Call for Abstracts

Your research can help solve mission-oriented challenges faced by the intelligence community.

Each summer, the Laboratory for Analytic Sciences (LAS) kicks off the annual planning cycle by releasing a call for abstracts. Academic and industry researchers are invited to submit one or more research project ideas. The call for abstracts is based on continuing LAS research interests and discussions with U.S. Intelligence Community partners about mission needs. The call for abstracts provides examples of technical areas and applications of interest for LAS.

LAS is a mission-oriented academic-industry-government research collaboration that works at the intersection of technology and tradecraft. We apply unclassified examples from academic and commercial partners to intelligence community goals. Collaborators selected for project funding will work with LAS staff to deliver mission-relevant solutions to better support today’s intelligence analysts.

Technical Areas

Operationalizing AI/ML

Interests related to understanding and evaluating the capabilities and behavior of artificial intelligence/machine learning (AI/ML) technologies, to facilitate their appropriate integration into operational intelligence environments.

Human-Machine Teaming

Interests related to improving how analysts can partner effectively with automation, particularly through the exploration of novel designs for user experiences that integrate state-of-the-art AI and ML capabilities.

Content Triage

Interests related to the scalable extraction and summarization of text, image, speech, audio and video content, in order to facilitate its discovery and use in intelligence analysis workflows.

Application Areas of Interest

Video Sensemaking

Video sensemaking helps analysts understand vast amounts of video data by making video content searchable, summarizing it, and alerting users to videos of interest, all while incorporating user feedback to improve the systems.

In this video, LAS staff outline the need for researchers who can make videos searchable, improve the design of user interfaces, and summarize hours of video data.

Audio Sensemaking

Audio Sensemaking involves the methods and technologies employed by language analysts to extract valuable intelligence from intricate, real-world foreign language audio. This process addresses significant challenges such as unwanted background noise, a multitude of speaking styles, subtle cultural nuances, and the sheer volume of data, all while incorporating user feedback to improve the systems.

In this video, LAS staff explain the value of extracting information and understanding from audio source data done by language analysts.

AI-Enabled Workflows

Research on AI-Enabled Workflows examines efforts to build trust and collaboration between humans and AI by addressing the common pitfalls of over- or under-reliance on automated tools through: enhancing credibility, strengthening approaches to evaluation, and transparently communicating performance and uncertainty.

In this video, LAS staff share examples of prior work and discuss the main components of AI-enabled workflows: enhancing sourcing, faithfulness, and attribution; evaluating explainable AI and language analysis; and communicating AI performance, uncertainty, and limitations.

AI Benchmarking

AI Benchmarking involves testing and comparing AI solutions using standardized methods to evaluate their performance, capabilities, and reliability, ultimately informing decisions about which AI models to use in various intelligence analysis applications.

In this video, LAS staff explain the lab’s interest in efforts to develop novel benchmarks that focus on use cases aligned with mission; ways to better understand the safety, security, and total cost of an AI system; and projects around test and evaluation of agentic and other forms of AI integrated systems that would incorporate additional tools to provide capabilities beyond just the AI model itself.

Edge AI/ML

Edge AI/ML processes data directly on devices or near the source to overcome limitations of cloud-based AI, especially in places with limited connectivity. Edge AI/ML improves real-time triage and analysis, and enables users to customize data filtering for specific insight.

In this video, a fictitious scene demonstrates how an edge device with its onboard AI can turn a chaotic and overwhelming moment into a manageable and efficient emergency response, ensuring that the most critical patients receive the life-saving attention they need first. LAS staff explain how this technology can also be used to triage intelligence information.

Agentic AI

Agentic AI research involves developing AI systems that can make decisions, perform tasks, and integrate with diverse information systems with minimal human input, aiming to enhance intelligence analysis efficiency while ensuring strict adherence to accuracy, objectivity, and compliance standards.

In this video, LAS staff outline the need for researchers who can contribute to designing and building trustworthy, autonomous, multimodal, and explainable AI agents for use in mission-critical environments.

Read the Call for Abstracts

Submit an Abstract

FAQs

Jul 07

Monday