On Wednesday, March 26, Amazon launched its new version of Alexa, its voice assistant, which is now powered by generative artificial intelligence to optimize its responses and include other features.

According to the company, generative artificial intelligence “will allow the popular voice assistant to have more personality, assess a user’s tone of voice, and even plan romantic dates.”

This version differs from the free version in that customers must pay Amazon a monthly fee of $19.99 to use Alexa+. Alexa+ will be free for Prime members, who already pay a monthly or annual fee for free shipping and other benefits.

During the launch, Amazon noted that Alexa+ “is capable of holding conversations with a more natural and human flow, and can learn more about a user—such as dietary preferences or allergies—the more it is used.”

Regarding its operation, it is known to be based on large language models that combine some developed by Amazon and others by Anthropic. It has “a model-agnostic system, allowing it to select the best AI model for the tasks it wants to complete,” according to Daniel Rausch, Amazon’s vice president of Alexa and Echo.

The monthly fee paid by customers will be used to cover operating costs, make the model more competitive, and continue investing in generative artificial intelligence.

M.Pino

Source: latimes

(Reference image source: Jonathan Borba on Unsplash)

Visit our news channel on Google News and follow us to get accurate, interesting information and stay up to date with everything. You can also see our daily content on X/Twitter and Instagram

La entrada Amazon unveiled the new Alexa+ powered by generative AI apareció primero en Bitfinance.

Facebook’s parent company, Meta, presented its latest developments in artificial intelligence (AI) tools for the metaverse, including BuilderBot, a system capable of building virtual universes from voice commands to an assistant.

“Building for the metaverse will require great advances in artificial intelligence“, said the company, which also discussed its research in this field as part of its Meta AI: Inside the Lab event.

Among the most notable new features is BuilderBot, an AI system showcased by Meta that allows users to generate items in the metaverse or import them from other sources using voice commands.

In a demo video in which CEO Mark Zuckerberg himself participated, Meta showed how, using voice commands led by “Hey, BuilderBot,” the system is capable of creating a tropical island in a minute: setting up the landscape (create a park, add a beach, add clouds), as well as adding things like tables, drinks, and even playing music.

The company also presented a series of AI tools with which it seeks to develop the future of the metaverse, starting with the purpose of creating a universal translator that works in real time.

Source: dpa

La entrada Meta would allow creating virtual universes with voice commands apareció primero en Bitfinance.

Google made improvements to its assistant that allow it to better understand the user’s context to perform tasks more efficiently, such as setting alarms and timers, improving the naturalness of conversations, and understanding and saying the names in the contact list more accurately.

The user will be able to help the assistant to “say and recognize” the names in his contact list in the same way that he pronounces them without the assistant being left with any recording of those pronunciations.

This process will allow you to better understand the user when they repeat the name, as well as to better pronounce it yourself. The tool will be available in English in the next few days, but Google assures in its blog that they hope to expand the service to other languages ​​”soon.”

Second, the tool now understands the context “more precisely” and understands what the user is trying to do when ordering something thanks to the improvement of its natural language understanding model. State-of-the-art BERT technology allows words to be processed in relation to one another, has made this breakthrough possible.

In this way, there is now “almost 100 %“ accuracy when programming alarms and timers, although the user expresses himself differently when programming them than when canceling them or cutting the phrase in half to modify the hour of the timer, for example.

The company ensures that in the future it will take this feature to other use cases so that the Google Assistant can continue to learn to better understand users. It is currently available in English for smart speaker alarms and timers in the United States only.

K. Tovar

Source: dpa

La entrada Google Assistant will add improvements in recognition apareció primero en Bitfinance.