Amazon is developing a new chatbot called “Metis,” which could be available as early as September, according to a report from Business Insider based on an internal document and anonymous sources close to the project. This chatbot will be powered by a new AI model called Olympus, instead of the Titan model that Amazon previously launched.
What is Metis and How Does it Work?
Metis will use an approach called Retrieval-Augmented Generation (RAG). According to Amazon, this method “redirects the language model (LLM) to retrieve relevant information from authorized and predetermined knowledge sources. This allows organizations to have greater control over the generated text and provides users with transparency on how the response is generated.”
In essence, RAG allows systems to retrieve data from external sources like APIs and document repositories, rather than relying solely on preloaded information. This data can be updated without needing to retrain the model, making it easier to provide more accurate and up-to-date responses.
The Team Behind Metis
The development of Metis is being led by Amazon’s CEO, Andy Jassy, and the chief scientist of the AGI team and senior vice president, Rohit Prasad. The team includes many veterans of Alexa AI and leverages technology developed for the upcoming “Remarkable Alexa” voice assistant.
Launch and Competition
Amazon plans to launch Metis in September, coinciding with its product launch event. However, there are concerns about whether this launch will be too late to compete with other major players in the AI field. OpenAI launched ChatGPT in November 2022 and Google introduced Bard (now known as Gemini) in March 2023. Although Amazon’s Titan model has not been as powerful as its competitors, the company is working on offering business-focused options to attract more customers.
Conclusion
Amazon’s new chatbot Metis represents a significant effort by the company to advance in the field of artificial intelligence and offer a competitive product. With RAG technology, Metis promises to deliver more accurate and controlled responses, which could be a major advantage for users and organizations seeking advanced AI solutions.
By: Nestor Castillo, ForAllTechNews Director
Photo by: CC BY-SA 2.0 / Atribution-Compartirigual 2.0 Generico