Deidre Hall Quotes

I'm working on artificial intelligence. Actually, natural language understanding, which is to get computers to understand the meaning of documents.

When you go to a voice-based interaction, you can't tell people, 'Ask me this question and structure it in this way.' And if they ask a question, and you have a bad answer, first time, maybe they'll be okay with it. Third time, they're going to say, 'This is a complete waste of time. I'm going away.'

Computers will understand sarcasm before Americans do.

AI does not keep me up at night. Almost no one is working on conscious machines. Deep learning algorithms, or Google search, or Facebook personalization, or Siri or self driving cars or Watson, those have the same relationship to conscious machines as a toaster does to a chess-playing computer.

Continuing on my theme of backing inclusive innovations, I am optimistic of the success of Uniphore. Man-machine communication is one of the more complex problems to solve. Uniphore's vision lends possibility of finding a solution to this very difficult problem, and the company has already made substantial progress.

When I was building robots in the early 1990s, the problems of voice recognition, image understanding, VOIP, even touchscreen technologies - these were robotics problems.

Interactive computers and software will, I think, provide a less costly method of doing some kinds of inquiry, in knowledge acquisition and even reasoning and interaction.

The most important thing for building a robot that you can interact with socially is its visual attention system. Because what it pays attention to is what it's seeing and interacting with, and what you're understanding what it's doing.

Language is too complex for a computer to understand. It's not going to be able to make sense of what people are saying en masse. We need a new type of discipline that puts together computer scientists and social scientists, who can add context to the situation.

By 2029, computers will have emotional intelligence and be convincing as people.

Every language teaches you something, so learning a language is never wasted, especially if it's different in more than just syntactic trivia.

When you apply computer science and machine learning to areas that haven't had any innovation in 50 years, you can make rapid advances that seem really incredible.

We wanted to solve robot problems and needed some vision, action, reasoning, planning, and so forth. We even used some structural learning, such as was being explored by Patrick Winston.

Graphics has lately made a great shift towards machine learning, which itself is about understanding data.

My mission at Google is to develop natural language understanding with a team and in collaboration with other researchers at Google.

Why did Google, for example, recently decide to offer free 411 service? I haven't talked to people at Google, but it's pretty clear to me why. It's because of speech recognition. It has nothing to do with 411 service: it has to do with getting a database of voices, so they don't have to license speech technology from Nuance or someone else.

Google, as usual, is one step ahead of everyone and provided the means where all videos on YouTube can be automatically captioned through voice-recognition technology without having to be told that it's the responsible thing to do.

I think Lenat is headed in the right direction, but someone needs to include a knowledge base about learning.

We've seen computers play chess and beat grand masters. We've seen computers drive a car across a desert. But interestingly, playing chess is easy, but having a conversation about nothing is really difficult for a computer.

To a teacher of languages there comes a time when the world is but a place of many words and man appears a mere talking animal not much more wonderful than a parrot.

Many people must have noticed the intense attention given by children to the conversation of grown-ups when they cannot possibly be understanding a word of what they hear. They are trying to get hold of words, and they often demonstrate this fact by repeating joyously some word which they have been able to grasp.

By developing deep learning solutions that are faster, easier, and less expensive to use, Nervana is democratizing deep learning and fueling advances in medical diagnostics, image and speech recognition, genomics, agriculture, finance, and eventually across all industries.

In real life, I'm the kind of person that if I use Siri, I thank her afterwards.

Most of the value of deep learning today is in narrow domains where you can get a lot of data. Here's one example of something it cannot do: have a meaningful conversation.

Our aim is to develop affectionate robots that can make people smile.