How the AI competition was lost by Siri, Alexa, and Google Assistant

It took more than a decade for virtual assistants to become indispensable. However, their clunky design and incorrect calculations left room for chatbots to rise.

Apple executives presented the fifth-generation iPhone to a packed auditorium in San Francisco on a gloomy Tuesday. The audience quickly got excited about a new feature on the phone, which had the same design as the previous model: A virtual assistant called Siri.

Scott Forstall, the head of software at Apple at the time, asked questions to Siri by pressing a button on his iPhone. “I’ve been in the A.I. field for a long time, and this still blows me away,” Mr. Forstall said after Siri checked the time in Paris (“8:16 p.m.”), defined the term "mitosis," which means "cell division in which the nucleus divides into nuclei containing the same number of chromosomes," and provided a list of 14 highly rated Greek restaurants, five of which were located in Palo Alto, California.

Twelve years ago, that was. People haven't been blown away by Siri or other artificial intelligence-powered assistants like Amazon's Alexa or Google Assistant since then. The technology has largely stagnated, and talking assistants have become the subject of jokes, such as in a 2018 "Saturday Night Live" sketch with a smart speaker for seniors.

A new type of virtual assistant is currently receiving rave reviews from the tech industry: chatbots. These AI-powered bots, like OpenAI's ChatGPT and the brand-new ChatGPT Plus, can quickly improvise responses to questions typed into a chat box. ChatGPT has been used to handle complicated tasks like writing fiction, drafting business proposals, and coding software.

Also, ChatGPT, which utilizes A.I. to think about what word comes straightaway, is quickly getting to the next level. It was incapable of writing a proper haiku a few months ago; It can now do so with enthusiasm. On Tuesday, OpenAI uncovered its cutting edge A.I. motor, GPT-4, which powers ChatGPT.

The excitement surrounding chatbots shows how Siri, Alexa, and other voice assistants, which once sparked similar enthusiasm, have lost their lead in the artificial intelligence race.

The products have faced difficulties over the past ten years. According to John Burkey, a former Apple engineer who worked on Siri, it took weeks to update the assistant's clunky code with basic features. According to former employees, Amazon and Google invested in areas where the technology rarely paid off because they miscalculated how it would be used. They claimed that companies' enthusiasm for the technology diminished after those experiments failed.

In an interview with The Financial Times this month, Microsoft CEO Satya Nadella stated that voice assistants are "dumb as a rock" and that newer artificial intelligence would take the lead. OpenAI has been closely collaborated with by Microsoft, which has invested $13 billion in the startup and incorporated its technology into the Bing search engine and other products.

Siri received no response from Apple. Google said it was focused on giving an extraordinary remote helper to assist with peopling on their telephones and inside their homes and vehicles; Bard, a chatbot, is the subject of separate testing by the company. The assistants and chatbots are powered by what are known as large language models, which are systems trained to recognize and generate text based on enormous data sets scraped from the web. Amazon claimed that it saw a 30% increase in global customer engagement with Alexa in the previous year and that it was optimistic about its mission to build world-class A.I. After that, they can come up with words to complete a sentence.

On the other hand, Siri, Alexa, and Google Assistant are, in essence, command-and-control systems. These are able to comprehend a limited number of inquiries and requests, such as "What is the weather like in New York City?" or "Turn on the lights in the bedroom." The virtual assistant simply states that it is unable to assist a user who requests a function that is not included in its code.

According to Mr. Burkey, who was given the task of improving Siri in 2014, Siri also had a cumbersome design that made it take a long time to add new features. Nearly two dozen languages are represented by an enormous list of words in Siri's database, including the names of musical artists and places like restaurants.

He described it as "one big snowball" as a result. He added, "it goes in one big pile" if someone wanted to add a word to Siri's database.

According to Mr. Burkey, even seemingly straightforward updates like adding some new phrases to the data set would necessitate a complete database rebuild, which could take up to six weeks. It could take nearly a year to add more complex features like new search tools. He stated that as a result, Siri could not develop into a creative assistant like ChatGPT.

According to former managers at Amazon and Google, the companies struggled to generate meaningful revenue with the assistants, which relied on technology similar to Siri's. Apple, on the other hand, successfully marketed its iPhones by utilizing Siri.)

According to a former Amazon leader involved with Alexa, the company hoped that the smart speaker powered by Alexa, the Echo, would help it increase sales in its online store by allowing customers to talk to Alexa to place orders. He went on to say that while Alexa's ability to respond to weather prompts and set alarms was fun to play with, few people asked Alexa to order things.

According to the former executive, Amazon may have overinvested in the production of new hardware that sold at or below cost, such as microwaves and alarm clocks that no longer function with Alexa.

According to the person, the company also underinvested in creating an ecosystem for users to easily expand Alexa's capabilities, similar to what Apple did with its App Store, which stoked interest in the iPhone. Although Amazon provided a "skills" store to enable Alexa to control third-party accessories like light switches, it was challenging for users to locate and configure skills for the speakers, in contrast to the streamlined process of downloading mobile apps from app stores.

Carolina Milanesi, a consumer technology analyst for Creative Strategies and an Amazon consultant, stated, "We never had that App Store moment for the assistants."

As a result of the company's 18,000 layoffs at the end of last year, a number of high-ranking Alexa executives have left the company.

An Amazon spokesperson, Kinley Pearsall, stated that Alexa was much more than just a voice assistant and that "we’re as optimistic about that mission as ever."

A former manager who worked on Google Assistant suggested that Alexa's failures might have led Google in the wrong direction. In an effort to imitate Alexa's capabilities, Google engineers spent years developing voice-controlled tablet screens and smart speakers for home accessories like light switches and thermostats. Ads were later incorporated into those home goods by the company, but they did not become a significant source of revenue.

According to the former manager, Google realized over time that the majority of people only used the voice assistant for a small number of straightforward tasks like starting timers and playing music. In 2020, when Prabhakar Raghavan, a Google chief, took over Google Collaborator, his gathering pulled together the virtual friend as a marquee highlight for Android cell phones.

When Google's parent company laid off 12,000 workers in January, the team developing operating systems for home devices lost 16% of its engineers.

A considerable lot of the huge tech organizations are currently hustling to think of reactions to ChatGPT. According to two people briefed on the program, Apple held its annual A.I. summit last month at its headquarters, an internal event for employees to learn about its large language model and other AI tools. According to the individuals, numerous engineers, including Siri team members, have been testing language-generating concepts weekly.

Google also stated on Tuesday that it would soon release generative artificial intelligence (A.I.) tools to assist software developers, governments, and businesses in developing applications with embedded chatbots and incorporating the underlying technology into their systems.

Experts in artificial intelligence predicted that chatbot and voice assistant technologies would eventually merge. As a result, chatbots will be able to be controlled by speech, and consumers of Apple, Amazon, and Google products will be able to ask virtual assistants to assist them with more than just weather-related tasks.

Aravind Srinivas, the founder of Perplexity, an artificial intelligence startup that provides a chatbot-powered search engine, stated, "These products never worked in the past because we never had human-level dialogue capabilities." We do now.