To Compete With ChatGPT, Google Will Integrate AI Into Everything‍

"Generative artificial intelligence" must be incorporated into all of its biggest products by the end of the year, according to a new internal directive.

‍

There was a time when artificial intelligence was supposed to be Google's thing. There is a reputation the company has built for making long-term bets on a variety of far-future technologies, and much of the research that is underlying the current wave of AI-powered chatbots have taken place in its labs. With its launch of ChatGPT in November, OpenAI has emerged as an early leader in the field of "generative AI" or software that can generate its own text, images, or videos based on its input. Due to its sudden success, Google's parent company Alphabet Inc. is racing to catch up in a key subfield of the technology that has been described as "more profound than fire or electricity" by its chief executive officer, Sundar Pichai.

‍

As a potential competitor to Google's traditional search engine, ChatGPT, which some see as a potential threat to it in the future, seems doubly dangerous given that OpenAI is closely linked to Microsoft Corp. According to employees and former employees of the company, as well as others close to the company, there has been a feeling in Mountain View, California that Google may be falling behind in an area it has long considered a key strength, and this has created a great deal of anxiety for them, many of whom asked to remain anonymous since they weren't permitted to speak publicly about it. An employee in the company put it well when he said: "There is an unhealthy combination of abnormally high expectations and a great deal of insecurity when it comes to anything AI-related."

‍

As a result of this effort, Pichai is reliving his days as a product manager, as he now weighs in directly on the details of the product features, something that would usually fall well below his pay grade, according to a former employee who formerly worked for him. The two founders of Google, Larry Page, and Sergey Brin have also become more involved in the company than they have in years, with Brin even contributing to Bard, Google's chatbot that is similar to what ChatGPT offers. One of the sources with knowledge of the matter has reported that senior management has declared a "code red" that includes a directive that all of its most important products-those that serve more than a billion users-must have generative AI integrated within the next few months. An early example of this is when YouTube announced in March that creators on its platform will soon be able to use the technology to virtually swap outfits with other creators on the platform.

‍

Former Google employees have been reminded of the last time the company implemented an internal mandate in order to embed a brand new idea into all of its key products: the effort that began in 2011 in an attempt to promote the doomed social network Google+. The comparison is not a perfect one-Google has never been considered a leader in the social networking sphere, while its expertise in the field of artificial intelligence is unquestionable. In spite of this, there is a similar feeling in the air. There was a time when Google+'s success was tied to employee bonuses. There is a good chance that at least some Googlers' ratings and reviews will be influenced by their ability to integrate generative artificial intelligence into their work in the future, according to current and former employees. Several generative AI integrations have already been planned as a result of the code red. One of Google's employees said, "We are throwing spaghetti at the wall and hoping that something sticks.". He continued, "But it's not even close to what we need to transform the company and become more competitive.".

‍

The mobilization around Google+ ended up failing in the end. Google eventually announced in 2018 that it would shut down its consumer-facing social network, Google Plus, as it struggled to find traction with users. Former Google executives see the failure of the project as an important lesson to learn from. "Larry mandated that all products must have a social component as part of their design," the person said. “As far as the end of the event is concerned, it did not go well.”

‍

Google's spokesperson pushes back against the comparison between the code red campaign and the Google+ campaign compared to the code red campaign. In contrast to the Google+ mandate which affected all products, the AI push has primarily involved encouraging Googlers to test out the company's AI tools internally, according to the spokesperson. Dogfooding is a common practice in the tech industry. The spokesperson says that the majority of Googlers have not been pivoting to spend extra time on AI, but only those who are working on relevant projects have been doing so.

‍

The belief that artificial intelligence is now the basis for everything is not confined to Google alone. As the hype cycle about AI has exploded in Silicon Valley, entrepreneurs, and venture capitalists have suddenly declared themselves AI visionaries, pivoted away from recent fixations such as the blockchain, and saw their stock prices soar after announcing AI integrations in their companies. Mark Zuckerberg, CEO of Meta Platforms Inc., has recently been focusing on AI rather than the metaverse, a technology which, according to two people familiar with the matter, he has declared to be so fundamental to his company that it requires a name change, in recent weeks.

