“We have a lot happening in the space,” Swami Sivasubramanian, Amazon’s vice president of database, analytics and machine learning, said at the March meeting, according to a recording obtained by The Washington Post. “We have a lot coming, and I’m very excited to share some of our plans in the future.”
When generative AI exploded onto the scene with the launch of ChatGPT in November, it was the formerly nonprofit research group OpenAI that won laurels. Tech giants Google and Microsoft (the latter of which invested billions in OpenAI) scrambled to catch up, launching chatbot products Bard and Bing, respectively, not long after.
But Amazon was nowhere in the mix, despite operating a massive cloud computing business, having the most employees and a market valuation of more than a trillion dollars.
Amazon Web Services has announced partnerships with AI companies like Stability AI and Hugging Face, which will allow other companies to use Amazon’s infrastructure to build artificially intelligent products. The company uses machine learning across many of its business divisions, including Alexa and search on Amazon.com. But its failure to launch a consumer-facing generative AI has left some to speculate that the company is lagging.
Recently, a select group of venture capitalists and AI company founders gathered at the Cerebral Valley AI conference, where Amazon appeared to be noticeably absent. Some attendees said it seemed like the tech giant had fallen behind its peers in the AI race.
At the recent all-hands meeting, executives seemed to push back on that idea.
Chatbots are just “one example of an application of these kind of models,” Sivasubramanian said. He said Amazon’s more than 100,000 machine-learning customers have expressed interest in using the company’s technology to improve personalization, search engine results and even to automate call centers.
“If you look at our track record on how we innovated in machine learning, we really paved the way for adoption of machine learning among mainstream enterprises and customers,” Sivasubramanian said. “We are excited to do the same in this space because this space is rapidly evolving, but we will do it in an Amazonian way. We will listen to customers and look at what areas they really need help and how to make them successful in this space.”
Amazon did not have immediate comment. Amazon founder and former CEO Jeff Bezos owns The Post.
Not every employee seemed convinced. One AWS staffer who commented on the all-hands video said Amazon is “not even close to the user being able to use any AWS service the way people can with ChatGPT.”
Another employee, seemingly unsatisfied by the company’s response, decided to ask a different source how Amazon can compete on AI: ChatGPT. The employee posted the chatbot’s response, which included a made-up Amazon technology, in the comments of the video, to the confusion of some of their colleagues.
Talk of generative AI technology — and how it will reshape art, work and life as we know it — has upended Silicon Valley as of late. But the early leaders of that conversation have been smaller start-ups, whose tools — like Midjourney’s image generator and OpenAI’s ChatGPT and DALL-E — have become wildly popular. Their success has raised questions about whether the tech behemoths have fallen behind on innovation. While companies like Microsoft, Google and Meta scramble to catch up, they are also contending with concerns about safety, as users flock to cutting-edge technologies that can behave unpredictably.
The surge has come at a time when economic concerns are forcing these large firms to cut back on spending, in some cases laying off the very people it hired to put AI safeguards in place.
Amazon chief executive Andy Jassy told the Financial Times in February that the company has a strategy for generative AI.
“Most large, deeply technical companies like ours have been working on these very large generative AI models themselves for a long time,” he said.
Amazon tends to be pragmatic: Typically, it doesn’t publicize experiments, preferring to launch products with a clear market strategy, according to a former Amazon employee who worked on AI and spoke on the condition of anonymity to describe sensitive internal matters.
Amazon Web Services already sells multiple tools that leverage artificial intelligence, like AWS Polly, which automatically turns written text into “natural-sounding human speech,” and Amazon Panorama, a computer vision tool that can interpret what a camera sees and automatically alert people to problems, like a broken product on a conveyor belt.
Amazon has also invested heavily in building Alexa, its voice assistant. But voice-activated chat technologies like Alexa and Siri are fundamentally very different from generative AI, which can learn and adapt rather than just answer questions. Amazon had hoped customers would eventually use Alexa for shopping and order products without picking up a phone or a computer. But while the speakers are popular, that behavior hasn’t caught on.
And as Amazon has scaled back and tried to find focus due to an uncertain economy, it has laid off thousands of employees, including hundreds who worked on Alexa.
But AWS CEO Adam Selipsky, who shared the stage with Sivasubramanian at the March all-hands, reiterated that when it comes to AI, there’s “a ton happening across all of Amazon.”
The company is working on “some really exciting projects and offerings,” he said.
“When they come out in due course, I think it’s going to be really cool.”
Gerrit De Vynck contributed to this report.
