Google Search undergoes a transformation with AI-generated answers
Google is set to revolutionize its Search feature by introducing AI-powered changes, as announced at the I/O, its annual conference for developers. The tech giant aims to evolve Search from a simple website finder into a tool, that directly provides answers and assists users in planning events and brainstorming ideas. Liz Reid, Vice President and Head of Google Search, stated, "[With] generative AI, Search can do more than you ever imagined."
Google responds to rise of generative AI
Google's changes to its primary revenue generator, Search, are a response to the rise of generative AI since OpenAI's ChatGPT was launched at the end of 2022. Several AI-powered apps and services like ChatGPT, Anthropic, Perplexity, and Microsoft's Bing (powered by OpenAI's GPT-4), have challenged Google by directly providing answers instead of merely presenting a list of links. Reid stated that Google aims to bridge this gap with its new features in Search.
AI overviews now available to millions
Starting immediately, most search queries will display complete AI-generated answers at the top of the results page in the US. This feature was first announced at Google I/O in 2023 as part of the company's Search Labs platform. Reid noted that users who opted to try the feature through Search Labs have used it "billions of times" so far. Google plans to make this facility accessible to over a billion people worldwide by year-end.
AI-generated answers boost traffic for publishers
Reid mentioned that any links included as part of the AI-generated answers, receive more clicks than if they appeared as a traditional web listing, which is a concern for publishers. "As we expand this experience, we'll continue to focus on sending valuable traffic to publishers and creators," Reid wrote. This indicates Google's commitment to supporting content creators and publishers, even as it revolutionizes its Search feature with AI technology.
Google Search to offer AI-organized results
In addition to AI Overviews, certain search queries in English in the US related to dining, recipes, and later movies, hotels, music, books, shopping and more will display a new search page where answers are organized using AI. Reid explained that when users are looking for ideas, Search will use generative AI to brainstorm with them, and generate an AI-organized results page for easy exploration.
Google Search Labs offers more AI-powered features
Users who opt into Search Labs will have access to more generative AI-powered features in Google Search. They can use AI Overview to simplify language or delve into a complex topic in more detail. According to Reid, Search Labs testers can also ask Google complex questions in a single query, to get the answers on a single page instead of conducting multiple searches, enhancing the search experience.
Google to introduce digital watermarks in AI-generated videos
Google has unveiled plans to integrate digital watermarks into its latest video-generation tools. The initiative aims at enhancing transparency regarding the origins of its AI-generated videos. The company's new Veo model, featured in the VideoFX app, will employ Google's SynthID system for embedding these watermarks.
What is SynthID?
SynthID, a digital watermarking system developed by Google, started its deployment to AI-generated images last year. The technology embeds nearly invisible watermarks into AI-created content, enabling detection tools to identify the artificial origins of the content. Notably, SynthID watermarks would also be extended to AI-generated text.
A response to rising misinformation concerns
The move toward watermarking comes as more companies express concerns that advancements in generative AI models could lead to a surge in misinformation. Systems like SynthID provide platforms like Google, with a framework for identifying AI-generated content that might otherwise be indistinguishable from human-created content. Major tech firms like TikTok and Meta have also announced plans to support similar detection tools on their platforms.
Debates surrounding the efficacy of digital watermarks
Despite the growing adoption of digital watermarks, there are ongoing debates about their effectiveness in combating deceptive AI content. Some researchers have demonstrated that these watermarks can be easily circumvented. However, the implementation of some form of detectability for AI-created content is widely considered an essential first step toward transparency.
Gemini will now search videos to help you solve problems
Google has introduced a groundbreaking AI feature in its search engine. Now, users can upload videos of issues they're facing. The generative AI then scours user forums and other online resources to find solutions. As demonstrated by Google representative Rose Yao, Gemini can effectively assist users in troubleshooting problems using video content.
Here's a real-life demonstration
Yao showcased Gemini's capabilities using a personal example. She had purchased a used turntable and was having difficulty getting the needle to sit on the record. After uploading a video of her issue, Gemini quickly located an explainer detailing how to balance the arm for that specific model. This instance highlights how Gemini can simplify problem-solving by searching with video content.
Google's vision: Video search beyond text queries
Google believes that "Search is so much more than just words in a text box." The company asserts that many queries are about the things people see around them, including objects in motion. By integrating video into its search engine, Google aims to save users time and effort in finding the right words to describe their issues. Gemini provides an AI Overview with steps and resources for troubleshooting, enhancing its ability to understand and solve complex problems.
Additional video features and future plans
In situations where the video alone doesn't clearly define the problem, Gemini allows users to add text or draw arrows pointing to the issue. This feature further enhances its ability to understand and find solutions for complex problems. Initially available to Search Labs US users in English, Google plans to expand this feature globally over time.
Gemini Nano to boost Android's TalkBack feature
Google has unveiled Gemini Nano at I/O 2024. It is designed to bring multimodal support to Android devices. This enhancement will allow Android to process text while simultaneously understanding visuals, sounds, and spoken language. The tech giant plans to integrate this new tool into its existing TalkBack feature, which vocalizes descriptions of images for users.
Gemini Nano to enhance image understanding in TalkBack
The integration of Gemini Nano's multimodal support into TalkBack is expected to provide a detailed understanding of images. Google revealed that TalkBack users encounter approximately 90 unlabeled images daily. The company believes that Gemini Nano will address this issue by providing missing information, such as the appearance of a clothing item or specifics of a newly received photo.
Nano to operate independently of network connection
Gemini Nano is designed to operate directly on a user's device, implying it should function correctly even without any network connection. This feature ensures that users can access detailed image descriptions regardless of their internet connectivity. While Google has not provided an exact release date, it has confirmed that TalkBack will get Gemini Nano's updated features later this year.
AI-based scam detection tool for Android phones
Google has unveiled a real-time scam detection tool, at the I/O 2024 conference. This tool, currently in testing stages, uses artificial intelligence (AI) to identify potential fraudsters during phone calls. The AI system will monitor conversation patterns commonly associated with scams and alert users if such patterns are detected.
Tool will operate directly on the handset
The scam detection tool will operate directly on the device, ensuring user privacy. Google provided an example of how the AI might flag a "bank representative" asking for personal information such as PINs and passwords, which are uncommon requests from banks. However, this feature will not be immediately available on Android 15, with more details to be shared later in the year. Users will need to opt-in to use this tool.