Can Intelligence Be Appropriated
Kate Lebedeva · 2020
In the modern world, artificial intelligence (AI) has become an overarching phenomenon, the “new electricity” that will transform each and every area of life. This chapter aims to provide an overview of existing, imperfect approaches to ownership over AI and suggests the possible solutions. AI is a complex phenomenon. It includes several components - such as algorithms, training data sets and trained models. A machine learning algorithm defines how a computer learns to link an input (e.g. a picture of an animal) to a specific output (e.g. whether it is a “cat” or a “dog”). Training data sets of the most sophisticated AI systems consist of millions of data inputs (e.g. photos of cats or videos of street traffic). Trained models are the statistical data generated through learning. They are best described as “black boxes” that can predict an output based on the input.