The Associated Press (AP) recently rolled out an AI-powered search capability that gives customers the ability to find needles in haystack after virtual haystack located throughout their vast archives.
AP is more than 176 years old and has photo and video archives – tens of millions of assets – that stretch back to the beginning of both of those formats.
The challenge: Metadata descriptions and captions have not been consistent over the archives’ long history, making it difficult to find exactly what is being looked for – especially for anything older than approximately ten years.
The solution: New AI tools allow customers to search visuals by both concepts and detailed descriptions.
The consequence: The results are dramatic. All users – customers, journalists and researchers – can find assets that have since been buried in the archive. For example, searchers can ask for – and discover – photos and videos of ‘people wearing yellow clothes drinking soda in the stands at a sporting event’ or ‘Winston Churchill feeding birds in a park’. Seek and there’s a good chance you will find!
But be aware: AI is not, currently, always best – sometimes traditional metadata produces better results. Join this session to hear lessons on when AI-powered search excels and when metadata search is more useful. This will lead to a discussion of what to expect next as search mechanisms become more powerful – it’s an exciting time.
