Conversations with AI Part IV: AI-assisted knowledge libraries and curation, the search for a trusted output

This Ipsos Views white paper explores the role of AI as part of curation and how human expertise and AI speed can work best together.

The author(s)
  • Ajay Bangia Global Scale leader, Ipsos UU
  • Jim Legg Ipsos UU, US
Get in touch

For many major businesses around the world, knowing what you know is a more difficult prospect than it can seem. What’s more finding the right information when you need it can be even harder. The challenge for many working in insights is locating the right sources and writing a thorough and succinct report that’s ready to socialize. The experts who do this are curators. 

We see the need to evolve our curators into certified AI curators who are experts at applying the latest advances in Generative AI to the curation process and who know how to apply analytical frameworks to a large corpus of data. The headline: we see speed as the primary benefit that AI could deliver. 

Ipsos is developing their own custom AI-assisted knowledge library and curation solution to align with client needs. To inform this solution, Ipsos ran two pilots exploring four key questions: 

  1.  Can AI effectively find relevant information in an unstructured data library? Both classic and generative AI were tested using ingested data and analytical frameworks to try to accurately identify relevant content faster. 
  2.  Once sources are found, can the AI accurately summarize them? Or is human validation still critical? 
  3. How can Ipsos take AI-generated summaries and connect them into a coherent overall report? 
  4.  What benefits should clients expect from AI-assisted curation solutions? The pilots compared human-only curation to an AI-assisted approach. The human approach involved curators reviewing and summarizing sources into a report. The AI approach used algorithms to identify, summarize, and combine relevant information. The human output helped assess the AI for hallucinations and gaps. 

Overall, the goal was evaluating if and how AI could speed up curation without sacrificing accuracy or the need for human judgment in conveying meaningful insights. The pilots aimed to balance efficient AI document handling with expert curation, storytelling, and drawing substantive conclusions. 

Further Reading:

Conversations with AI: How generative AI and qualitative research will benefit each other

Conversations with AI Part II: Unveiling AI quality in qualitative workstreams

Conversations with AI Part III: How AI boosts human creativity in ideation workshops

Conversations with AI Part IV: AI-assisted knowledge libraries and curation

The author(s)
  • Ajay Bangia Global Scale leader, Ipsos UU
  • Jim Legg Ipsos UU, US

Society