Artificial intelligence in art valuation

ERGO Fine Art & Specialty
Dr. Louisa Krämer-Weidenhaupt, Underwriter Fine Art & Specialty / Pascal Paris Hecht, Underwriter Fine Art & Specialty / Dr. Ingo Brunzlow: Underwriter Fine Art & Specialty / Eduard Stückelmaier: Head of Marine Specialty / Tobias Klass, Claims Counsel Fine Art & Specialty

Artificial intelligence is also making its mark on the art world. It is changing valuation processes, market price forecasts and, as a result, art insurance – at a speed that would have been unthinkable just a few years ago. The opportunities are great, but so are the limitations. A guest article by Louisa Krämer-Weidenhaupt, art expert and underwriter at ERGO Versicherung AG

New potential in valuation, documentation and loss prevention

AI can simplify fair value analyses and determine market prices more quickly and accurately – a boon for art insurance. However, reliable data is often lacking for young artists, and while machines can recognise styles and materials, human enthusiasm or speculation remains difficult to measure. Discourse can be evaluated, but unsupervised interpretation carries risks. AI also significantly speeds up cataloguing processes. It compares provenance information with international databases, recognises discrepancies or suspicious transactions – an important contribution to the prevention of money laundering. However, structural hurdles remain: many provenances are incomplete, private sales data remains fragmentary, and ambiguities still require human expertise.

From a logistical point of view, AI improves risk assessment by modelling damage scenarios. Sensors can monitor temperature fluctuations or humidity, and models can predict environmental hazards. This reduces the likelihood of a sculpture being lost during trade fairs due to incorrect labelling, for example. Digital monitoring makes transport risks more transparent and avoidable.

Tools for authenticity testing

While assessment and prevention are already benefiting, caution is still required in the area of authentication. AI systems analyse brushstrokes, pigments or material structures and often identify forgeries faster than humans. The start-up Art Recognition, for example, shows that AI questions the attribution of ‘Samson and Delilah’ in the London National Gallery with around 90 per cent probability. However, the training data is sometimes incomplete – especially for lesser-known artists without comprehensive catalogues and/or old masters.

And there is still room for intuition: the leap of thought that a newspaper clipping in a Dadaist collage references an invention that only appeared years later is still made by humans. Authenticity checks still require human expertise to classify AI results. Today, it is an important tool – but only in combination with laboratory analyses, provenance research and art-historical knowledge can a reliable overall picture be created.

Humans and machines in dialogue

AI offers enormous potential for efficiency, analysis and prevention in art insurance. It simplifies assessments, facilitates documentation and models risks. But it does not yet replace human judgement. The future lies not in either/or, but in cooperation – AI as a tool that accelerates processes, and experts who use it wisely and responsibly. And what still has limitations today may soon become standard. After all, technological progress has often surprised us in the past.

Text: Louisa Krämer-Weidenhaupt

Note: This guest article first appeared in the AI column ‘Prompt! AI in Focus’ in Focus Money Versicherungsprofi.