Aug 30, 2024

Amazon's Anthropic Alexa

How much of the voice assistant will be powered by the AI startup?

Well, this may answer a few (of the many) questions about Amazon's upcoming "Remarkable Alexa" revamp:

Amazon's revamped Alexa due for release in October ahead of the U.S. holiday season will be powered primarily by Anthropic's Claude artificial intelligence models, rather than its own AI, five people familiar with the matter told Reuters.

Amazon plans to charge $5 to $10 a month for its new "Remarkable" version of Alexa as it will use powerful generative AI to answer complex queries, while still offering the "Classic" voice assistant for free, Reuters reported in June.

But initial versions of the new Alexa using in-house software simply struggled for words, sometimes taking six or seven seconds to acknowledge a prompt and reply, one of the people said.

That's why Amazon turned to Claude, an AI chatbot developed by startup Anthropic, as it performed better than the online retail giant's own AI models, the people said.

For months now, there have been various reports about Amazon's struggles to remake Alexa in a post-ChatGPT world. While the Echo devices had first-mover advantages, these clearly ended up being an albatross around the neck of the company as they've tried to come up with answers, quite literally. Internal fighting seemed to be further fueling problems, so a move to rely more heavily on an outside source to help power Alexa would seemingly make sense.

What's unclear from the report is if this is similar to what Apple is doing for Siri in their partnership with OpenAI – that is, basically using ChatGPT as a backstop of sorts for anything the in-house "Apple Intelligence" can't handle. This Amazon/Anthropic tie-up makes it sound like a deeper partnership, where the startup would actually be powering many queries of Alexa on the backend perhaps without the user knowing what answer is coming from where. I had originally wondered if this was the path Apple was going to take with Siri since, as with Alexa, Siri went from a first-mover to a mess to obsolete seemingly overnight.

When Bensinger asked Amazon for comment on the report of Anthropic powering Alexa, their response was actually sort of interesting:

"Amazon uses many different technologies to power Alexa," a company spokeswoman said in a statement in response to detailed Reuters questions for this story.

"When it comes to machine learning models, we start with those built by Amazon, but we have used, and will continue to use, a variety of different models - including (Amazon AI model) Titan and future Amazon models, as well as those from partners - to build the best experience for customers," the spokeswoman said.

That's definitely not a "no". In fact, it at least implies a partial "yes". But again, it's not clear just how much of Alexa is going to be powered by Claude. That only reads like not all of Alexa will be.

One other wrinkle in this: numerous reports have indicated that a key element of 'Remarkable Alexa' is that it will be a paid service. That's also very different from what Apple is doing with the new version of Siri, at least for now. It's one thing to revamp a voice assistant, it's another to ask people to pay for it. And it's another still if you're relying on a third-party service to power the paid offering, even if just in part. This makes Amazon's ability to make money here even more murky if they also have to pay a partner (which Apple currently does not).

That leads to a thought I first brought up in June while thinking through how to fix Alexa:

Amazon pretty clearly should just buy Anthropic, but obviously cannot in the current regulatory environment. They already have a lot riding on that startup, but can they cut an even larger one, similar deal to the deal Microsoft has with OpenAI? Or can they try to run Microsoft's Inflection playbook and ~~buy~~ totally not buy the company but obtain it through other means?

My prediction record got a boost yesterday on the news that Apple is likely to invest in OpenAI, something I corrected guessed back in May, and kept doubling-down on. I've done the same here, just a few days ago:

Well, a week later, they made that "hackquisition", ~~buying~~ hiring much of Adept, another highly-funded and well-regarded AI startup that had lost its way in the maze of the hyped-up AI funding environment. Obviously, there hasn't been enough time for that team to have had any real meaningful impact here, but if the launch of "Remarkable Alexa" leads to a flurry of "Unremarkable Alexa" headlines, I suspect that Anthropic hackquisition chatter will pick up...

I probably should have guessed that Amazon might get ahead of such negative headlines by actually using Anthropic to power Alexa. Call it a test before an eventual acquisition. Though beyond the regulators, Google may have a thing or two to say about such a move, as they also have billions of dollars invested here.

It's almost like this is building up to an OpenAI vs. Anthropic battle, with Microsoft and Apple on Team OpenAI, while Amazon and Google are on Team Anthropic. Of course, it sounds like NVIDIA also may soon join Team OpenAI, which could tilt some scales and might push Anthropic down a path...

🗣️

Do you hear that?...