Microsoft reflects on the new Bing: 71% gave AI answers a thumbs up, Bing can become repetitive after 15+ questions, a few sessions over two hours, and more
A little over a week ago, we shared an all new, AI-powered Bing search engine, Edge web browser, and integrated Chat, that we think of as Your Copilot for the Web.
Context & Ripple Effects
Microsoft had just paired a custom OpenAI model with Bing and Edge as an AI web copilot, while an early hands-on showed the product using citations for factual answers. The feedback now identifies the limit behind that initial promise: the Bing-and-Edge launch created a conversational search surface whose quality can deteriorate in extended exchanges.
That constraint matters as Bing seeks repeat use, not merely first impressions. Later coverage reports that the preview drew millions of active users and that about a third were new to Bing, making sustained chat quality relevant to whether new Bing users become durable search users.
First-order effects
- Bing users receive a positive early signal on answer quality, but people pursuing longer research-style chats can encounter repetitive responses after roughly 15 questions.
- Microsoft has an explicit reliability and session-length weakness to address in Bing Chat and its Edge integration, rather than treating favorable answer ratings as sufficient validation.
Second-order effects
- Microsoft's ability to turn Bing's AI preview into repeat usage depends more on improving multi-turn conversations, especially among the new-to-Bing users identified in subsequent usage reporting.
- As Microsoft adds task-specific features such as AI-generated shopping guides, weak long-session behavior becomes a product constraint across more than general web search.
Third-order effects
- If conversational search becomes a durable interface, competition will shift from impressive single answers toward maintaining context, variety, and usefulness across longer user sessions.
- Bing's later role as a search service for Copilot, ChatGPT, and other platforms suggests that conversational-answer quality can become infrastructure-level rather than a feature confined to Bing's own interface.
The trend: AI search is moving from a launch-era focus on answer quality toward proving that conversational interfaces can sustain repeated, multi-step use across distributed AI products.