Data & AI

Recognize AI as a computing tool that adapts its functions by acquiring knowledge from organized data inputs and outputs. e.g., AI tools improve their tasks by comparing outputs to correct answers such as a photo-sorting app checks if its ‘cat’ labels match human-provided tags, then updates its sorting rules to reduce mistakes

Recognize that many inputs and outputs can be organized into a structure that is easily readable by a machine (e.g., data-table).

Describe in plain language how AI uses and builds upon data in multiple ways. e.g., AI systems identify patterns in data by processing thousands of input-output pairs, and the system adjusts its internal mathematical model to minimize error, enabling it to predict outputs for new inputs such as a spam filter

Identify how issues in data, such as bias, missing data, and errors, can affect the output of an AI tool and the training of an AI tool from the input-output pairs it learns from. e.g., If AI only sees pictures of cats in sunlight, it would fail to recognize cats in shadows

Describe the basic mathematical features of an AI model in terms of independent variables (e.g., inputs), dependent variables (e.g., outputs), and predictors or weights (e.g., slopes of many variables). e.g., AI models use math to weigh inputs, such as a music recommendation model might calculate:(play_count × weight₁) + (listen_duration × weight₂) + (skip_count × weight₃) = recommendation_score, and weights are adjusted automatically to minimize mismatches between predicted and actual user preferences.

Describe and explore how it is possible for data in a variety of formats (e.g., images) to be translated into organized, numerical information for an AI model to process.

Identify how biases in training data can lead to biases in AI models by directly affecting predictors or weights. e.g., If AI only sees pictures of cats in sunlight, it would fail to recognize cats in shadows

Identify and label a simple prediction algorithm or equation for a very basic AI prediction model. e.g., a basic AI model’s equation looks like: Prediction = (weight₁ × input₁) + (weight₂ × input₂), such as a college admission model might weight GPA (input₁) and test scores (input₂) to output an acceptance likelihood, and training involves automatically adjusting weights to match historical data.

Understand that algorithms use cost functions to measure errors and adjust predictions.

Recognize that some AI tools can be used to explore complex data with many variables.

Recognize the types of problems that are ideal for using an AI tool to analyze complex data.