Amazon held its annual press event to show off its new hardware devices today. It also used the event to talk about its own plans for developing generative AI features.
In its live blog on the event, Amazon revealed it is developing a new Large Language Model in combination with its Alexa digital assistant. It will be available in preview form soon on all of Amazon's Echo devices for US residents.
Our latest model has been specifically optimized for voice and the things we know our customers love—like having access to real-time information, efficiently controlling their smart home, and getting the most out of their home entertainment.
The live blog included a description of a demo of the Alexa-based generative AI chatbot. Retiring Amazon hardware head Dave Limp sometimes paused to address the media in the audience, but was able to return to chatting with Alexa right where they left off. There's no more need to say "Alexa" to start a conversation, and Amazon's new voice recognition tech will adjust "your natural pauses and hesitation to deliver a more free-flowing conversation."
Beyond today's demo, Amazon says it is working on another generative AI model. The blog stated:
Instead of first converting a customer’s audio request into text using speech recognition, and then using an LLM to generate a text response or an action, and then text-to-speech to produce audio back—this new model will unify these tasks, creating a much richer conversational experience.
In 2024, Amazon will allow third-party developers to access its new LLM to create new experiences with Alexa. Some of those companies are already working on those new generative AI features. Amazon stated:
We’re working with BMW Group to develop conversational in-car voice assistant capabilities, which provide a natural way of interacting with and getting to know your car including quick how-to instructions and answers on your vehicle’s features.
Other companies are working on ways to use Alexa AI for other things, like creating songs or having natural conversations with AI-based characters.
Even more features are coming for Alexa-based devices. They include Call Translation for Echo Show products. Amazon says it will caption Alexa calls in real-time. It will launch later in 2023 in 10 different languages.