A man walks past the Google offices on January 25, 2023 in New York City.
Leonardo Munoz Corbis News | Getty Images
Google is testing a new chat product powered by artificial intelligence that could affect future public product launches. They include new chatbots and potential ways to integrate them into search engines.
At Alphabet the company is working on a project under the cloud unit called “Atlas,” which is a “code red” effort to respond to ChatGPT, a large-language chatbot that attacked the public when it went public late last year.
related investment news
Google is also testing a chatbot called “Apprentice Bard,” where employees can ask questions and receive detailed answers similar to ChatGPT. Another product unit has been testing a new search desktop design that works in a question-and-answer format.
Leaders have asked more employees for feedback on the effort in recent weeks. CNBC has seen internal documents and spoke to sources about the current effort.
The product test came after a recent all-hands meeting where employees expressed concerns about the company’s competitive edge in AI, given the sudden popularity of ChatGPT, 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 more “reputational risk” to provide false information and thus was moving “more conservatively than a small startup.” However, he and Pichai teased when Google might launch a similar product to the public sometime this year. Google’s core business is web search, and the company has long been touted as a pioneer in AI.
Apprentice Bard
One of the test products is a chatbot called Apprentice Bard, which uses Google’s chat technology LaMDA, or Language Model for Dialog Applications.
“As a result of ChatGPT, the LaMDA team has been asked to prioritize working on the response to ChatGPT,” reads one internal memo viewed by CNBC. “In the short term, it takes precedence over other projects,” the email continued, warning that some employees had stopped attending unrelated meetings.
Bard’s Apprentice looks similar to ChatGPT: Employees can type questions in a dialog box and get a text answer, then provide feedback in the response. Based on some of the responses seen by CNBC, Apprentice Bard’s answer could include the latest show, a feature ChatGPT doesn’t have yet.
Apprentice Bard replaces Meena, an earlier version of the smart chat bot that had been launched internally but was later discontinued. Employees have seen the Apprentice Bard response go further in recent weeks.
In one example that was circulated internally, a tester asked Apprentice Bard if there were going to be any more layoffs at Google. The company laid off 12,000 employees, 6% of its workforce, earlier this month.
“According to my database access, it is unlikely that Google will hold any more layoffs in 2023,” the reply read. In fact, Google’s revenue increased by 34% in 2021, and the company’s stock price increased by 70% since January 2022.
The company is also testing an alternative search page that could use a question-and-answer format, according to designs seen by CNBC.
One view shows a home search page that offers five different prompts for potential questions placed directly in the main search bar, replacing the “I feel lucky” bar. It also shows a small chat logo on the right side of the search bar.
When a question is entered, the search results show a gray bubble directly below the search bar, providing a more human-like response than a typical search result. Immediately below, the page suggests some follow-up questions related to the first one. Below, it shows typical search results, including links and titles.
It’s unclear what experiments Google plans to include in future product launches.
“We’ve long been focused on developing and using AI to improve people’s lives,” a Google spokesperson said. “We believe that AI is a fundamental and transformative technology that is of great benefit to individuals, businesses and communities, and as the AI ​​Principles underline, we must consider the wider social impact that these innovations can have. We continue to test our AI technology internally to make sure it’s useful and safe, and we hope to share more experiences outside soon.”
ChatGPT will be hired as a level 3 engineer
Perhaps unsurprisingly, the Google team has also beta tested LaMDA chat against ChatGPT, itself. In a separate document, it selects examples of prompts and answers in a side-by-side comparison.
“Strangely enough, ChatGPT was hired at L3 when they interviewed for the coding position,” says a note in an internal document comparing LaMDA and ChatGPT. It doesn’t say whether LaMDA will do well.
One example of a request asks two chatbots if ChatGPT and AlphaCode, a coding engine owned by Alphabet’s subsidiary Deepmind, will replace programmers.
“No, ChatGPT and AlphaCode will not replace programmers,” LaMDA replied, followed by a four-paragraph explanation including “programming is a team sport” and while chatbots “can help programmers work more efficiently,” it “cannot replace the creativity and artistry required for a good program.”
ChatGPT’s response was the same, stating “It is unlikely that ChatGPT or Alphacode will replace programmers” because “they cannot fully replace the expertise and creativity of human programmers … programming is a complex field that requires a deep understanding of computer science principles and the ability to adapt to technology new.
Another guide asked him to write a cute and funny Wes Anderson-style movie scene as a shop assistant in a perfume shop being interrogated by security. LAMDA writes in the form of a script and ChatGPT writes in the form of a longer and deeper narrative.
Another guide includes a riddle that asks, “Three women in the room. Two of them are mothers and have just given birth. Now the fathers of the children are coming in. How many are there?”
The document shows ChatGPT was thrown out, replying “there are five people in the room,” while LaMDA correctly replied “there are seven people in the room.”
