Home  >  Article  >  Technology peripherals  >  To deal with the challenge of ChatGPT, Google plans to embed AI into all products and services

To deal with the challenge of ChatGPT, Google plans to embed AI into all products and services

王林
王林forward
2023-04-12 17:55:051154browse

To deal with the challenge of ChatGPT, Google plans to embed AI into all products and services

News on March 9th, Google has always been known for its long-term bets on various future technologies, and most of what is currently needed to explode AI chatbots Technologies have been studied in its laboratories. However, a startup called OpenAI became an early leader in so-called generative artificial intelligence (AIGC) when it launched ChatGPT last November. The sudden success of OpenAI has Google and its parent company Alphabet rushing to catch up in the field of AIGC, with its CEO Sundar Pichai even comparing the technology to the application of fire or electricity.

ChatGPT is regarded by many as the ultimate challenger to Google’s traditional search engine. Coupled with the close relationship between OpenAI and Microsoft, ChatGPT seems to pose a double threat to Google. Falling behind rivals in areas where it believes it may have a key advantage has stoked anxiety at Google headquarters, according to current and former Google employees and others close to the company. As one current employee said: "As with any AI-related initiative, Googlers have unusually high expectations and a great deal of insecurity, which is an unhealthy combination."

This This situation reminds Pichai of his days as a product manager, because he is used to being directly involved in the detailed development of products. According to one former employee, Pichai should not have been directly involved in such specific tasks. Google co-founders Larry Page and Sergey Brin are also more involved in the company than they have been in years, with Brin even submitting a submission for Google's ChatGPT-style chatbot Bard changed code.

Google management has issued a so-called "Code Red," which includes a directive that all of its most important products, those with more than 1 billion users, must AIGC support will be added in a few months. The company already announced in March that creators on its YouTube video platform would soon be able to use the technology to virtually dress up.

The latest efforts in AI remind some Google employees of the last time the company issued a similar internal order, in 2011 when it began efforts to promote the ill-fated social network Google. But unlike last time, Google was never seen as a leader in social networking, whereas its expertise in AI was recognized. The parallel is that employee bonuses were once closely tied to Google's success, and now the performance of at least some Google employees may be affected by their ability to integrate AIGC into their work. Code Red has spawned dozens of plans to integrate AIGC technology. "We're throwing spaghetti at the wall to see what sticks," said one Google employee. "But that's far from what the company needs to transform and stay competitive."

Ultimately, the effort around Google failed, and the social network struggled to attract users. Google eventually said in 2018 that it would shut down the product for consumers. One former Google executive viewed the failure as a cautionary tale. A person familiar with the matter said: "Larry's request was that every product must have a social element, and the results looked pretty bad."

A Google spokesperson responded to someone who compared "Code Red" to Google. defend. While Google's mission spans across all products, he said, current AI efforts primarily encourage Google employees to test the company's AI tools internally: a common practice in the tech industry nicknamed "internal testing." The spokesperson also said that most Google employees don't spend extra time on AI, but only those working on related projects are involved.

Google is not the only company that firmly believes that "AI is now everything." Silicon Valley has entered a full-scale hype cycle, with venture capitalists and entrepreneurs suddenly declaring themselves visionaries in the field of AI, far removed from recent topics such as blockchain, and the companies they back often see their stock prices soar after announcing the integration of AI capabilities. In recent weeks, Mark Zuckerberg, CEO of Facebook parent company Meta, has been focusing on AI rather than the Metaverse, according to two people familiar with the matter. He has previously stated that the Metaverse is the cornerstone of the company, even changing the company's name to do so.

The new mobilization order is good news for many Google employees, who are well aware of the company's history of conducting speculative research but faltering in commercializing it. Some team members already working on the AIGC project hope they will now be able to "release more products and have greater product impact rather than just doing some research work," according to a person familiar with the matter.

In the long run, considering that Google has done a lot of preliminary work, OpenAI’s sudden rise in the past few months does not seem to be able to shake Google’s position. As early as 2016, Pichai began calling Google an "AI-first" company. Google has used machine learning technology to power its advertising business for years, while also integrating AI into key consumer products like Gmail and Google Photos. In these products, Google uses the technology to help users write emails and organize photos, among other things.

In a recent analysis, research firm Zeta Alpha reviewed the 100 most cited AI research papers between 2020 and 2022 and found that Google dominates the field. “It looks like Google is like a sleeping giant that’s still lagging behind but is catching up,” said Amin Ahmad, a former Google AI researcher and co-author of Vectara Co-founder of , a startup that provides conversational search tools for enterprises. Ahmed added: "I think Google has done a very good job of applying this technology to many of their core products and is well ahead of other companies in the industry."

Google has also been working hard. Address the tension between its business priorities and dealing responsibly with emerging technologies. There is good evidence that automated tools tend to reflect biases present in training data sets, and many companies are concerned about the impact of testing tools on the public before they are ready. AIGC comes with a number of risks that make Google reluctant to rush its technology to market. In the world of search, for example, chatbots can provide simple answers, some even directly from the company that developed it, as ChatGPT seems to have become the voice of OpenAI. Fundamentally, this is a riskier activity than providing a list of links to other websites.

Operation Code Red appears to have disrupted the way Google calculates risk-reward, worrying many experts in the field. Emily Bender, a professor of computational linguistics at the University of Washington, said Google and other companies following the AIGC trend may not be able to steer their AI products away from bias. In response, a Google spokesperson said that Google’s efforts are subject to its AI rules. The company published a set of guidelines for responsible development of AI technology as early as 2018 and has always adhered to a cautious approach.

Yet other companies have shown they are willing to move forward quickly whether Google does so or not. One of the most important contributions to the field by Google researchers was a landmark paper titled "Attention is All You Need," in which the authors introduced converters that help AI models focus system based on the most important information in the data they are analyzing. Transformers are now a key component of large-scale language models, and the technology powers today’s chatbots, where the “T” in ChatGPT stands for transformer. Five years after the paper was published, all but one of the authors left Google, citing a desire to escape the constraints of a large, slow-moving company.

They are among dozens of AI researchers who have jumped to OpenAI, and others have helped found startups including Character.AI, Anthropic and Adept. Several startups founded by former Google employees, such as Neeva, Perplexity AI, Tonita, and Vectara, are currently looking to reimagine search using large language models. Sara Hooker, a Google AI researcher who currently works at AI startup Cohere, said that only a few key places have the knowledge and ability to build these models, making competition for these talents fierce.

It is not new for people or certain organizations to have made significant contributions to the development of a breakthrough technology, only to see others realize astonishing financial gains without them. Keval Desai, who formerly worked at Google and is now a managing director at venture capital firm Shakti, cited Xerox Parc as an example. Xerox Parc laid the foundation for the personal computer era that would later be followed by Apple and Microsoft and build multi-trillion dollar empires on its foundation. Desai said: "Google wanted to make sure that it was not the Xerox Parc of that era. Even though all the innovation was happening there, it was not well known." (Xiao Xiao)

The above is the detailed content of To deal with the challenge of ChatGPT, Google plans to embed AI into all products and services. For more information, please follow other related articles on the PHP Chinese website!

Statement:
This article is reproduced at:51cto.com. If there is any infringement, please contact admin@php.cn delete