Microsoft’s Bing team published a framework describing how indexing requirements change when the goal is to ground AI answers rather than to rank search results.
The post identifies five measurement areas where the company says the two systems diverge. It also names “abstention” as a design choice for AI-powered retrieval.
What Microsoft Described
The post argues that traditional search indexing and grounding indexing share the same foundation but serve different goals.
Traditional search, the team writes, asks “which pages should a user visit?” The grounding layer asks “what information can an AI system responsibly use to construct a response?”
Microsoft identifies five categories where the measurement requirements differ.
On factual fidelity, the team notes that some ranking mismatch is tolerable in traditional search because a user can click through and evaluate. In grounding, the post describes breaking content into retrievable chunks as a process that “can distort page substance in ways that never appear in any ranking signal.”
For source attribution quality, the Bing team calls attribution helpful in traditional search but “a core signal” in grounding. Not all indexed content matters equally as evidence for an AI answer, the team adds.
On freshness, Microsoft notes a clear difference in cost. Stale content in search is a ranking problem. In grounding, the post says, “a stale fact produces a misleading response.”
For coverage of high-value facts, the post explains that a missed document in search is recoverable because alternative results exist. In grounding, the index must ensure “the specific facts and sources that people are likely to ask about are actually available and groundable.”
On contradictions, traditional search can surface one source above another and let the user decide. A grounding system can’t do that. “An AI system that silently arbitrates between contradictory sources is one that may confidently assert the wrong thing,” the team says.
Abstention And Iterative Retrieval
The post also covers two design differences between the systems.
Microsoft calls declining to answer “abstention.” For a grounding system, that’s a valid outcome when support is missing, stale, or conflicting. Traditional search doesn’t need to make this judgment because it presents options for a human to evaluate.
Iterative retrieval is the other difference. Traditional search is typically a single interaction where a query goes in and ranked results come out. Grounding systems may need to ask follow-up questions, refine retrieval based on intermediate results, and combine evidence from multiple sources.
Errors in early retrieval steps “compound through subsequent reasoning steps in ways that no human reviewer would catch in real time,” the post adds.
Context
This blog post comes after a series of moves by Microsoft to build out its grounding tooling and give publishers visibility into it.
In February,…
Source link
Disclaimer
We strive to uphold the highest ethical standards in all of our reporting and coverage. We blogs.grocliq.com want to be transparent with our readers about any potential conflicts of interest that may arise in our work. It’s possible that some of the investors we feature may have connections to other businesses, including competitors or companies we write about. However, we want to assure our readers that this will not have any impact on the integrity or impartiality of our reporting. We are committed to delivering accurate, unbiased news and information to our audience, and we will continue to uphold our ethics and principles in all of our work. Thank you for your trust and support.
Website Upgradation is going on for any glitch kindly connect at [email protected]
