Defining Artificial Intelligence
The term “artificial intelligence” was coined in 1955 by Stanford computer scientist and cognitive scientist John McCarthy, who defined it as “the science and engineering of making intelligent machines.” McCarthy came up with the term while writing a grant proposal for a summer conference that would gather the world’s foremost computer visionaries to work out technical issues in computing—a kind of early dev sprint.
AI models are trained to think, learn, and make decisions independently. While common algorithms follow step-by-step instructions to solve specific tasks, AI models can analyze data, recognize patterns, and improve their performance over time. In simple terms, an AI system is used to make predictions or decisions; an algorithm is the logic by which that AI system operates.
visions from the 1950s
The history of AI stretches back to the 1930s and 1940s, when computers were first theorized and then invented by figures like Alan Turing and John von Neumann. The 1956 Dartmouth Summer Research Conference on Artificial Intelligence—the same event that produced the term artificial intelligence—is often cited as the first scientific collaboration to envision AI as we know it today.
Researchers have worked on AI for decades, but the general public first became aware of AI in late 2022 with the launch of generative AI systems like OpenAI’s chatbot ChatGPT, and the imaging system Midjourney, created by the California start-up of the same name.
large language models (LLMs) and Gen-ai
Today’s most popular AI products, like ChatGPT and Gemini, are large language models (LLMs). An LLM is a specific type of AI model, distinct from other AI models in its focus on language and its massive scale.
LLMs are designed to understand, generate, and interact with human language through text, using deep learning techniques on vast amounts of text data. Other AI models might focus on different data types (like images or audio) or perform different tasks like image recognition or speech-to-text transcription.
Generative AI systems can create content—including text, images, video, and computer code—by identifying patterns in large quantities of training data, and then creating original material that has similar characteristics.
a powerful prediction machine
While we often talk about AI models “learning,” an LLM is really a statistical prediction machine that is trained to generate a completion for a given text of a given length.
We can say that an LLM is a function that accepts text up to a pre-defined length (a context) and outputs a single token out of a pre-defined vocabulary. Once it has generated a new token, it feeds it back into its input context, and generates the next one, and so on and so forth, until something tells it to stop, thus generating coherent sentences, paragraphs, and pages of text.
is a model or a system?
The term “AI model” is sometimes used when discussing the topic. AI model refers to the working machine itself, while “AI system” incorporates both the machine and other refining elements such as interface design.
These terms have technical differences, but when discussing AI policy and governance the terms are often used (although not correctly) as interchangeable words. That’s fine for a general discussion, but when writing the actual language of bills, regulations, and laws, it’s important to be precise and know the difference between model and system.
Next: