Antiques in the Age of AI

Although the study of machine intelligence began in the 1950s, the field accelerated dramatically in 2022 with the release of OpenAI’s ChatGPT — the first widely accessible large language model available through a simple, free web interface. Its rapid uptake across industries soon followed. The antiques world, often slower to embrace new technologies, did not adopt AI immediately, but its use is now growing across auction houses, museums, and private collecting. As in many fields, those who adapt tend to move forward.

What do specialists really think about AI’s growing role in their work? How are auction houses and museums using it, and where do they see risks or ethical concerns? And for collectors, where can AI genuinely help when buying antiques, and where does human expertise still matter most? Read on to find out!

AI and Antiques

AI in antiques is shaped less by the technology itself and more by the professional contexts in which it appears. Dealers, auction specialists, and museum professionals encounter it through different pressures, expectations, and ethical frameworks, which means its impact is uneven — sometimes transformative, sometimes contentious. What is clear is that AI is no longer peripheral: it is becoming woven into the daily mechanics of the trade, even as its limitations remain seen.

For dealers, often working independently or in small teams, AI has become a practical tool for streamlining admin, research, and cataloging. Feeding basic descriptions and images into tools such as Copilot can generate draft catalog notes, while AI also supports social media planning, content creation, and even image generation. These tools can compress days of work into hours, but only when used responsibly. Miscataloging caused by unchecked AI outputs is now a recognized issue, particularly in categories where subtle distinctions in material, technique, or maker define value. AI can misread glaze types, conflate regional carving styles, or misattribute workshop production to named artists. The ethical issue here is not simply error — it is accountability. Dealers must ensure that AI-generated text does not replace connoisseurship, and that speed does not erode the interpretive care that underpins trust in the market. There is also a quieter tension: as more dealers rely on the same tools, catalog language risks becoming homogenized, flattening the individuality of expertise that traditionally distinguishes one dealer’s eye from another’s.

In auction houses, AI’s presence extends beyond workflow and into the objects being sold. Christie’s New York’s 2025 “Augmented Intelligence” sale — the first major auction of AI-generated artworks — demonstrated both the market’s appetite for novelty and its discomfort with the legal and ethical ambiguity surrounding AI-generated art. More than 6,000 artists signed an open letter arguing that many of the works on offer relied on models trained on copyrighted material without permission. Despite the backlash, the sale achieved $728,784, yet no comparable auction has followed, suggesting that the commercial sector is still unsure how to position AI‑generated objects within a market built on authorship, provenance, and artistic intent. However, individual AI‑generated works have been sold at other auctions; for example, “A.I. God. Portrait of Alan Turing,” a mixed‑media canvas created by the humanoid robot Ai‑Da, sold for $1.08 million at Sotheby’s in 2024, making it the highest price ever achieved at auction for an AI‑generated portrait.

Behind the scenes, however, AI is quietly reshaping auction workflows: assisting with comparables, analyzing past sales, and supporting marketing teams with automated copy and audience segmentation. Even so, specialists remain aware that valuation and connoisseurship rely on knowledge built through handling thousands of objects, something AI cannot replicate, but more on that below.

Museums, operating within heritage rather than commercial frameworks, use AI differently. Their priorities are preservation, access, and interpretation. AI supports virtual guides, personalized tours, collections management, and search tools. It can also improve accessibility: at the Museum of Tomorrow in Rio de Janeiro, AI provides real-time sign-language translation and personalized audio descriptions. At the Victoria and Albert Museum in London, machine-learning projects are underway to identify and flag problematic terminology in historic cataloging — part of a broader ethical conversation within the sector about language, representation, and institutional responsibility.

AI also intersects with collecting practices. While not yet “antique,” today’s digital artifacts will become tomorrow’s historical objects. In 2023, the Victoria and Albert Museum acquired a July 2022 issue of “The Economist” featuring a cover generated by an early version of Midjourney (a generative AI program that creates images and animations from text prompts), capturing a cultural moment in which intangible digital processes become embodied in tangible objects. These acquisitions raise forward-looking questions: how do institutions preserve the context of AI-generated material, and how will future researchers interpret objects created by systems whose training data may never be fully transparent?

