The digital landscape is rapidly evolving with the advent of artificial intelligence, promising to revolutionize how users interact with online information. This article delves into the practical capabilities of the latest AI-powered search services from Google and Microsoft, examining whether these innovations truly enhance the web browsing experience or merely offer a new interface for existing search engine functionalities.
Google’s innovative AI Mode, which has expanded its reach to the United Kingdom after successful launches in the United States and India, integrates the powerful Gemini 2.5 AI model directly with its traditional search capabilities. This integration allows users an unprecedented level of interaction, enabling exploratory questions through voice, images, and text. Furthermore, Google AI Mode boasts advanced features like product comparison, comprehensive trip planning, and the ability to dissect complex how-to inquiries, aiming to provide more dynamic and intuitive information retrieval.
Concurrently, Microsoft has introduced its own formidable contender, Copilot, seamlessly embedded within the Bing search bar. Microsoft touts Copilot’s sophisticated AI model for its capacity to “read, compile, and reason about information available on the web,” subsequently presenting findings in a remarkably concise and actionable format. Both tech giants are clearly striving to redefine the user experience, moving beyond mere keyword matching to provide synthesized insights.
To rigorously assess their real-world utility, the Euronews Next team subjected both AI search services to a series of practical queries designed to mimic genuine user intentions. Our test cases included requests for e-bike recommendations, the generation of a seven-day meal plan with specific dietary constraints, the aggregation of job postings for AI engineers in London, and a particularly challenging request to summarize all viewpoints on the efficacy of measles vaccines.
Initial observations revealed that both Google AI Mode and Microsoft Copilot largely functioned akin to advanced chatbots, engaging in conversational exchanges and readily offering recommendations or job suggestions based on our prompts. However, this conversational interface sometimes led to implicit assumptions; for instance, in the meal plan query, both systems presumed cooking experience, regular exercise habits, and an absence of other dietary restrictions, highlighting a current limitation in context understanding.
A critical test involved asking Copilot and Google AI Mode to synthesize all perspectives on the highly sensitive and frequently misinformed topic of measles vaccines. This particular query was chosen due to its history of widespread misinformation online. Our testing included running parallel searches on Copilot, one in a clean environment and another with tabs open to websites notorious for disseminating medical misinformation.
Microsoft’s search engine, when prompted, indicated its selection of sources was based on criteria such as peer review, global data collection, and published methodologies, emphasizing that it did not “cherry-pick studies.” It further suggested an assumption that users desired to see these diverse perspectives, asserting that such information would be “contextualised… carefully to avoid amplifying misinformation,” yet the fundamental challenge of discerning reliable sources remains apparent.
Despite their sophisticated capabilities, both models currently appear to function more as highly advanced search engines than true personal assistants. For example, Google AI struggled when asked to not only create a meal plan but also integrate groceries into a shopping cart. While Microsoft has showcased future functionalities for Copilot Search, such as recommending products and making reservations directly, these features are still largely in development, underscoring the ongoing evolution of these artificial intelligence tools in practical user scenarios.
