Google has showed how it is building on artificial intelligence across its businesses, including a beefed-up Gemini chatbot and improvements to its prized search engine as it races to compete with AI rivals.
The flurry of announcements underscores Google’s efforts to refresh its products since Microsoft-backed OpenAI’s 2022 launch of ChatGPT dazzled the public, threatening the incumbent’s long reign over online search and AI.
Among Google’s latest salvos was an addition to its family of Gemini 1.5 AI models known as Flash that is faster and cheaper to run; a prototype called Project Astra, which can talk to users about anything captured on their smartphone camera in real time; and search results categorised under AI-generated headlines.
“This is a moment of growth and opportunity,” Google CEO Sundar Pichai told reporters when asked if the AI updates could risk Google’s most profitable business.
The product presentation at Google’s annual I/O developer event in Mountain View, California followed a shorter showcase by rival OpenAI on Monday. OpenAI demonstrated how ChatGPT could voice answers with human-like intonation to any written or visual prompt. The start-up’s CEO, Sam Altman, wrote that OpenAI had delivered software that “feels like AI from the movies”.
Google’s news at times covered similar ground, underscoring the fierce competition between the two AI developers.
For instance, Google DeepMind has worked to build technology that can carry out day-to-day tasks for consumers. Early results have manifested in Project Astra, a tool that can use a smartphone camera and draw conclusions about the world around it.
Universal AI agent
In a demo video shown during Google I/O, a user deployed it to identify a speaker and locate glasses they had left in another part of the room. The company also teased how it could pair Project Astra with what it calls Gemini Live, a potentially more natural-sounding voice and text aide than its Google Assistant of the past.
Demis Hassabis, CEO of Google DeepMind, said of the work behind Project Astra: “We wanted to build a universal AI agent that can be truly helpful in everyday life.”
Another area in which Google showed how it is facing off against competitors was video generation. The company teased Veo, an AI model that can spin up 1080p-resolution videos lasting longer than a minute, available to approved creators on a preview basis, filmmaker Donald Glover among them. OpenAI has promoted film-conjuring software of its own among Hollywood executives, enthralling and worrying the creative industry.
Read: Google says its AI is ready for business
Google also announced improvements to its Gemini Pro 1.5 model that is capable of making sense of a massive amount of data. On Tuesday, it said it was doubling that amount, to two million tokens, meaning the AI potentially could answer questions when given thousands of pages of text or more than an hour of video to ingest.
The Pro model — starting with prompt sizes of up to a million tokens, or pieces of data — will also be available to subscribers to Google’s Gemini Advanced service.
Google also shed light on its efforts to power AI with new computing chips and revamp its namesake search engine.
The company announced a sixth-generation tensor processing unit (TPU), which aims to give it and its Google Cloud customers an alternative to industry heavyweight Nvidia’s powerful processors. The new chip will be available to its cloud customers in late 2024, Google said.
Meanwhile, for US users of Google Search browsing the Web in English, the company said it soon will use AI to help organise search results for queries on dining, recipes, and eventually movies, books and other content.
Also for Google Search, the company is rolling out AI Overviews to all users in the US this week, after a long period of public testing since last year’s I/O event. The feature uses generative AI to synthesise information and answer more complex queries for which there is no simple answer on the web.
Analyst Jacob Bourne of eMarketer said: “The AI Overviews launch reception this week will be an indicator of how well Google can adapt its Search product to meet the demands of the generative AI era.”
He added: “To maintain its competitive edge and satisfy investors, Google will need to focus on translating its AI innovations into profitable products and services at scale.”
Ads will remain in slots throughout a given Google webpage, the company said, and AI Overviews will roll out to more than a billion people by year-end.
The company also showed off an experiment that will let users ask questions of videos they upload to Google Search, as they can do with images today. The company demonstrated how this could help diagnose what is wrong with a broken record player. — Max A Cherney and Jeffrey Dastin, with Anna Tong, Yuvraj Malik and Noel Randewich, (c) 2024 Reuters
