cd /news/artificial-intelligence/the-search-agent-that-stopped-foolin… · home topics artificial-intelligence article
[ARTICLE · art-83936] src=pub.towardsai.net ↗ pub= topic=artificial-intelligence verified=true sentiment=· neutral

The Search Agent That Stopped Fooling Itself

A team from the National University of Defense Technology has identified a failure mode in reinforcement learning-trained search agents, called retrieval-equivalence collapse, where agents repeatedly retrieve the same documents despite generating different queries. Their proposed fix, Harness-G, replaces free-form queries with a fixed menu of choices and improves performance by nearly 11 points on average across six benchmarks at smaller model sizes and about 4 points at larger ones.

read1 min views3 publishedAug 2, 2026
The Search Agent That Stopped Fooling Itself
Image: Pub (auto-discovered)

Member-only story

Why teaching an AI to pick from a menu beats letting it write its own questions #

Here is a strange failure that shows up when you train an AI agent to search for answers using reinforcement learning. You ask it to research a question. It writes a search query, gets some results, decides it needs more information, and writes a new query. On paper this looks like exploration. The agent is trying different phrasings, chasing different angles, behaving like a curious researcher.

But if you actually check what documents it pulled back, something disappointing happens. Ten different rewritten queries for the same question often return the same handful of documents. The agent thinks it is exploring. It is actually going in circles wearing different outfits.

A team from the National University of Defense Technology ran the numbers on this and gave the problem a name: retrieval-equivalence collapse. Their paper, “Harness-G: A Graph-Structured Harness for Search Agents,” does two things. First, it proves this collapse is real and explains why it quietly wrecks the training process. Second, it proposes a fix that swaps free-form search queries for a fixed menu of choices, paired with a smarter way to reward the agent for good decisions. The result beats the strongest existing method by close to 11 points on average across six question-answering benchmarks at smaller model sizes, and by about 4 points at larger ones.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @national university of defense technology 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/the-search-agent-tha…] indexed:0 read:1min 2026-08-02 ·