How AI is transforming the way we discover art and culture online

The quick wins that make a big difference

Before diving into the exciting possibilities, we focused on some fundamental issues that were creating friction for users. These might sound technical, but they have a real impact on whether people can find what they're looking for.

Search that's too generous
Tate's current search uses an 'OR' operator between terms, meaning a search for "opening times" returns pages containing either "opening" or "times" - not necessarily both. This floods users with non-specific results, making it harder to find what they actually need.

Past events dominating results

With years of archived exhibitions and events, historical content was often ranking higher than current, relevant information. We collaborated with Tate to implement a 'decay' function that gradually reduces the prominence of past events as they get older.

Missing connections
Users searching for "JMW Turner" weren't finding the artist page for "Joseph Mallord William Turner" because the search couldn't make that connection. Implementing smart synonyms would bridge these gaps.

These fixes alone would dramatically improve the search experience, but Tate were keen that this project pushed further in search of a more forward-thinking user experience.

Exploring the art of the possible

The most exciting part of this project was prototyping three completely different approaches to search, each powered by vector search and AI - technologies that understand meaning and context rather than just matching words.

Prototype 1: Search journeys

Imagine a search that doesn't just give you results, but curates a journey through content. Search for "Turner" and instead of a simple list, you get a guided tour: the artist's page, followed by key artworks, then a related contemporary artist, ending with a current exhibition you could visit. Behind the scenes, an LLM analyses the content and builds these intelligent pathways, creating a dynamic 'tour guide' that explains why each result might be relevant to you. A persistent sidebar follows you around the site, maintaining your search context as you explore.

Prototype 2: Intelligent grouping

Rather than a flat list of results, this approach clusters content intelligently using AI embeddings to understand relationships between different types of content. Events appear together in one block, whilst a custom artist section might show their artwork alongside biographical information and upcoming exhibitions. Most intriguingly, it supports visual search - type "mountain" and find artwork containing mountains, or "red" to discover artwork dominated by red tones.

Prototype 3: Conversational discovery

This prototype treats search as a conversation, using retrieval augmented generation (RAG) to pull relevant content and pass it to an LLM for natural language responses. Users can refine their queries naturally: "Show me abstract artwork... now show me green ones... more geometric..." The system remembers context throughout the conversation and can answer complex questions by intelligently combining information from across Tate's collection to provide rich, informed responses.

The bigger picture

What struck us most about this project was how it highlighted the difference between finding information and enabling discovery. Traditional search is about retrieval - users know what they want and need to find it quickly. But in cultural contexts, search becomes about serendipity, connection, and inspiration.

The prototypes we built aren't just about better technology - they're about understanding that someone searching for "Impressionism" might be a student researching for an essay, an art lover planning a visit, or someone who's just curious about a painting they saw on social media. Each has different needs, but all deserve an experience that enriches their understanding.

Making it happen

The beauty of this approach is that it doesn't require throwing everything away and starting fresh. The immediate improvements could be implemented quickly and would make a real difference to users today. The more adventurous concepts could be tested and refined over time.

We've prioritised the recommendations based on impact and effort, starting with the technical fixes that would immediately improve the current experience, then moving towards the more innovative approaches that could transform how people discover art and culture online.

Looking ahead

This project reinforced something we see across all our work: the best digital experiences happen when you solve immediate user problems whilst also pushing the boundaries of what's possible.

Search might seem like a technical challenge, but at its heart, it's about human curiosity and the desire to learn, explore, and connect. In cultural institutions like Tate, search has the potential to be more than functional - it can be transformational.

The future of museum discovery is about meeting people where they are, understanding what they're trying to achieve, and helping them find not just what they're looking for, but what they didn't know they were looking for.