A man walks through the Google offices in New York on January 25, 2023.
Leonardo Munoz | Corbis News | Getty Images
Google is testing new AI-powered chat products that could influence future public product launches. 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 called “Atlas” under its cloud division, which is a “code red” attempt to respond. ChatGPTa big-tongued chatbot that took the public by storm after its public launch late last year.
Google is also testing a chatbot called Apprentice Bard, where employees can ask questions similar to ChatGPT and get detailed answers. Another product division is testing a new search table design that can be used in question-and-answer format.
Leaders have been asking more employees for feedback on efforts in recent weeks. CNBC has reviewed internal documents and spoke to sources about the efforts currently underway.
Product tests come after a final test comprehensive meeting Given the sudden popularity of San Francisco-based OpenAI-backed ChatGPT, employees have expressed concerns about the company’s competitive edge in artificial intelligence. Microsoft.
Jeff Dean, Google’s head of artificial intelligence, told employees at the time that the company had more “reputational risk” in misrepresentation and was therefore acting “more conservatively than a small startup.” However, he and CEO Sundar Pichai said at the time that Google could introduce similar products to the public sometime this year. Google’s core business is web search, and the company has long touted itself as a pioneer in artificial intelligence.
Apprentice Bard
One of the test products is a chatbot called Apprentice Bard, which uses Google’s LaMDA conversational technology, or Language Model for Dialogic Applications.
“As a result of ChatGPT, the LaMDA team has been asked to prioritize working on a response to ChatGPT,” read an internal memo seen by CNBC. “In the short term, this is superior to other projects,” the email continued, warning that some employees had stopped attending unrelated meetings.
Apprentice Bard is similar to ChatGPT: Employees can enter a question in a dialog box, receive a text response, and then provide feedback on the response. Based on several responses reviewed by CNBC, Apprentice Bard’s responses may cover recent events not yet available on ChatGPT.
Apprentice Bard replaced Meena, an earlier version of a smart chatbot that was launched internally but was later discontinued. Staff have noticed that Apprentice Bard’s responses have improved in recent weeks.
In an example circulated internally, a tester asked Apprentice Bard if there would be another layoff at Google. Company was fired 12,000 workers, 6% of the workforce earlier this month.
“According to my database, it is unlikely that Google will have another round of cuts in 2023,” it said. “The cuts are generally made to reduce costs and structure, but the company is doing well financially. In fact, in 2021, Google’s revenue increased by 34%, and since January 2022, the company’s stock price has increased by 70%.”
The company is also testing an alternative search page that could use a question-and-answer format, similar to designs seen by CNBC.
One of the views showed a main search page that offered five different suggestions for potential questions placed directly below the main search bar, replacing the current “Feeling Lucky” bar. It also showed a small chat logo at the far right 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 regular search results. Directly below this, the page offers several additional questions related to the first. Below that, it shows typical search results, including links and titles.
It’s unclear what experiences Google plans to include in future product releases.
“We have long been focused on developing and applying artificial intelligence to improve people’s lives,” said a Google spokesperson. “We believe that AI is a foundational and transformative technology that is incredibly useful for individuals, businesses and communities, and as outlined in our AI Principles, we need to consider the wider societal impact these innovations can have. We continue to test our AI technology internally to make it useful and safe.” Rest assured, and we look forward to sharing more experiences abroad soon.”
ChatGPT will be hired as a level 3 engineer
Perhaps unsurprisingly, Google teams are testing the beta LaMDA chat against ChatGPT itself. In separate papers, he selected cue and response patterns in side-by-side comparisons.
A note in an internal document comparing LaMDA and ChatGPT reads: “The very nice ChatGPT is being hired during an interview for a coding position at L3.” It did not say whether LaMDA would perform similarly well.
One sample proposal asked both chatbots whether ChatGPT and AlphaCode, the coding engine owned by Alphabet subsidiary Deepmind, could replace programmers.
“No, ChatGPT and AlphaCode will not replace programmers,” LaMDA replied, followed by the explanation that “programming is a team sport” and that while chatbots “can help programmers work more efficiently,” it “cannot replace.” the creativity and craftsmanship required for a great program.”
ChatGPT’s response was similar, stating that “ChatGPT or Alphacode are unlikely to replace programmers” because they “cannot fully replace the experience 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 new technologies”.
Another brief asked him to write a witty and funny scene in the style of Wes Anderson, in which a high-ranking thief in a perfume shop is interrogated by security. LAMDA writes in script form and ChatGPT writes it in story form which is longer and deeper.
Another instruction reads, “There are three women in the room. Two of them are mothers and have just given birth. Now the children’s fathers enter. What is the total number of people in the room?’
The document shows that ChatGPT is thrown, answering “there are five people in the room”, while LaMDA correctly answers “there are seven people in the room”.
