Amazon has introduced a new generative AI model for its voice assistant, Alexa, aimed at enhancing user experience through improved conversation, real-world applications, personalization, personality, and trust.
The announcement was made at the company’s Devices & Services Fall Event at Amazon’s HQ2 in Arlington, Virginia, where Dave Limp, Senior Vice President of Amazon Devices and Services, and Rohit Prasad, Senior Vice President and Head Scientist of Amazon Artificial General Intelligence, took the stage. So far we’ve learned about the new Echo Show 8.
The new Alexa model focuses on five foundational capabilities:
During the event, Limp demonstrated the conversational capabilities of the new Alexa model. The assistant was able to engage in a fluid conversation about Limp’s favourite college football team, Vanderbilt, and even expressed its own preference for the Seattle Seahawks.
Prasad highlighted the technical innovations behind the new model, including a speech recognition system that adjusts to natural pauses and hesitations. The system also allows for more expressive and context-sensitive responses. A future update, referred to as “speech-to-speech,” aims to unify tasks like speech recognition and text-to-speech, offering a richer conversational experience.
Heather Zorn, Vice President of Alexa, also took the stage to announce new tools for developers. Starting next year, developers will be able to integrate their content and APIs with Amazon’s LLM, offering richer and more engaging experiences through a simple no-code solution. “It’s important we bring our partners along the journey with us. This thinking is in our DNA across Amazon,” said Zorn. She also noted that over one million brands have used Amazon’s tools to grow their businesses and that this approach will continue with the introduction of advanced LLMs.
Customers in the U.S. will soon have access to these new features through a free preview on existing Echo devices, including the very first Echo device shipped in 2014.