Across the trade, AI is a tool that accelerates certain tasks, challenges others, and forces institutions to articulate what expertise means in an age of machine-generated knowledge. The antiques world — traditionally slow to adopt new technologies — now finds itself negotiating a landscape where efficiency, ethics, and authenticity intersect in increasingly complex ways. The question is no longer whether AI belongs in the sector, but how its presence reshapes all tasks related to the discovery, research, promotion, and selling of antiques.

Industry Expert Insights

An interview with Otto Billström, Head of Modern Decorative Art & Design at Bonhams London, offers a clear view of how AI is used within a commercial auction house setting.

AI tools are a valuable addition to his work, particularly as large language models such as ChatGPT, Copilot, or Gemini can serve as a useful first step in identifying well-known artists, marks, or objects. However, as with most technologies, there are significant limitations. Billström notes: “While it can recognize basic shapes and patterns, the tools are far from refined enough to tell makers who work in similar styles apart and will usually err on the side of a more well-known maker. Similarly, it is not capable of telling a genuine work apart from a fake. It is therefore important to always do your own research and arrive at your own conviction.”

Despite their usefulness, AI tools cannot replace human expertise or relationships. Auction specialists, dealers, historians, and even enthusiastic collectors remain indispensable within the antiques ecosystem. The trade is built on relationships and human interaction, often spanning generations of buyers returning to the same sellers. It becomes richer when emotional stories and personal connections are shared about objects and their meaning to families or individuals. AI can scan the internet for patterns, but it cannot form the deeper connections or develop the lived expertise that, as Otto says, is the “bedrock of the market.”

When discussing valuations and market predictions, Billström emphasizes that AI tools are not yet advanced enough to distinguish sale context or provenance, both of which can significantly influence value. Single-owner collections often achieve higher prices, particularly when items have noteworthy provenance or historical importance. For example, in May 2026, Christie’s hosted “MASTERPIECES: The Private Collection of S.I. Newhouse.” One of the most influential collectors of the last century and co-owner of the Condé Nast publishing empire, Si Newhouse’s collection realized $630,825,000, with works by artists such as Pablo Picasso and Constantin Brancusi. “An AI will simply look at the price achieved, meaning in normal market conditions, the result may be anomalous and therefore create false expectations,” says Billström.

How You Can Use AI When Buying Antiques

You can use AI much like professionals do: as a helpful starting point when researching objects, forming first impressions, and organizing your own collection. While it should never replace expert advice or connoisseurship, AI can make early research faster, clearer, and more accessible.

AI image‑recognition tools can assist with object identification by offering initial suggestions about makers, styles, or periods based on a photograph. This can be particularly useful when you browse markets, fairs, or online listings, giving you a quick sense of what you might be looking at. These results, however, should always be treated cautiously, as AI may misidentify similar styles or default to well‑known names.

AI can also help you deepen your understanding of an object by summarizing historical context, explaining stylistic features, or outlining typical materials and techniques. It can translate marks or inscriptions, provide short biographies of makers, or clarify differences between related styles. It can also help suggest other resources, such as books and articles that could deepen your knowledge. This kind of interpretive support can make unfamiliar categories easier to navigate, especially if you are new to collecting.

Finally, AI is a practical tool for cataloging your personal collection. It can turn photographs and brief descriptions into draft catalog entries, help you organize objects by maker, period, or material, and create consistent terminology across your records. If you want to maintain a clear archive of your acquisitions — whether for insurance, valuation, or personal interest — AI can streamline the process significantly. Accuracy remains essential, so all AI‑generated information should be checked before being added to a permanent record.

AI and antiques form a revealing intersection between tradition and modernity. The technology accelerates research, expands access, and reshapes workflows, yet it also exposes how much of the field still relies on human judgment, material literacy, and cultural interpretation.

As dealers, auction houses, and museums negotiate its possibilities and limits, AI becomes less a disruptive force and more a lens through which the sector reconsiders expertise, ethics, and the future of collecting.

Its presence does not replace the foundations of the trade; instead, it highlights them, reminding us that innovation in antiques is most meaningful when it strengthens, rather than shortcuts, the knowledge that has always defined the field.