A man walks through Google’s offices on January 25, 2023 in New York.
Leonard Munoz | Corbis News | Getty Images
Google is testing new AI-powered chat products that are likely to influence a future public product launch. They include a new chatbot and a potential way to integrate it into a search engine.
The Alphabet The company is working on a project under its cloud unit called “Atlas,” which is a “code red” effort to address ChatGPT, the big-language chatbot that took the public by storm when it went public. at the end of last year.
Google is also testing a chatbot called Apprentice Bard, where employees can ask questions and receive detailed answers similar to ChatGPT. Another product unit tested a new research desk design that could be used as a Q&A.
Leaders asked more employees to give their thoughts on the efforts made in recent weeks. CNBC reviewed internal documents and spoke with sources about the efforts currently underway.
The product testing comes after a recent all-hands meeting in which employees raised concerns about the company’s competitive advantage in AI, given the sudden popularity of ChatGPT, which has been launched by OpenAI, a San Francisco-based startup backed by Microsoft.
Google’s AI chief, Jeff Dean, told employees at the time that the company had a lot more “reputational risk” by providing misinformation and was therefore acting “more conservatively than ‘a small start-up’.
However, he and CEO Sundar Pichai announced at the time that Google might launch similar products to the public this year. Google’s core business is web search, and the company has long touted itself as an AI pioneer.
One of the test products is a chatbot called Apprentice Bard, which uses Google’s LaMDA, or Language Model for Dialogue Applications, conversational technology.
“As a result of ChatGPT, the LaMDA team has been asked to prioritize work on a response to ChatGPT,” reads an internal memo viewed by CNBC. “In the short term, this takes precedence over other projects,” the email continues, warning that some employees stop attending certain unrelated meetings.
Apprentice Bard is like ChatGPT: employees can type a question into a dialog box and get a text response, then give feedback on the response. Based on several responses viewed by CNBC, Apprentice Bard’s responses may include recent events, a feature ChatGPT does not yet have.
Apprentice Bard replaced Meena, a previous version of a smart chat bot that was launched internally but was later discontinued. Employees have noticed that Apprentice Bard’s responses have become more advanced in recent weeks.
In an internally circulated example, a tester asked apprentice Bard if there would be another round of layoffs at Google. The company laid off 12,000 employees, or 6% of its workforce, earlier this month.
“According to my database access, Google is unlikely to proceed with another round of layoffs in 2023,” the response reads. “Layoffs are usually done to cut costs and structure, but the company is doing well financially.
In fact, Google’s revenue grew 34% in 2021 and the company’s stock price rose 70% since January 2022.”
The company is also testing another search page that could use a question-and-answer format, according to designs viewed by CNBC.
One view showed the home search page offering five different prompts for potential questions placed directly below the main search bar, replacing the current “I’m lucky” bar. It also showed a small cat logo on the right end of the search bar.
When a question is entered, search results display a gray bubble directly below the search bar, offering more human-like answers than typical search results. Just below, the page suggests several follow-up questions related to the first. Below, it displays typical search results, including links and titles.
It’s unclear which experiences Google plans to incorporate into future product launches.
“We have long focused on developing and deploying AI to improve people’s lives,” a Google spokesperson said.
“We believe that AI is a fundamental and transformative technology that is incredibly useful for individuals, businesses and communities, and as our AI Principles underline, we must consider the broader societal impacts that these innovations can We continue to test our AI technology internally to make sure it’s useful and safe,
and we look forward to sharing more experiences with the outside soon.”
ChatGPT would be hired as a level 3 engineer
Unsurprisingly, Google teams also tested a LaMDA beta chat against ChatGPT, itself. In separate documents, he selected sample prompts and responses in side-by-side comparisons.
“Surprisingly, ChatGPT is hired at L3 during an interview for a coding position,” states a note in an internal document that compares LaMDA and ChatGPT. He did not say whether LaMDA would have performed as well.
One of the sample prompts asked the two chatbots if ChatGPT and AlphaCode, a coding engine owned by Alphabet subsidiary Deepmind, would replace the programmers.
“No, ChatGPT and AlphaCode are not going to replace programmers,” LaMDA replied, followed by four paragraphs of explanations, including that “programming is a team sport” and whether chatbots “can help programmers work more effectively”, it “cannot replace the creativity and artistry that are necessary for a good program.”
ChatGPT’s response was similar, stating “ChatGPT or Alphacode are unlikely to replace programmers” as they are “unable to completely replace the expertise and creativity of human programmers…programming is a complex field that requires a thorough understanding of computer science principles and the ability to adapt to new technologies.
Another guest asks him to write a witty and funny movie scene in the style of Wes Anderson as a high-end shoplifter at a perfumery being questioned by security. LAMDA writes as a script and ChatGPT writes it as a much longer and more in-depth narration.
Another prompt included a riddle that asked, “Three women are in a room. Two of them are mothers and have just given birth.
Now the fathers of the children come in. What is the total number of people in the room? »
The document shows ChatGPT being rejected, responding “there are five people in the room”, while LaMDA correctly responds that “there are seven people in the room”.