The Free Software Foundation (FSF) is preparing a statement of criteria for free machine learning applications.

In this presentation the FSF will present the major considerations for defining what user freedom means in the rapidly developing field of machine learning, why it is different from defining software freedom, and why it matters.

• We will talk about why we use the term Machine Learning over Artificial Intelligence
• We will take a moment to discuss user freedom, what it means and why it matters
• We will discuss how machine learning applications differ from software when considering user freedom
• We will review what kinds of machine learning applications we should take into account when considering the broader issue
• We will take the question of what (outside of software) elements of the application need to be considered and explain the FSF's position on it (think training data, model parameters, weights)
• We will talk about how a machine learning application could have ethical reasons to remain nonfree, and how the FSF's criteria deal with this