A smarter version of Amazon's Alexa is coming — but it might not be powered by Amazon's tech.
According to a new report by Reuters, the new version of Alexa will "primarily" be powered by Anthropic's Claude AI models, instead of Amazon's own AI.
Anthropic is an AI company founded by ex-OpenAI employees, which claimed that Claude 3's Opus model, which launched in March this year, was in some ways more powerful than comparable models from OpenAI and Google. Amazon is a big investor in Anthropic, having invested $4 billion into the company (Google is another investor, having thrown at least $2 billion into Anthropic itself).
The new, AI-powered Alexa is internally referred to as "Remarkable," and it will cost between $5 and $10 when it launches in October, Reuters reported in June. Classic Alexa – the one that can perform basic tasks but cannot chat with you nearly as well as OpenAI's ChatGPT can – will remain free.
When asked about this, Amazon gave Reuters a canned response, saying it "uses many different technologies to power Alexa," though it did say it will continue to use its own Titan model, future Amazon models, "as well as those from partners."
The report sounds quite a bit more damning than that, with Reuters' sources claiming that Amazon's attempts to build the new Alexa with in-house software had it "struggling for words," and that it was painfully slow, sometimes taking "six or seven seconds to acknowledge a prompt and reply." Ouch.
Alexa is a very important part of Amazon's ecosystem, given that it powers the vast number of smart devices in Amazon's lineup. And yeah, with Google making its things smarter via Gemini and Appel preparing to supercharge Siri, Amazon's Alexa could probably use some AI-based sprucing up.
Topics Artificial Intelligence
