The Intel Community’s Approach to AI

While public sector strategies on the use of AI still remain rare these days, the Intelligence Community (IC) has moved forward with issuing a framework to incorporate such technologies in strengthening its mission. Released last month, the Augmenting Intelligence using Machines (AIM) Initiative describes the IC’s approach in implementing artificial intelligence, automation and augmentation (AAA) technologies to create a sound digital base and take its agencies to the next generation of technological innovation.  

According to the document, the initiative seeks to “guide the IC to accelerate the adoption of digital and analytics transformation, identify mission use cases, build a coherent data ecosystem, acquire the appropriate AI tools, reshape the workforce, adapt new workflow processes, and change the culture.”

The framework is centered on four objectives, in chronological order, detailing the priorities and technology investments set by AIM:

1. Immediate and ongoing – Digital Foundation, Data, and Science and Technical Intelligence (S&TI): achieve a digital infrastructure ripe for data sharing and accessibility. Enable automated solutions in vetting data and establish standards for evaluation. Dedicate resources for the research of AI, in both its uses and adversaries.
2. Short term – Adopt Commercial and Open Source Narrow AI Solutions: pursue ready opportunities in the commercial sector. Establish an AIM Center to sustain innovative prototypes of AI and ML solutions. Strengthen partnerships with dedicated public sector research institutions to accelerate adoption of AAA technologies.
3. Medium term – Invest in the Gaps (AI Assurance and Multimodal AI): develop AI solutions to process and relay information from various data sources while overcoming traditional challenges and silos.  
4. Long term – Invest in Basic Research Focused on Sense-Making: assist and enhance the work of IC analysts and officers in understanding complex, multisource information.

The guidance provides a summary table for the four objectives:

Source: The AIM Initiative

While the pursuit of AAA, particularly AI, technologies is apparent throughout the objectives, the AIM strategy details other considerations that must be made in using AI and ML including assurance, reliability and antagonistic concerns. In particular, the document specifies that the IC must understand how AI algorithms could fail. 

Contractor Implications

The AIM initiative paves the way for contractors in working with the IC when it comes to AI solutions. In addition to workforce, policy and communication strategies outlined in AIM and the technologies called out in the four objectives, the initiative pointedly details an Industry Partnership Strategy. Recognizing the necessity of private sector collaboration, AIM calls for acquisition friendly agreements such as “analysis-as-a-service” and allowing access to geospatial and other intel-related data for algorithm development. Moreover, contractors are likely to see flexibility and additional spending in AI-related research from the IC and keen observance of data and workforce sharing policies and oversight.