Machine Intelligence

Machine Intelligence (MI) is the science and engineering of making intelligent machines. The ultimate goal of this field is to make machines reach human-level intelligence. As humans, we take for granted a huge number of capacities and functions that our brains and body are able to seamlessly perform, but are extremely hard to translate to the world of machines. In this talk I plan to discuss the challenges of building machines that can understand, learn, act, and interact in ways that approach human intelligence.

Although machines boast impressive computing speeds and memory, they famously lack many abilities considered basic to human intelligence. To approximate these abilities, machines would have to process sensory data, text and other media and use it to reason, make predictions, and perform actions in the real world. More importantly, they would have to learn from their experiences as well as possess a certain amount of common sense knowledge. For example, a child may never be explicitly told that a coat is used for keeping warm in winter, but she may nonetheless act on such knowledge when it starts to snow. Comparably intelligent behavior might require a machine to have concepts corresponding to winter, snow, warm, and coat, as well as a rich interconnected set of structures that help tie these together and guide its behavior in the context of snow.

Progress toward true machine intelligence would enable a wealth of applications across many domains. This talk will focus on highlighting the main challenges in this area and discuss some of the promising avenues toward new solutions.

Neha is a senior software engineer at Google, where she has worked since 2010. She graduated from Carnegie Mellon University with a Masters in Software Engineering and is currently working with the Machine Intelligence team at Google, contributing to the effort towards understanding common sense knowledge from text and other media. Her main areas of interest include using cognitive methods for language understanding and reasoning.

In her previous role, she worked on improving the quality of news-­related results on Google.com and on Google News. Prior to that, she worked on understanding social interaction signals to improve the quality of advertisements on mobile devices.