Amazon is reportedly developing a ‘killer’ AI chatbot that will compete directly with OpenAI’s ChatGPT.
With some perceptions that the company is falling behind in the race to create generative AI, Business Insider has detailed the response to follow from the tech giant.
The chatbot has been codenamed Metis, powered by an Amazon LLM known as Olympus which is being trained on a massive two trillion parameters of data which is around twice the capability of GPT-4. This increased scope should make Metis more capable of human-like responses, reducing the margin for error in context and reasoning.
Metis will enable Microsoft to take its place among the other big tech players such as Google and Microsoft which offer their own bespoke AI assistants and in a crowded, competitive AI space, many start-ups like Amazon-backed Anthropic, and Perplexity are also providing an offering.
Jeff Bezos questions Amazon’s position
The new offering from Amazon is expected to “deliver text and image-based answers in a conversational manner” and to provide links to source responses, suggest follow-up points of interest, and generate images. With the added capacity from Olympus, Amazon is said to want the AI chatbot to be able to return information from beyond the original data source to deliver updated information, such as stock prices.
Metis has also been tipped to work as an “AI agent”, moving toward the evolution of the technology as agents will be able to perform additional tasks seamlessly and autonomously, using data, decision-making and multitasking, based on activity patterns and set algorithms.
Some of the criticism that Amazon was falling behind in the pecking order came from an unlikely source in Jeff Bezos, with the company’s founder and former CEO stating his concern in emails to senior staff asking why more AI firms were not using its cloud services.
There have been reports of a proposed launch for Metis in late September this year.
Image credit: Via Ideogram
