Microsoft and Google have turned the rapid rise of generative artificial intelligence into a direct contest over internet search. Microsoft on Tuesday unveiled a new version of Bing and its Edge browser built around a next-generation OpenAI model, offering conversational answers, source-linked summaries and content generation. The launch came one day after Google announced Bard, an experimental conversational service powered by its LaMDA technology and designed to bring generative AI into search and other products.

Microsoft describes the new Bing as an “AI copilot for the web.” Its Feb. 7 launch announcement says the service combines a conventional results page with a sidebar that synthesizes information, a chat mode for more complex questions and tools that can generate material such as itineraries, emails and interview preparation. The system is entering a limited desktop preview, with wider access planned as Microsoft gathers feedback.

The new Bing changes the basic interaction model of a search engine. Instead of requiring users to move among multiple links and assemble an answer, the system can produce a synthesized response and allow follow-up questions. Microsoft says it will cite sources so users can move from the generated answer to underlying web pages.

That approach creates a new competitive opportunity for Bing, which has long operated far behind Google in search usage. It also introduces risks familiar from large language models: confident factual errors, incomplete sourcing and the possibility that a generated response could obscure disagreement among sources. Microsoft acknowledges that mistakes are possible and says it is applying safeguards developed with OpenAI around misinformation, harmful content and data safety.

An Associated Press report described the launch as Microsoft’s attempt to challenge Google by making conversational AI a central feature of search rather than a separate demonstration. OpenAI Chief Executive Sam Altman appeared at Microsoft’s event, underscoring how quickly the research partnership between the companies is becoming a consumer-product alliance.

Google answers with Bard and new AI features

Google moved first this week, announcing Bard on Monday. Chief Executive Sundar Pichai said in an official company post that Bard is initially being opened to trusted testers before wider public availability. It uses a lighter version of Google’s Language Model for Dialogue Applications, or LaMDA, to reduce computing requirements while the company gathers feedback.

Pichai also said Google plans to add AI-powered features to Search that can synthesize complex information and multiple perspectives for questions that do not have a single simple answer. Google’s challenge is different from Microsoft’s: it must adopt a technology that could improve the search experience without undermining a dominant advertising business built around the existing format of queries, results and paid placements.

A Guardian account of Microsoft’s launch emphasized that the two companies are now competing not merely over chatbot novelty but over one of the internet’s most valuable commercial gateways. The ability to answer complex questions conversationally could affect how users discover websites, how publishers receive traffic and how advertising is displayed alongside search results.

Microsoft’s OpenAI investment provides the foundation

The Bing launch follows Microsoft’s Jan. 23 announcement of a new multiyear, multibillion-dollar investment in OpenAI. Microsoft’s partnership statement says Azure will remain OpenAI’s exclusive cloud provider and that the companies will work together on supercomputing, model deployment and new consumer and enterprise experiences.

OpenAI’s own announcement says the investment follows Microsoft’s earlier commitments in 2019 and 2021 and will help expand the computing systems needed to train and deploy increasingly capable models. The arrangement gives Microsoft unusually direct access to OpenAI technology for products while providing OpenAI with the capital and computing infrastructure required for systems that are expensive to train and serve at scale.

Microsoft says Bing uses a next-generation OpenAI large language model that is more capable than ChatGPT and customized for search. The company also describes a proprietary orchestration approach called Prometheus that combines the model with current web information and Bing’s ranking systems. Those details matter because search requires freshness and attribution in ways a standalone language model trained on a fixed data set may not.

Accuracy, attribution and economics become the next tests

The first competitive question will be whether users find conversational search reliably useful. The deeper question is how the technology changes the economics of the web. Search engines currently send users to publishers, merchants and other sites while selling advertising around those pathways. If an AI system answers more questions directly, publishers could receive less traffic even when their reporting or data contributed to the answer.

Attribution is therefore not just a user-interface issue. Microsoft’s decision to include citations is an acknowledgment that generated answers depend on external information and that users need a way to verify claims. Google faces the same challenge as it begins integrating generative models into search.

Computing cost is another constraint. Large language models require substantially more processing than a traditional query, and both companies will have to determine how to operate these systems at billions-of-query scale without eroding search margins. Google’s decision to begin Bard with a lighter LaMDA model reflects that problem directly.

For now, the announcements mark an inflection point. ChatGPT demonstrated public demand for conversational AI; Microsoft is attempting to turn that demand into a challenge to Google’s search dominance, while Google is moving its own advanced models from research toward consumer products. The competition is no longer about whether generative AI will enter mainstream software. It is about who can make it accurate, fast, affordable and trustworthy enough to reshape how people navigate the web.