Microsoft's Antithesis: Amazon, which missed the biggest opportunity of the AI era

In the midst of the AI wave sweeping the globe, tech giants are all racing to take the lead. However, while Microsoft has become the leader in AI with ChatGPT, Amazon, holding Alexa, has found itself in an embarrassing situation of missed opportunities.

According to media reports on Friday, more than a dozen former Amazon Alexa team members revealed that this tech giant is seriously lagging behind competitors Google, Microsoft, and Meta in the AI race, and is struggling to catch up.

Amazon launched a new Alexa voice assistant driven by generative AI in September last year, showcasing its natural conversation and task processing capabilities at a grand event.

At that time, Amazon stated that the Alexa Large Model would soon be available for preview on Alexa devices in the U.S. market for free.

Rohit Prasad, Amazon's Senior Vice President and head of Alexa, stated that this news marked "a significant change for our beloved assistant" and referred to the new Alexa as a "super assistant."

However, after the event, it was as if nothing had happened.

Alexa remained the same familiar Alexa, and in the following months, there was hardly any further progress heard about this new version of Alexa.

Why was the new Alexa project aborted?

According to interviews with more than a dozen former employees who worked in the Alexa AI department, the launch of the new Alexa version was repeatedly delayed, due to technical challenges and internal organizational flaws at Amazon.

Although the core large language model of the new Alexa is benchmarked against GPT , the parameter scale is only 100 billion, far below the industry-leading level. Additionally, due to a lack of sufficient high-quality data and advanced chip equipment, the Alexa team struggled to catch up in a short period of time.

"The deadlines given by management are always unrealistic, they may not even understand what it takes to build a product like ChatGPT." a former Alexa researcher said.

Bloated organizational structure and conflicts between departments also hindered the process of updating Alexa.

The report stated that in order to achieve natural conversation, many of Alexa's original functional modules needed to be reconnected and deeply collaborated with the core language model. However, the leadership was reluctant to maintain both the old and new systems, preferring to "burn down the old version of Alexa first."

Various territorial disputes and short-sighted behaviors have led many employees to feel frustrated and choose to resign.

To make matters worse, Amazon has repeatedly lowered the priority of the Alexa project and placed it behind the development of generative AI technology for AWS.

Even the AI startup Anthropic, in which Amazon invested $4 billion last year, has a Claude large model that can be compared to GPT, but due to privacy concerns and internal politics, it has not effectively empowered the Alexa team.

It is worth mentioning that the Alexa business unit has been in a loss-making state for many years, mainly seen as a means to maintain user stickiness and acquire data. In contrast, the AWS cloud computing business is the company's profit backbone "They are questioning whether (the new Alexa) will eventually succeed," an industry insider who has recruited talent for the Alexa team multiple times said, many former employees have lost confidence in this project.

Although many believe that Amazon is fully capable of bringing Alexa into the era of generative AI, more and more evidence suggests that Amazon may not be ready yet.

"Alexa AI is plagued with technical and bureaucratic issues," said Mihail Eric, former senior machine learning scientist at Alexa AI, a long time ago, which has led to many advanced technologies not being able to be implemented.

Several dozen former employees interviewed by the media in the past month have echoed Eric's remarks.

Large models outperform old-school machine learning, Amazon rushes to respond

In fact, long before ChatGPT shocked the world, Amazon launched Alexa in 2014 as a digital assistant with the Echo smart speaker.

After Alexa was launched, it quickly gained popularity, and by 2017, over 20 million devices had been sold.

Alexa does not use a generative large language model like ChatGPT, but instead follows a traditional natural language processing technology route, with its core being pieced together by many small machine learning models and a large number of manually written rules.

Although this architecture was successful in the early stages of smart speakers, it revealed a lack of flexibility and adaptability after the emergence of ChatGPT.

To keep up with the pace of the new generation of AI assistants, Alexa may need a thorough overhaul at the algorithm, model, and data levels.

Due to the lag in this core technology, once ChatGPT was launched, Amazon was caught off guard.

"Internal leadership clearly has grand plans, but they don't really know what they are doing," a former intern at the Alexa team said.

In the following months, Amazon's Alexa team has been in chaos, frantically transforming Alexa from a rigid command-based robot into a truly conversational and helpful agent.

At the same time, overnight, Amazon's non-generative AI projects were all downgraded, and executives urged developers and product managers to find ways to ensure that Amazon could provide generative AI products to customers.

A former project manager for Alexa artificial intelligence described the atmosphere at the company as "a bit panicked"