Skip to Main Content

Getting Started with AI

Learn how to use generative AI effectively and safely.

Types of AI

The term "AI" generally refers to a subset of artificial intelligence called Generative AI. All of the popular AI models out today that you may have heard about are a type of Generative AI. There is no true Artificial General Intelligence yet, which would be machines that can fully think for themselves. Technology is not yet at that stage. This can be contrasted to Simple AI, which includes simple input/output structure like Alexa or Siri answering an easy question but becoming confused for more complex ones. The ranking of AI based on intelligence is: Simple AI > Generative AI > Artificial General Intelligence.

Types of Generative AI:A chart of language models by intelligence. ChatGPT is in the lead followed by Claude.

Large Language Models (Chatbots)
  • Designed to generate human-like text
  • Produces quality that depends on its training data
  • Includes, but not limited to, the following examples:
    • ChatGPT
    • Gemini
    • Claude
AI Art Generators
  • Designed to create images from a prompt
  • Includes, but not limited to, the following examples:
    • Stable diffusion
    • Midjourney
    • Adobe Firefly

How does AI Work?

Generative AI. . .

  • Learns from and mimics large amounts of data to create text, images, music, videos, code, and more content based on inputs or prompts.
  • Lacks true intelligence and functions purely from pattern prediction.
  • Requires large datasets for training the generative models.

This means that generative AI essentially predicts the next word or pixel that should come next. Everything it does is a probability-based guess, which means it will not always be correct.

In the shortest terms, generative AI is a highly sophisticated autocomplete feature.

Read a more in-depth explanation of how AI works at One Useful Thing (https://www.oneusefulthing.org/p/thinking-like-an-ai).