Vision-language models (VLMs) have made interactive digital museums increasingly feasible by connecting 3D digitization with natural-language artifact exploration. However, in cultural heritage domains such as ancient Greek pottery, reliable VLM assistance is limited by two challenges. First, open-ended interpretation requires grounding fine-grained 2D/3D visual evidence in specialized curatorial knowledge, yet the retrieval process may introduce weak sources and unverifiable references. Second, when the available evidence is incomplete, noisy, or ambiguous, VLMs often produce confident but unsupported answers instead of calibrated uncertainty. To address these challenges, we propose VaseMuseum, a lightweight and modular multimodal agent framework for intelligent digital museums of ancient Greek pottery. VaseMuseum combines an interactive virtual museum with VaseAgent, which supports both 2D images and 3D artifacts through multimodal perception, 3D-aware reasoning, external knowledge retrieval, and inference-time reliability control. Specifically, VaseAgent retrieves evidence from authoritative web and museum knowledge sources, and source-level control selects diverse and verifiable evidence before generation. Meanwhile, response-level control checks generated claims against the evidence pool and encourages neutral, evidence-bounded answers when support is insufficient or conflicting. Moreover, a training-free GRPO-style selection mechanism favors responses with valid references and calibrated confidence without updating the VLM backbone. Experiments in a realistic digital museum simulation show that VaseMuseum improves citation validity, reduces hallucinations on knowledge-intensive queries, and produces more neutral answers under ambiguity compared with search-enabled VLM baselines.
Alex: Welcome to another episode of ResearchPod.
Sam: Today we're looking at a new research framework designed to help people explore ancient Greek pottery in a museum — guided by artificial intelligence. And the central challenge the researchers are tackling is one that might surprise you: AI systems tend to act like confident experts even when they don't actually know the answer.
Alex: That sounds like a real problem in a museum context. What does that actually look like in practice?
Sam: Researchers call it a "hallucination." If you ask an AI about a specific vase, it might invent a painter's name or a historical date — something that sounds completely plausible but is entirely made up. The model isn't lying on purpose. It's just trained to be helpful, and being helpful often means sounding certain, even when certainty isn't warranted. In a museum, where you're trying to teach people real history, that's a serious problem.
Alex: So the question the researchers are asking is: how do you build an AI guide that's honest about what it doesn't know?
Sam: Exactly. Their framework — called VaseAgent — treats every visitor question as an evidence-seeking process. Rather than drawing on internal memory alone, the AI actively searches for verified, scholarly sources and anchors its answer in that evidence. The authors argue that for cultural heritage, you can't afford to bluff.
Alex: It's the difference between a student who guesses on a test and one who actually looks up the source material before answering. How does the system avoid getting buried under too much information?
Sam: It uses a two-step filter. First, it gathers information from the web. Then it applies what the researchers call "source-level control." Think of it like a librarian who checks the credentials of every book before putting it on the shelf. Weak or unverifiable sources get discarded, so only high-quality, relevant evidence enters the AI's working space before it even attempts a response.
Alex: That makes sense for filtering the input. But what happens when the evidence itself is conflicting — when two experts genuinely disagree on who painted a vase?
Sam: That's where a second layer called "response-level control" comes in. After drafting an answer, the system checks that draft against the evidence it just gathered. If the evidence is missing or contradictory, the system is designed to stay neutral — it might say, "Scholars are divided on this attribution," rather than picking a side just to sound authoritative. It's built to express uncertainty when the historical record is genuinely unclear.
Alex: That's a meaningful shift. But does it require retraining the entire model every time new historical data comes to light?
Sam: Not at all. They use a technique called Group Relative Policy Optimization — GRPO. Rather than rewriting the AI's underlying knowledge, it works more like a final editorial review. The system generates several candidate answers, then selects the one most supported by the gathered evidence. The AI's core doesn't change — just the process of choosing what to say.
Alex: So it's less about making the AI smarter in the traditional sense, and more about building a layer around it that enforces honesty about what it knows and what it doesn't.
Sam: Right. And the researchers tested how far that layer can scale. They looked at what happens when you increase the number of candidate answers the system considers before settling on one. Think of it like inviting more experts into a room to debate a question — you tend to get a more reliable conclusion, but it takes longer for everyone to speak.
Alex: So at some point, the visitor is just standing there waiting for the AI to finish its internal debate?
Sam: Precisely. They found that reliability improves as you add more candidates, but those gains level off. Beyond a certain point, the response time keeps growing while the quality stays flat. So the researchers identified a practical sweet spot — enough candidates to get a well-supported answer, without making the visitor wait.
Alex: What about the opposite problem — a rare vase that barely exists in any online archive? Can the system still function, or does it just fail silently?
Sam: That's an important limitation the researchers are candid about. The system's performance is strictly bounded by what's actually available in external sources. If a specific, obscure artifact has almost no documented history online or in accessible archives, the system cannot generate knowledge from nothing. In that case, it will report that it couldn't find sufficient evidence — which is exactly the cautious, honest behavior the researchers were aiming for.
Alex: So it's not a box that knows everything. It's more like a diligent researcher who knows when to stop and say, "I don't have enough to be confident here."
Sam: That's a fair description. And the researchers argue that for cultural heritage specifically, that "I don't know" response is a feature, not a failure. It's a clear improvement over a generic model that might confidently invent a plausible-sounding history just to fill the silence. The logical next step, they suggest, would be connecting these systems to private museum archives — records that aren't publicly available online. That would let the AI function as a precise research assistant not just for casual visitors, but for art historians who need to cross-reference verified, specialist records.
Alex: It's a thoughtful approach to a genuine problem — building technology that respects the complexity and uncertainty of history, rather than papering over it with false confidence. Thanks for listening to ResearchPod.