Using AI in an Intelligence Context Future Scenario Workshop

This workshop report presents a summary of observations and discussions related to an event on the use of AI in intelligence by 2040, held in February 2024. The future scenario workshop was held to examine the use and implications of AI in a hypothetical 2040. It was based on a hypothetical scenario in which the relationship between a technically advanced economic and military power called Roland and the neighbouring island state Islay deteriorates. It unfolded in three parts:

1. The build-up of military exercises by Roland near Islay and the potential for invasion.
2. Roland’s full-scale invasion of Islay.
3. Roland establishing civilian authority over Islay.

The workshop was structured around the scenario phases. The following are key observations that participants made in the context of each scenario phase but have relevance to the broader question of what AI use in an intelligence context will look like by 2040 and what implications and questions follow.

Part 1: Pre-Invasion

• The adversary’s widespread AI integration in military structures likely constitutes a strategic advantage, increasing accuracy and efficiency, especially for data analysis.
• However, AI dependency and integration also constitutes an additional risk where access can be achieved to the adversary’s systems and data, for example for information on the adversary’s movements of troops and equipment.
• The quality, not just the quantity, of data is decisive for intelligence purposes. Authenticating information is therefore a key capability.
• Communication and assessment of information is likely challenging in this scenario, including the communication of accurate information and analysing available information at fast pace.
• It is unclear how the UK achieves deterrence in this context or what deterrence looks like in a zero-trust, AI-fuelled environment. This relates to both the communication of information and the sharing of capabilities for deterrence purposes.

Part 2: Full-Scale Invasion

• AI can be used to design efficient evacuations, for example to identify British individuals and their next of kin, or to design and optimise exit plans.
• UK support to allies includes sharing intelligence capabilities and assisting integration and coordination of datasets.
• To be prepared for a full-scale invasion scenario, the UK government needs to invest in and support the development of key enablers, such as computing power, communication infrastructure and critical technologies.
• The UK government also needs more credible technology leadership to implement such aims.
• Procurement needs to become more agile to enable shorter innovation cycles.

Part 3: Civilian Authority Established

• Information operations may be less effective in a context of information chaos.
• There is a significant need for authentication methods.
• It is important that both AI systems and humans are trained in cultural comprehension. At the same time, AI can help to understand culture, history and their lessons.

Participants further assessed wider implications, such as the need for the UK to expand AI training for all sectors and skills levels, and identified further research questions based on these observations.