What Are The Risks When It Comes To AI Search?
Google has introduced AI Mode, a tab designed to answer complex questions in a single response capsule, including links. This feature utilizes a method called "query fan-out," which breaks down queries into multiple parts and performs numerous searches…
Google has introduced AI Mode, a tab designed to answer complex questions in a single response capsule, including links. This feature utilizes a method called "query fan-out," which breaks down queries into multiple parts and performs numerous searches to compile comprehensive answers. AI Mode is powered by the Gemini 2.5 model.
Additionally, Google has previewed two new features: Deep Search, which generates referenced reports quickly, and Search Live, which enables real-time camera chat. AI Mode is expected to provide more web links than traditional search results, supporting content creators.
Mozilla's Firefox currently relies heavily on revenue from Google as the default search engine. A potential antitrust ruling could affect this partnership, compelling Mozilla to explore alternative options such as Microsoft's Bing. Mozilla has previously experimented with AI answer engines like Perplexity, indicating a shift towards AI search tools.
Google has introduced AI Mode, a tab designed to answer complex questions in a single response capsule, including links.
A study by the Tow Centre for Digital Journalism evaluated 8 AI search tools and found that over 60% of queries resulted in incorrect answers. ChatGPT and Gemini displayed limitations in handling certain topics, while tools like Grok and Perplexity retrieved paywalled content, indicating issues with the Robot Exclusion Protocol. The study highlights the need for AI search tools to improve accuracy and reliability.
Research indicates that AI tools often misattribute sources, leading to issues with traffic and credibility for publishers. For instance, Google’s Deep Search incorrectly credited sources in a significant number of instances. Enhancing citation practices is crucial for maintaining trust in AI-generated content.
Researchers suggest implementing public error dashboards, clear crawler identifiers, and transparent lists of content sources. These measures would enable editors to monitor AI tool performance and advocate for necessary corrections. Additionally, the introduction of enforceable "do not crawl" options and visible source panels could enhance accountability and user trust.
Based on reporting by techround.co.uk.
