Users naturally want their voice assistants to be smarter, and since the appearance of ChatGPT, there have been continuous questions online about “how to integrate ChatGPT into my phone”. Moreover, voice assistant applications based on GPT have already been launched.
However, this is not enough for giants to make strategic decisions. Although Alexa helped Amazon to seize a share of the smart speaker market, it does not have a clear path to commercialization, and even engages in loss-making businesses. Amazon’s ambition is to use voice assistants to lead users to e-commerce consumption, but over time, most users only use Alexa to execute basic commands such as “listen to music” and “turn off the lights”.
At the end of last year, Amazon announced layoffs of 17,000 people, of which the “Global Digital” department responsible for Alexa became a heavily affected area, and some executives left the company. According to BusinessInsider, this department lost $3 billion in the first quarter of 2022 alone, twice the total losses of other departments, and the culprit is Alexa.
The report also stated that the hardware team is expected to lose $10 billion in 2022.
Although Google Assistant has a high share in the voice assistant market and is the number one voice assistant in the US market, it seems to have been misled by Amazon and has taken the same wrong path. In the five years from 2016 to 2021, Google released eight major hardware products for embedded main search speakers/smart displays, and has since stagnated.
In 2022, Google removed support for its assistant for two internal product lines, Nest Wifi routers and Fitbit wearables. The same year, The Information reported that Google wants to reduce investment in developing voice-assisted search functions for Google Assistant.
Google Assistant, like Alexa, has no profitable path, no advertising, no subscription model, and is just a diligent assistant that often cannot help.
Using large language models behind chatbots such as ChatGPT and Bard on voice assistants can naturally improve the user experience, but it still cannot solve the problem of product commercialization.
As for Apple, it is deeply constrained by “protecting user privacy.” A few years ago, Siri was involved in a scandal of “eavesdropping” and withdrew all outsourcing companies for Siri projects, showing greater respect for the privacy of conversations between users and Siri. But for large language models, this means a lack of training data.
Whether it is Siri, Alexa, or Google Assistant, they have developed for many years and need to solve internal resource allocation problems to make major changes.
Siri from Apple has experienced ups and downs from the beginning. Steve Jobs personally guided the project but died on the second day of the product launch. After a year of release, Siri had already gone through three responsible persons, with constant internal disagreements, not only in management, but also in search, speech recognition, and natural semantics.
In 2022, three key figures, Srinivasan Venkatachary, Steven Baker, and Anand Shukla, left, they were the founders of machine learning startup Laselike, later acquired by Apple for $150 million. This departure led them into the arms of Google.
To make Siri take a major turn, it is bound to involve the re-allocation of internal personnel and resources: who will lead the project? What path should be taken? Should the past personnel stay or leave? These are all difficult problems for Cook.
By comparison, Google has already made a choice. At the end of March, CNBC reported on changes within Google’s team. According to a leaked internal memo, Google Assistant seems to be making way for the company’s new superstar Bard, and is ready to integrate Bard with Google Assistant.