According to Fortune, Alexa is nowhere near its goal to be “the world’s best personal assistant,” much less Amazon founder Jeff Bezos's ambition to create a real-life equivalent of the helpful Star Trek computer.
Instead, Amazon’s Alexa risks becoming an obsolete digital artefact, a cautionary tale of a potentially revolutionary technology that became entangled in the wrong game.
Amazon wanted to incorporate AI into an upgraded Alexa but found it is encountering issues. The company's more conversational, context-aware voice assistant, which it showcased at its autumn hardware event last year, has yet to be released beyond a limited preview. According to Fortune’s coverage, while Amazon might eventually introduce an improved LLM-based Alexa, it will fall short of its potential.
Many former employees interviewed by Fortune stated they left partly because they believed the new Alexa would never be ready or would have been surpassed by competitors when it was launched. Its most significant disadvantage, compared to firms like OpenAI and its widely discussed ChatGPT, is that it must “navigate an existing tech stack and defend an existing feature set,” as Fortune states.
Essentially, the old Alexa is obstructing the new Alexa. Fortune’s informants claim Amazon has yet to determine how to merge Alexa’s current capabilities with those promised for the latest Alexa last autumn—a better, smarter, more conversational assistant. One employee told Fortune that the message within the company following the demonstration event was that “we need to essentially destroy the bridge with the old Alexa AI model and shift to exclusively working on the new one.”
Fortune reports that Amazon has struggled to enable its Alexa LLM to consistently and effectively execute API calls, which is how the current Alexa interacts with other devices, such as third-party smart home gadgets and music services. It has also faced challenges in training the LLM to comprehend natural language, as while it has millions of devices in use, its customers have conditioned themselves to use “Alexa language” and do not engage conversationally with the device.
Another challenge reported has been Amazon’s decentralised organisational structure, where the thousands of individuals working on Alexa are divided into numerous teams, leading to conflict and dissatisfaction.
Mihail Eric, a research scientist who departed the company in 2021, stated on X (formerly Twitter) that he attributes the failure of his work on Alexa—which he claims could have been the start of an Amazon ChatGPT (well before ChatGPT was released)—to the company’s organisational chart and the insistence that research be linked to a product launch.
Amazon claims it is committed to the development of its voice assistant.
“Generative AI presents a tremendous opportunity to enhance Alexa for our customers. We have already incorporated generative AI into various components of Alexa. We are diligently working on large-scale implementation—in the over half a billion ambient, Alexa-enabled devices in homes globally—to provide even more proactive, personal, and reliable assistance to our customers. We are enthusiastic about what we are creating and look forward to its delivery to our customers,” a spokesAmazon said.