Client
The Economist Intelligence Unit (EIU)
Service
Data & AI
Geopolitical scenario mapping with The Economist Intelligence Unit (EIU)
EIU help their clients make decisions based on an understanding of the geopolitical landscape. Part of this is understanding the impact that geopolitical events may have on various sectors. Within a lab environment, EIU wanted to explore how AI can assist analysts when producing scenario pathways, to improve the ability to support their clients at scale.
Focussing on the right problem
Our AI Lab helped EIU explore how AI can scale the role that Analysts play within their org. Scenario pathways form an important part of the EIU offering to their clients and it was the generation of these pathways we selected as a use case with high potential value for AI.
Scenario pathways are important, but time consuming to produce given the amount of content, logical progression, lateral thinking and the structured format they follow. These factors meant this type of analysis felt like a good candidate for AI exploration, and allowed us to frame the key lab question: Can AI improve EIU’s ability to carry out scenario pathway analysis?
The lab
We successfully built two iterations of a proof-of-concept scenario pathway generator tool. Using a Vector Database, ChatGPT LLM and the framework Dash, the tool generates structured output in response to a prompt.
We worked with EIU to agree a problem statement, and designed two scenarios to test the tool’s output against, which were: how the Ukraine War will end, and the effects of a Trump Tariff regime.
Variables between the two iterations included changing the input data to include EIU articles, changing the underlying prompts, specific context data & creativity. We could then explore whether they impacted the quality of output, and in turn added more value to EIU users.
The result
- The second proof-of-concept met significantly more user needs, including the need for accurate analysis, comprehendible results & quickly generating different scenarios.
- Effectively handled ‘Wildcard’ scenarios, in areas of less understanding, with less existing material to draw from.
- Produced creative results, which helped analysts in considering a wider range of scenarios, ultimately improving the richness of their analysis.
- Saved large amounts of time & effort for EIU analysts in carrying out this part of their role.
- Formed a basis (/initial solution) for EIU to carry out further testing, and iteratively improve on going forward.
2 POC's
Demonstrating significant improvement
10
EIU Analysts further testing our build