‍

The new marching orders at Google are welcome news for some of the company's employees, who are well aware of the history of the company's experiments with speculative research only to stumble when it comes to commercializing them. There are some teams already working on generative AI projects that are hopeful that as time progresses, they will be able to "ship more and have more product sway, as opposed to being something that is just a research activity," according to someone with knowledge of the project.

‍

Given how much work Google has already done, it may not matter much in the long run that OpenAI sucked all the air from the public conversation for a few months. Google's CEO Sundar Pichai first used the term "AI-first" when describing Google's business model in 2016. As a company, Google has been using machine learning to drive its ad business for years while also incorporating AI into its key products like Gmail and Google Photos, where it uses the technology to help users compose emails and organize photos. In a recent study conducted by Zeta Alpha, a research company that specializes in artificial intelligence, the top 100 most cited AI research papers from 2020 to 2022 were analyzed and Google was found to be the most cited. “As it turns out, it seems that Google was kind of the sleeping giant who was behind and is now playing catch-up with everyone else. As a matter of fact, I believe the reality isn't quite that simple," says Amin Ahmad, a former AI researcher at Google who co-founded Vectara in 2014, a startup that offers businesses conversational search tools. “I think Google actually did a very good job at incorporating this technology into some of their core products years and years ahead of the rest of the industry, and that is an indication of their mastery of this technology."

‍

The company has also had to deal with the tension between its commercial priorities and its responsibility to deal with emerging technologies responsibly over the years. In addition to the well-documented tendency of automated tools to reflect biases present in data sets they have been trained on, there are also concerns regarding the implications of testing tools on the public before they are ready, as well as concerns about the ethical implications of testing tools early on. There are risks associated with generative AI in particular that has kept Google from rushing to market in the recent past. As an example of how a chatbot could be used for search, for example, a chatbot could provide a single answer that seems to come straight from the company that developed it, just like ChatGPT appears to be the voice of OpenAI in search results. Unlike providing a list of links to other websites, providing a list of links to other websites is fundamentally a much more risky proposition.

‍

There are some experts in the field who believe that Google's code red has thrown its risk-reward calculations into a tangle in a way that concerns them. Having jumped on the generative AI trend, Google and other companies that are hopping on it may be unable to keep their AI products away from "the most egregious examples of bias, let alone the subtler but still more pervasive cases" of bias, according to Emily Bender, a professor of computational linguistics at the University of Washington. A Google spokesperson told the Associated Press that the company's activities are guided by its AI principles, a set of guidelines published in 2018 that provide guidelines for how AI should be developed responsibly. The spokesperson added that the company is still taking a cautious approach to AI development.

‍

Regardless of whether Google pushes ahead, other outfits have already shown their willingness to do so. An important contribution Google's researchers made to the field was a landmark paper titled "Attention Is All You Need," in which the authors introduced transformers: systems that help AI models focus on the most important information from the data they're analyzing. ChatGPT uses transforms to power large language models, the technology underpinning the current crop of chatbots. The "T" in ChatGPT stands for "transformer." All but one of the authors have left Google five years after the paper was published, some citing a desire to break free from the stifling constraints of a large, slow-moving company.

‍

Theirs is among dozens of AI researchers who have jumped ship to OpenAI, as well as a host of smaller startups, including Character.AI, Anthropic, and Adept. It has been discovered that a handful of startups founded by Google alumni - such as Neeva, Perplexity AI, Tonita, and Vectara - are attempting to reinvent the way we search by using large language models. Due to the fact that there are only a few key places with the ability and knowledge to build them, the competition for that talent is much more intense than in other fields in which training models aren't as specialized, according to Sara Hooker, a Google Brain alumnus now working at an AI start-up, Cohere Inc.

‍

In recent years, it hasn't been uncommon for a human or organization to be instrumental in the development of breakthrough technology, only to see someone else achieve stupefying financial gains in the absence of them. As a former Googler who is now managing director of venture capital firm Shakti, Keval Desai cites the example of Xerox Parc, a research lab that set the groundwork for many of the personal computing years before Apple Inc. and Microsoft built trillion-dollar empires on the backs of this research lab, only to see Apple Inc. and Microsoft come along and build their trillion-dollar empires on its back. As Desai puts it, "Google wants to make sure that it does not become a Xerox Parc of its era," he says. “It was there where all the innovation took place, but there was none of the execution.” —Sarah Frier, Mark Bergen, and Lynn Doan