The science of making machines intelligent, so they can recognize patterns and get really good at helping people solve specific challenges or sets of challenges.

Artificial intelligence is in use when a computer program makes a decision on a prediction—this could be through straightforward rule-based systems or heuristic methods, such as “if rain, then umbrella.” In machine learning on the other hand, the decisions are learned.

Formally, machine learning is a subfield of artificial intelligence. However, in recent years, some organizations have begun using the terms artificial intelligence and machine learning interchangeably.

A person who writes code to generate images, sounds and writing. Creative technologists work in diverse fields, such as design, research, art, filmmaking, music, and creative writing, and they often build their own interactive digital tools. 

Machine learning provides many exciting tools for creative technologists.

Read about some of AMI's and Google Arts & Cultures collaborations with creative technologists here.

AI researchers approach problems by representing data in probabilistic models. These models automatically recognize and classify information. They can also generate images, sounds, and text. 

 A common example of a generative model is the GAN (generative adversarial network) known as BigGAN.

Practically speaking, a generative model does either of the following:

(1) A model that creates (generates) new examples from the training dataset. For example, a generative model could create poetry after training on a dataset of poems. The generator part of a generative adversarial network falls into this category.

(2) A model that determines the probability that a new example comes from the training set, or was created from the same mechanism that created the training set. 

For example, after training on a dataset consisting of English sentences, a generative model could determine the probability that new input is a valid English sentence.

In biological brains, neurons form networks of living tissue that transmit nerve impulses. Digital neural networks inspired by organic brains transmit information to recognize and generate data, and to automate decision making. 

 Researchers layer cascades of these neural nets to create deep learning systems.

Latent space is a multidimensional field of compressed data. In latent space, similar data points are grouped closer together according to their most relevant features of regarding the algorithm’s query. 

Latent space allows a dataset to be clustered, visualized, and analyzed according to the features that the machine learning algorithm has been trained to analyze. 

A subfield of artificial intelligence that comprises techniques and methods to develop AI, by getting computer programs to do something without programming super-specific rules.

There are many ways to get a computer program to learn something. Most relevant to our list is supervised learning, in which the program learns to make predictions—like your commute time—from hundreds of thousands of examples. 

Other popular approaches are unsupervised, semi-supervised, and reinforcement learning, but we’ll leave those for another day (or you can learn the technical details on your own with our Machine Learning Glossary for developers).

Letters and words, musical notes, and audio waveforms can all be represented as a sequence of characters, which can be learned by neural nets.

 Following an input (“The cow jumped over ”), a text generator predicts the sequence most likely to follow (“the moon.”) GPT-3 and BERT are prominent sequence generating neural net models.

Supervised machine learning methods rely on large corpora of labelled data called datasets or training sets. These define a neural net’s expertise.

For example, a neural net trained on labelled photos of dogs (like DeepDream) can recognize and generate images of dogs. Datasets are constructed and labelled by people, and inherit the biases of their creators.