Advanced Computing in the Age of AI | Sunday, February 12, 2023

Google Will Take on Microsoft in AI Search 

Google’s search engine dominance has been threatened overnight by an obscure AI technology that came out of the blue, and the tech giant is now taking steps to catch up.

Since its introduction late last year, the ChatGPT chatbot has been considered a significant threat to the way people search for information. The chatbot is diverse and can answer people’s questions, compile essays or poetry, or even write code.

The conversational AI’s ability to present cohesive answers is considered a threat to Google’s search engine, which for decades has been the standard bearer in the way people search for information on the internet.

OpenAI’s ChatGPT can customize answers to specific questions asked by a user, which can save time scrolling through websites.

Source: YouTube screenshot

The overnight success of ChatGPT has forced Google to call a “code-red” to take on the threat posed by the AI chatbot on its search engine business, according to a report published in December by The New York Times.

Google is holding an event on February 8 where the company promises to reimagine “how people search for, explore and interact with information, making it more natural and intuitive than ever before to find what you need,” according to the event description on the livestream page on YouTube.

Google is promising “greater access to information for people everywhere, through Search, Maps and beyond.”

The event points to a revamp of its search engine by implementing AI research projects like LaMDa, which is Google’s competitor to ChatGPT, into its mainstream search engine.

Google’s accelerated search revamp comes as Microsoft weaves ChatGPT into its Bing search engine. A ChatGPT-powered Bing search engine apparently appeared for a short time and then disappeared, according to a news report by The Verge.

Microsoft in late January said it was making a multi-billion dollar investment in OpenAI to independently “research and develop AI that is increasingly safe, useful, and powerful.”

Google LaMDa, which was announced in 2021, was largely under wraps and kept away from the public. ChatGPT – which is still in research mode – stole the thunder when it became available in an easy-to-use interface, leaving Google scrambling to catch up.

Google has an array of AI technologies in research besides LaMDA. Its ground-breaking PaLM (Pathways Language Model (PaLM), scales up to 540 billion parameters, and is significantly larger than GPT-3.5, which is the large language model powering ChatGPT.

“In the coming weeks and months, we’ll make these language models available, starting with LaMDA so that people can engage directly with them,” said Sundar Pichai, Alphabet’s CEO, according to a transcript of the earnings call on Thursday.

Pichai repeated some of LaMDa’s benefits. The company has previously talked about how AI could be used to compile or complete emails or written work, or summarize a complicated report.

“Very soon, people will be able to interact directly with our newest, most powerful language models as a companion to Search in experimental and innovative ways. Stay tuned,” Pichai said.

Google will provide tools and APIs for developers, creators, and partners, Pichai said.

“This will empower them to innovate and build their own applications and discover new possibilities with AI on top of our language, multimodal, and other AI models,” Pichai said.

Alphabet, which is Google’s parent company, is creating a tighter bond across all business units to bring AI technology to core business offerings.

DeepMind, which is a unit of Google’s parent company Alphabet, is trying to create “artificial general intelligence,” or AGI, which aims to replicate the way human intelligence works. The concept of AGI is to combine Deepmind’s research into one offering to complete human chores, find information, play games and go scientific research. The AGI concept will include computer vision, speech and natural language processing.

Google is already using AI to improve search results and in products like Google Cloud. But the delay of implementing LaMDa commercially was possibly because of Google’s pledge for responsible and ethical use of AI.

The LaMDa webpage states “being Google, we also care a lot about factuality (that is, whether LaMDA sticks to facts, something language models often struggle with), and are investigating ways to ensure LaMDA’s responses aren’t just compelling but correct.”

The battle between Microsoft and Google is now on AI-powered search engines, but also between Google and OpenAI on the core AI technologies. Large language models from Google and OpenAI are headed to parameters in excess of 1 trillion, which will make responses from AI search engines better and more accurate.

Google has claimed that it invented the transformer technology on which ChatGPT is built. Transformers do a better job in establishing relationships between words and other elements in a passage to generate more relevant answers.

For example, transformers can establish a meaning of a word, sentence and paragraph, and establish the context, relationship and connections across a written passage. That is especially important for domains like healthcare, where reliable relationships and context need to be established for accuracy.

Add a Comment