Menu
Log in
Forgot password

Member Login for Full Site Content

Log in


AI Overview & Key Concepts

AI & Machine Learning

Artificial intelligence (AI) is technology that mimics human thinking and capabilities by "perceiving" input, learning from recognized "rules," and making decisions.

Modern AI models "learn" complex hierarchies of rules governing their decisions and output through machine learning (ML). They process extraordinary amounts of data and identify patterns without explicit human programming. ML-enabled AI can do tasks like write, calculate, summarize and prioritize (and more) in ways that seem similar to what an average person can do.

AI models need guidance from real people. Without oversight, AI output is vulnerable to mistakes, oversimplification, bias, and other problems. Despite appearances or bold claims, it cannot replace human powers of perception, logic, creativity, or moral reasoning.


Types of Machine Learning

  • Supervised Learning: Algorithms train using a labeled dataset where each piece of data has a 'correct' output. Ex. image classification, spam detection, predicting housing prices.
  • Unsupervised Learning: Algorithms train using an unlabeled dataset, requiring the algorithm to find patterns within the data. Ex. Anomaly detection
  • Self-Supervised Learning: A subset of Unsupervised Learning where the computer creates its own labels. Ex. Predicting words, image colorization.
  • Semi-Supervised Learning: A combination of supervised/unsupervised learning. A small number of the provided data may be labeled.
  • Reinforcement Learning: The agent learns from their environment, while being reinforced to produce desired outcomes. Ex. Training autonomous vehicles, robotics.

Most Common Type of AI: GACPA Acronym

Generate - Text, images, video, and audio, including combinations to produce complete deliverables.

Automate/Control - Mechanize virtual or tangible tasks, such as traffic lights, safety sensors, scheduling, and self-driving cars, trains, etc.

Correlate/Predict/Discover - Answering what did, is, should, or could happen if...by simulating and processing information at a rate much faster than scientist-directed experiments. 

Problem Solve/Reason - Problem identification, analysis, evaluating solutions, and recommendations. 

Agent/Virtual Assistant - Your AI personal assistant that can perform complex virtual tasks.

Information from Demystifying Artificial Intelligence for You by Bruce Kinney is licensed under CC-BY-NC-SA 4.0 International.

Generative AI

Generative AI (GenAI) generates new content (language, images, and more) from user prompts based on patterns learned from large amounts of training data. GenAI is "a type of artificial intelligence (AI) that is able to create new content, such as text, images, music, or entire datasets, based on patterns and information it has learned from existing data. Unlike traditional AI that simply analyzes data, GenAI actively produces new material, simulating a level of creativity once thought unique to humans."1

1. GenAI in Teaching and Learning Toolkit by Gwen Nguyen is licensed under a CC-BY-NC 4.0 International license.

GenAI Models vs. GenAI Tools

Model: a neural network like an AI 'engine'. Ex. Large Language models (like GPT-4, GPT-4o, Gemini, Claude 3.5), Diffusion models

Tool: the software used to interact with the model. Ex. Microsoft Copilot for Web, (uses the GPT-4 & GPT-3.5 models), ChatGPT Free (uses the GPT-4o & GPT-3.5 models), DALL-E (uses a diffusion model).

Information from Different Generative AI Options by University of Sydney under a CC-BY-NC 4.0 license.

Resources:

Large Language Models

Large Language Models (LLMs) are a type of GenAI that produce language. Popular examples include ChatGPT, Claude, Microsoft Copilot, and Gemini. LLMs allow users to prompt (or, give instructions/ask questions) an AI tool using natural language. In other words, you can "talk" to the AI the same way you talk to another person. When the tool responds, you can use its conversational capabilities to ask further question, request edits, and more.

Common uses of LLMs include:

  • Generating text or code

  • Summarizing

  • Translating

  • Editing grammar, structure, or tone

  • Providing information (a common but NOT recommended use of LLMS due to false or misleading outputs)

Image Generators: Firefly, DALL-E 3, Microsoft Copilot, Ideogram

Other Tools: Perplexity, Canva, Glasp, Goblin Tools, Gamma, Grammarly, MagicSchool.ai, Diffit, Almanack, Slidesgo

Resources:



TEST


Follow us on social media:


State Partner for Missouri

MLA is a 501(c)3 non-profit organization.

PO Box 684

Jefferson City, MO 65102

MLAC2C@molib.org

Powered by Wild Apricot Membership Software