Google plans to launch an AI chatbot against ChatGPT, tentatively named "Apprentice Bard"

Zhitong Financial App learned that Google (GOOGL. US) is testing new AI chat products that could impact future public product launches. These include a new type of AI chatbot, and a way to potentially integrate it into search engines. The company is working on a project under its cloud computing unit called "Atlas," a response to ChatGPT's "Code Red." ChatGPT is a large-scale language-based artificial intelligence chatbot that attracted a lot of public attention after entering the market late last year.

Google is also testing an AI chatbot tentatively called Apprentice Bard, where employees can ask questions and get detailed answers similar to ChatGPT. Another of Google's product divisions has been testing a new search desktop design designed for question-and-answer searches.

In recent weeks, Google executives have been asking more employees for feedback on the measures. The product test was conducted at a recent plenary session, thanks to a report by Microsoft (MSFT. ChatGPT, launched by OpenAI, a San Francisco startup backed by US), has suddenly gained popularity around the world, and Google has expressed concern about the company's competitive advantage in artificial intelligence.

Jeff Dean, Google's head of artificial intelligence, told employees at the time that the company had a greater "reputational risk" in providing misinformation and was therefore "more conservative than smaller startups." However, he and CEO Sundar Pichai joked at the time that Google might launch a similar product to the public sometime this year. Google's main business is web search, and the company has long touted itself as a pioneer in the field of artificial intelligence.

"Apprentice Bard" VS ChatGPT

One of Google's test products is a chatbot called Apprentice Bard, which uses Google's conversational technology LaMDA, the language model for conversational applications.

An internal memo reads: "Because of ChatGPT, the LaMDA team was asked to prioritize responses to ChatGPT." "In the short term, it takes precedence over other items," the email continued, warning some employees to stop attending certain unrelated meetings.

Apprentice Bard looks similar to ChatGPT: employees can enter a question in a dialog box, get a text answer, and then feedback on the answer. According to people familiar with the matter, Apprentice Bard's answer may include recent events, a feature that ChatGPT does not yet have.

Apprentice Bard replaced Meena, a previous version of an intelligent chatbot that was launched inside Google but was later discontinued. Google employees have also noticed that the apprentice Bard has become increasingly responsive in recent weeks.

In one example circulating internally, a tester asked apprentice Bud if Google would have another round of layoffs. Earlier this month, the company laid off 12,000 employees, or 6 percent of its workforce.

Apprentice Bard wrote in his response: "Based on my database access, Google is unlikely to make another round of layoffs in 2023." "Layoffs are usually about reducing costs and structure, but the company is in good financial shape. In fact, Google's revenue grew 34% in 2021, and the company's share price has risen 70% since January 2022. ”

According to the design seen by sources, the tech company is also testing an alternative search page that could use a question-and-answer mode. It's unclear what experiments Google plans to include in future product launches.

A Google spokesperson said: "We have long focused on developing and deploying artificial intelligence to improve people's lives." "We see AI as foundational and transformative technology that will be very useful to individuals, businesses and communities, and as outlined in our AI principles, we need to consider the broader societal impact that these innovations may have." We will continue to test our AI technology internally to ensure it is helpful and safe, and we look forward to sharing more experiences with the outside soon. ”

In the face of the soul Q&A thrown by humans, how do AI chatbots react

Perhaps unsurprisingly, the Google team is also testing a chat tool based on beta LaMDA technology against ChatGPT itself. In different documents, it selects examples of tips and answers that are compared side by side.

In an internal document comparing LaMDA and ChatGPT, it reads: "Surprisingly, when interviewing for a coding position, ChatGPT was hired at L3." It did not say whether LaMDA would perform just as well.

One example prompt asks whether two chatbots, ChatGPT and AlphaCode (the coding engine owned by Alphabet subsidiary Deepmind), will replace programmers.

LaMDA's response was, "No, ChatGPT and AlphaCode will not replace programmers." LaMDA then uses four paragraphs of explanation, including that "programming is a team sport" and that while chatbots "can help programmers work more efficiently," they "cannot replace the creativity and artistry necessary for great programs."

ChatGPT responded similarly, saying that "ChatGPT or Alphaccode are unlikely to replace programmers" because they "cannot completely 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 new technologies."

Another tip asked it to write a witty and funny movie scene in Wes Anderson's style, depicting an upscale store pickpocket being interrogated by security guards in a perfume store. LAMDA takes the form of a script, while ChatGPT takes the form of a longer, deeper narrative.

Another tip includes a classic riddle: "Three women in one room." Two of them are mothers who have just given birth. Now, the fathers of the children are coming in. How many people are in the room?"

The documentation shows that ChatGPT was constrained by the question, answering "There are five people in the room," while LaMDA correctly answered "There are seven people in the room."