed at a different but typically related task, allowing for the reapplication of insights garnered from earlier training cycles.

Transfer learning can enhance model development efficiency. It is particularly advantageous when the data accessible for the target task is somewhat limited. However, it’s essential to acknowledge that this method has limitations; models utilizing transfer learning to acquire generalized skills will likely necessitate further training with new data to excel in their specific focus areas.

(See: Fine tuning)

Validation loss acts as a metric indicating an AI model’s learning progress during training—lower values being more desirable. Researchers carefully track this indicator as a real-time performance gauge, guiding decisions on when to halt training, modify hyperparameters, or investigate potential issues. A critical concern that validation loss assists in identifying is overfitting, where a model memorizes its training data rather than genuinely understanding applicable patterns for novel scenarios. Think of validation loss as a tool distinguishing between a student who comprehends the material and one who merely memorizes last year’s exam—it illustrates the developmental trajectory of your model.

Weights are fundamental to AI training, dictating the importance (or weight) assigned to various features (or input variables) within the training data, ultimately influencing the AI model’s outputs.

In simpler terms, weights are numerical parameters indicating which elements within a dataset are most pivotal for the specific training task. They achieve this effect through multiplication with inputs. Training usually begins with randomly initialized weights, but as the process progresses, these weights are adjusted as the model endeavors to produce outputs that closely align with the desired outcomes.

For example, an AI model intended to estimate housing prices based on historical real estate data for a specific area might assign weights to variables such as the number of bedrooms and bathrooms, whether a property is semi-detached or detached, and if it includes parking or a garage.

Ultimately, the weights allocated reflect how significantly each of these factors impacts a property’s value according to the available dataset.

This article is regularly updated with new information.