Amazon has unveiled its latest AI innovation for its virtual assistant Alexa, which promises to make it more conversational and human-like.

The new feature, called “let’s chat”, will use a Large Language Model (LLM) that can generate natural language responses and messages for users.

The company announced the new feature at its HQ2 headquarters in Arlington, Virginia, USA, where it demonstrated how Alexa can now hold more engaging conversations, adjust its tone, and provide real-time information.

The feature will be available to existing Echo owners in the next few weeks as an early preview.

This latest model has been specifically optimised 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 LLM will enable Alexa to compose messages on behalf of users when asked, and to answer questions in more natural ways. For example, when Limp asked Alexa what its favourite sports team was, it said the Seahawks and followed up with stats and game times.

The LLM will also allow Alexa to understand conversational commands for smart home devices, such as “look spooky” to change lights or “there’s a mess in here” to launch a robot vacuum.

However, the presentation was not flawless, as CNBC reported that there were some “hiccups” during the demonstration, including Alexa lagging and questions being repeated.

Amazon is not the only tech giant that is investing in AI models that can generate natural language. Earlier this year, Amazon Web Services launched a series of AI products that help other companies develop their own chatbots and image-generation services. Microsoft and Google Cloud are also working on similar products, following the success of ChatGPT.