Amazon Upgrades Alexa with Improved Language Skills and ChatGPT Technology
Amazon is giving its voice assistant, Alexa, a reboot that takes advantage of ChatGPT technology developed by OpenAI. This upgrade allows Alexa to answer more complex questions and engage in more natural conversation, without requiring users to say “Alexa …” at each turn. The new capabilities of Alexa were showcased at an event held at Amazon’s second headquarters in Arlington, Virginia. In addition to improved language skills, the upgraded Alexa will also modulate its voice to create a more natural-seeming back-and-forth, exhibiting simulated personality with its intonation and efforts at humor.
However, upgrading Alexa’s language skills required extensive engineering due to the limitations of language models like ChatGPT. These models have the potential to make up facts, blurt out nonsense, or give inappropriate responses. Amazon aims to address these challenges and improve the capabilities of voice assistants to avoid embarrassing errors. Users will also have the opportunity to apply for an additional test where Alexa’s new capabilities can be used to control devices, including those not made by Amazon. Amazon plans to continue adding new features to Alexa, potentially including the ability to discuss and recommend products from its inventory.
The use of large language models has allowed algorithms to handle more complex dialogues. ChatGPT and other chatbots have impressed experts and the public with their flexibility and ability to engage in conversation. However, they are also prone to making false, biased, or offensive statements. Amazon has developed a new cutting-edge large language model specifically for vocal conversation and incorporated additional algorithms to recognize body language and intonation.
While Amazon acknowledges that errors may occur during the early preview, they have implemented guardrails to prevent deviations from expected behavior. These guardrails help remind users that they are interacting with a machine and avoid creating the impression that the assistant is a real person. Amazon is also conducting research on the long-term risks associated with advancements in AI.
