Amazon announced a preview of an upgraded Alexa that will be powered by a generative AI large language model. Here’s what we know. 
If you’ve been wondering when voice assistants would get the generative AI treatment, Amazon has your answer.
On Wednesday, the retail giant previewed an evolved version of Alexa, powered by generative AI. Soon, Amazon’s voice assistant technology will be powered by a proprietary large language model (LLM) that’s « custom-built and specifically optimized for voice interactions, and the things we know our customers love—getting real-time information, efficient smart home control, and maximizing their home entertainment, » said the announcement.
Ever since OpenAI’s ChatGPT kicked off a generative AI frenzy almost a year ago, techies have wondered if and when voice assistants would get the much-needed upgrade. Integrating LLMs with voice assistants seemed like a perfect use case. But the process is more complicated than deploying new code or updating the software, since voice assistants use less dynamic methods of AI. With current voice assistants, machine learning and natural language processing are applied to a finite database of words and phrases. Alternatively, LLMs are able to generate new information and build on top of their knowledge to become smarter over time. So essentially, Alexa has the same name but entirely new inner workings.
Some of the new Alexa features highlighted in the announcement include: lower latency, sensors within Amazon Echo devices to understand non-verbal cues, integration with third-party APIs including fictional character app Character.AI. Overall, the new Alexa will be able to understand context, carry over information from previous conversations, and become more personalized to your household the more you use it.
Since this is just a preview, Amazon didn’t share much detail about the new Alexa’s timeline. But it did note that there will be a free preview for Alexa customers in the U.S. soon.
