Google is not about let Microsoft or anyone else take the search crown without a fight. The company announced today that it will release a chatbot called Bard “in the coming weeks.” The launch appears to be a response to ChatGPT, the sensationally popular AI chatbot developed by Microsoft-funded startup OpenAI.
Sundar Pichai, Google’s CEO, wrote in a blog post that Bard is already available to “trusted testers” and is designed to put the “breadth of the world’s knowledge” behind a conversational interface. It uses a scaled-down version of a powerful AI model called LaMDA that Google first announced in May 2021 and is based on technology similar to ChatGPT. Google says this will allow the chatbot to be offered to more users and collect feedback to help address issues related to the quality and accuracy of the chatbot’s responses.
Google and OpenAI build their bots on text-generating software that, while eloquent, is prone to fabrication and can repeat objectionable speech styles picked up online. The need to mitigate these shortcomings, and the fact that this type of software cannot be easily updated with new information, poses a challenge to hopes of creating powerful and profitable new products based on this technology, including the assumption that chatbots can restore web- search.
Notably, Pichai has not announced any plans to integrate Bard into the search field that drives Google’s revenue. Instead, it demonstrated a novel and cautious use of core artificial intelligence technology to improve conventional search. For questions that don’t have a single, agreed-upon answer, Google synthesizes an answer that reflects different opinions.
For example, the question “Which is easier to learn to play the piano or the guitar?” will be met with: “Some say the piano is easier to play because the finger and hand movements are more natural… Others say the guitar is easier to learn chords.” Pichai also said that Google plans to make the underlying technology available to developers through an API, as OpenAI does with ChatGPT, but did not offer a timetable.
The heady hype generated by ChatGPT has led to speculation that Google is facing a serious challenge to its web search dominance for the first time in years. Microsoft, which recently invested about $10 billion in OpenAI, is holding a media event tomorrow related to its work with the creator of ChatGPT, which is believed to be about new features for the company’s second-largest search engine, Bing. OpenAI CEO Sam Altman tweeted the photo about himself with Microsoft CEO Satya Nadella shortly after Google’s announcement.
Quietly launched by OpenAI last November, ChatGPT has become an internet sensation. Its ability to answer complex questions with apparent consistency and clarity has many users dreaming of revolutionizing education, business, and everyday life. But some artificial intelligence experts advise caution, noting that the tool doesn’t understand the information it provides and is inherently prone to making things up.
The situation may be particularly vexing for some of Google’s AI experts because the company’s researchers developed some of the technology behind ChatGPT — a fact Pichai alluded to in a post on the Google blog. “Six years ago, we refocused the company around artificial intelligence,” Pichai wrote. “Since then, we’ve continued to invest in AI across the board.” He name-checked both Google’s AI research division and work at DeepMind, the British AI startup that Google acquired in 2014.
ChatGPT is built on top of GPT, an artificial intelligence model known as a transformer first invented at Google that takes a string of text and predicts what will happen next. OpenAI gained notoriety for publicly demonstrating how feeding massive amounts of data into transformer models and increasing the power of the computer running them could create systems capable of generating language or images. ChatGPT improves on GPT by allowing people to provide feedback on different responses to another AI model that adjusts the output.
Google, by its own admission, has chosen to tread carefully when it comes to adding LaMDA technology to products. In addition to hallucinating misinformation, AI models trained on text taken from the Internet tend to display racial and gender biases and repeat hate speech.
Those limitations were highlighted by Google researchers in a 2020 draft research paper that argued caution should be taken with the text-generating technology, which irked some executives and led to the company firing two prominent ethical AI researchers, Timnit Gebra and Margaret Mitchell.
Other Google researchers who worked on LaMDA’s technology became frustrated with Google’s hesitation and left the company to create startups using the same technology. It seems that the appearance of ChatGPT has inspired the company to accelerate the implementation of text generation capabilities in its products.