It is thought that Amazon is constructing a separate Large Language Model (LLM) to strengthen its voice assistant, ‘Alexa’. The intention is to totally utilize Alexa, which has lost its competitiveness attributable to the emergence of ChatGPT, in business.

Enterprise Beat, a specialized media outlet, reported on the twenty ninth (local time) that Amazon CEO Andy Jassy announced this through a performance announcement and promised to strengthen AI overall.

In accordance with this, CEO Jassy said, “Alexa is already getting used by lots of of hundreds of thousands of endpoints reminiscent of shopping, entertainment, and smart homes.

As well as, he explained that Amazon had put a number of effort into LLM through the years. He announced that he had developed his own chip dedicated to AI called ‘Trainium’ to learn the LLM called ‘Inferentia’, and now each the LLM and the chip have developed a second version.

Specifically, it was emphasized that this LLM won’t be developed for a broad purpose like ChatGPT or Bard, but shall be specialized for Alexa and Amazon businesses, reducing hallucinations or misuse problems of existing chatbots.

As such, Amazon announced its intention to separate LLM for Alexa and LLM for business services.

On the 14th of last month, Amazon announced plans to launch ‘Bedrock’, a generative AI cloud service for enterprises, including Amazon’s language model ‘Titan’ and startup AI21’s language model ‘Jurassic-2’. You should utilize Antropic’s language model ‘Claude’ and Stability AI’s image generation AI model ‘Stable Diffusion’.

“Bedrock is a sexy model optimized for enterprise customers,” said CEO Jassy. “Unlike Bedrock, LLM for Alexa will not be intended to be released for general use.

Reporter Lim Dae-jun ydj@aitimes.com