Google CEO Sundar Pichai announced on Thursday that the company will soon be incorporating cutting-edge A.I. technology into its search engine.
In response to the rising threat posed by Microsoft-backed chatbot ChatGPT, Google has been testing some of these new features among its employees. The features include “Apprentice Bard”, a chatbot, and new search desktop designs that could be used in a Q&A format. Pichai emphasized that the company’s conversational A.I. model LaMDA (Language Model for Dialogue Applications) will soon be made available to the public where users can “interact directly” with its advanced language model.
A.I. was a major theme during the company’s Q4 earnings call this Thursday, with Pichai stating that “A.I. is the most profound technology we are working on today.” This focus on A.I. comes amid challenges facing the company’s advertising business and increasing competition from Microsoft. The company reported its fourth consecutive quarter of missed earnings and revenue expectations, with a 2% decline in Google’s Search and Other revenue and an 8% drop in YouTube’s advertising revenue.
Google is under pressure from ChatGPT, which has the potential to disrupt the entire web search industry with its creative answers to complex queries. Microsoft is rumored to be considering integrating ChatGPT into its search engine, Bing, which has reportedly sparked the interest of Google co-founders Larry Page and Sergey Brin.
In addition to these search improvements, Google also announced that its A.I. arm DeepMind will be consolidated into Google, instead of being included in the “Other Bets” segment, which includes long-term projects such as self-driving cars and venture capital investments. Pichai also stated that the company will be providing new tools and APIs for developers and partners to “discover new possibilities with A.I.”
However, Pichai warned that the implementation of large language models is still in its “early days” and the company will need to scale slowly. Jeff Dean, a Google employee, stated that the company has a greater “reputational risk” in providing incorrect information, which is why it is proceeding more cautiously than smaller startups.
Prior to the announcement, this week researchers at Google unveiled the research results of a new music A.I. tool dubbed “MusicLM.”
Unlike other A.I. music tools, the tool is based on a limited set of descriptive words, which operates similarly to language models like ChatGPT, using machine learning algorithms trained on vast amounts of data to produce human-like results.
The team behind MusicLM believes the tool could provide a quick and easy way for anyone to create high-quality music selections. However, the researchers have also recognized some risks associated with the machine learning process, including biases present in the training data that may affect music generation for underrepresented cultures.
The team plans to continue to study the system’s results and limit biases through further development and testing. The team also aims to improve the system by incorporating lyrics generation, text conditioning, and better voice and music quality in the future.