The World Wide Web (WWW) and WWW browsers have permeated our lives and revolutionized the way we get information and entertainment, socialize, and do business.
Researchers at Stanford University have created a new version of the World Wide Web (WWvW) that allows people to fully navigate using new tools that make developing voice-based agents easier and cheaper. Is proposing. Using voice.
About 90 million Americans already use smart speakers to stream music and news, perform tasks such as grocery ordering, booking scheduling, and lighting control. But at least in the United States, two companies basically control these voice gateways to the voice web. Amazon is a pioneer of Alexa. And Google has developed the Google Assistant. In effect, the two services are walled gardens. These oligopolies create a great imbalance, making technology owners prefer their products to those of their rivals. They control what content is available and what fees are charged to act as an intermediary between the enterprise and its customers. On top of that, the unique smart speaker puts your privacy at risk. Eavesdropping on conversation As long as they are connected.
Stanford Team led by Professor of Computer Science Monica Lamb and Stanford Open Virtual Assistant Lab (OVAL) has developed an open source privacy protection virtual assistant called Genie and a cost-effective voice agent development tool that can provide an alternative to its own platform.Scholars also hosted a workshop on November 10th to discuss their work and propose a design for the World Wide Voice Web (Watch). Full event).
What is WWvW?
Like the World Wide Web, the new WWvW is decentralized. The organization publishes information about voice agents on its website. This information can be accessed from the Virtual Assistant. According to Ram, in WWvW, voice agents are like web pages, providing information about services and applications, and virtual assistants are browsers. These voice agents can also be made available as chatbots or call center agents, and can also be accessed from a computer or phone.
“WWvW has the potential to reach even more people than WWW, including those who aren’t tech savvy, who can’t read or write well, or who can’t even speak a written language,” Lam says. For example, Krispiev, an assistant professor of computer science at Stanford University, and Musadumbuya and Lisa Einstein, graduate students, Working on the development of voice technology for three African languages It can help bridge the gap between illiteracy and access to valuable resources, including agricultural information and health care. “Unlike Amazon and Google-led commercial voice web, which is only available in some markets and languages, decentralized WWvW is voice information and services for all uses in any language, including education and other humanitarian purposes. Bringing to society. There are great financial benefits, “says Lam.
Why haven’t these tools been created before? The Stanford University team said: Creating voice technology is very difficult. Amazon and Google have invested enormous amounts of money and resources to provide their assistants with AI natural language processing technology, and have hired thousands of people to annotate their training data. “The technology development process is expensive and very laborious, which creates a barrier to entry for anyone looking to provide a commercial-grade smart voice assistant,” says Lam.
Unleash the genie
For the past six years, Lam has worked with Giovanni Campagna, a PhD student at Stanford University and a professor of computer science. James RandyWhen Christopher Manning, Professor of Computer Science and Linguistics, at OVAL has developed a new voice agent development methodology with sample efficiency that is two orders of magnitude higher than current solutions. The open source Genie Pre-trained Agent Generator they created offers dramatic cost and resource savings in developing voice agents in a variety of languages.
Interoperability is an important component that allows devices to interact seamlessly with each other, Lam said. At the heart of Genie technology is a distributed programming language created for virtual assistants called ThingTalk. Enable interoperability of multiple virtual assistants, web services, and IoT devices. Stanford is currently offering the first course of ThingTalk. Conversational virtual assistant with deep learningThis fall.
As of today, Genie has pre-trained agents for the most popular voice skills such as music playback, podcasts, news, restaurant recommendations, reminders, timers and more, supporting over 700 IoT devices. These agents are public and can be applied to other similar services.
Worldwide Voice Web Conferencing
The OVAL team has these concepts Workshop focusing on the World Wide Voice Web November 10th.
The conference will be attended by academia and industry speakers with expertise in machine learning, natural language processing, computer-human interactions, and IoT devices, with panelists building a voice ecosystem, pre-trained agents, and more. We talked about the social value of the voice web. The Stanford team also gave a live demonstration of the genie.
“We want others to participate in building the World Wide Voice Web,” says Ram, a faculty member at Stanford University. Human-centered Artificial Intelligence Laboratory.. “The original World Wide Web grew slowly at first, but once it became popular, we couldn’t stop it. We hope we can see the same with the World Wide Voice Web.”
Genie is an ongoing research project funded by the National Science Foundation, Alfred P. Sloan Foundation, Verdant Foundation, and Stanford HAI.