Abstract
Intelligence agencies today have to collect and analyse intelligence on numerous individuals, state and non-state actors in an environment of many complex hybrid threats and overlapping interests. Additionally, there is a glut of data from several sources that need to be processed quickly and accurately. Artificial Intelligence (AI) presents a viable way to maximise the value of the All-Source intelligence products. Despite all the promise AI holds for the Intelligence Community, the technology is far from perfect.
By Kwei Quaye-Foli
Introduction
In a purportedly leaked document from a U.S. National Security Agency (NSA) analyst in 2011, the analyst (hereafter Agent X) paints a vivid picture of how too much data can lead to Analysis Paralysis. Agent X wrote, ‘you are standing in a shopping aisle trying to decide between jam, jelly or fruit spread, which size, sugar-free or not, generic or Smucker’s. It can be paralyzing’ [1]. In 2013, this analogy would come to highlight the challenges the NSA faced in making sense of the data it gathered under its “Collect It All” mass-surveillance strategy of capturing, archiving and analysing electronic communications worldwide [2]. Almost a decade later, Artificial Intelligence (AI) may just provide the solution for overcoming the practical and technical limitations of mass surveillance Agent X was concerned with.
History of Technology & Intelligence
Innovations in computing and communications technologies over the last century have doubtlessly altered how we pursue our business, academic and leisurely endeavours. These advancements have also changed the way countries compete with each other. Intelligence and defence organisations seldom take long to find creative ways to incorporate the latest technologies into their operations and activities. This trend is just as evident with AI as it was with the radio, encryption, satellites and Unmanned Aerial Vehicles (UAVs).
The United Kingdom’s Office for Artificial Intelligence defines AI as ‘the use of digital technology to create systems capable of performing tasks commonly thought to require intelligence. It (AI) generally involves machines using statistics to find patterns in large amounts of data’ [3]. In terms of security and intelligence, AI enhances and facilitates tasks like facial and other biometric recognition, language translation, sentiment analysis, encryption and decryption among many others.
According to the U.S. National Initiative for Cybersecurity Careers & Studies (NICSS), All-Source Analysis ‘analyzes threat information from multiple sources, disciplines and agencies across the intelligence community; synthesizes and places intelligence information into context and draws insights about the possible implications’ [4]. Simply put, All-Source Analysis feeds off the output from multiple intelligence branches or channels like Human Intelligence (HUMINT), Signals Intelligence (SIGINT), Geospatial Intelligence (GEOINT), Open-Source Intelligent (OSINT), and Imagery Intelligence (IMINT), among many others [5].
What does all this Mean in Practical Terms?
In more practical and relatable terms, the Central Intelligence Agency (CIA), the Defense Intelligence Agency (DIA) and the Bureau of Intelligence & Research (INR) are members of the United States’ Intelligence community (IC) with All-Source Analysis capabilities. On the other hand, the National Security Agency (NSA) and the National Geospatial Intelligence Agency (NGA) respectively specialise in SIGINT and GEOINT, making both agencies single-source intelligence (Single-INT) organisations [6]. The contemporary threat landscape is scattered with wide-ranging, complex and hybrid threats from individuals, non-state actors and governments each with increasing capabilities and overlapping interests. Therefore, an intelligence agency’s ability to collect, analyse and disseminate intel about a large number of actors and their activities in a fast, accurate and efficient manner is vital to that agency’s customers. Artificial Intelligence makes such capabilities possible.
AI and the Intelligence Community
Considering the challenges a Single-Int agency like the NSA faced with analysing all the data it collected in 2013, it does not require much imagination to picture the exponentially more complex problem All-Source Analysis agencies like the CIA had to contend with during that period. It is especially remarkable that the U.S. faced these problems, a country with one of, if not the most sophisticated and well-resourced intelligence and national security enterprises in the world. Perhaps it is rather unsurprising that this well-funded and oiled intelligence machinery would inevitably find the solution to the problem by leveraging the significant innovations in AI, computing power, digital storage and networked systems.
The United States’ AIM Strategy (Augmenting Intelligence using Machines) exploits AI to reduce the number of steps and time spent between data collection and analysis, by the IC, to the dissemination of the intelligence product to the consumers. The former U.S. Director of National Intelligence Dan Coats explained, ‘the pace at which data are generated and collected is increasing exponentially and the Intelligence Community (IC) workforce available to analyze and interpret All-Source, cross-domain data is not’ [7]. Coats went on to explain that ‘leveraging Artificial Intelligence, automation and augmenting technologies will advance and enhance the IC’s ability to provide needed data interpretations to the decision makers’” [8]. Other countries like the United Kingdom, China and Israel have similar AI-based programmes operated by their defence and intelligence services.
The timing of this AI revolution could not be more serendipitous for the IC. The explosion of analytical capability that AI offers the IC exploits the boom in data quantity, availability and access. According to U.S. Air Force Director of Intelligence (Warfighter Support) Lt. Gen. N.T. Shanahan, ‘when it comes to intelligence, surveillance and reconnaissance, we (the U.S.) have more platforms and sensors than any other time in Department of Defense History’ [9].
Governments and the private sector players have invested heavily in developing AI-based solutions for the security and intelligence industry due to the wide range of possible applications. For example, the CIA’s venture capital arm In-Q-Tel has invested in AI-centric companies like Forge A.I. [10] & Cylance [11].
Conclusion
All the wonders of AI notwithstanding, the technology still and will most likely continue to have inherent and systematic limitations for the near future. For instance, the output from AI processes is as good as the quality of the algorithms and training data used to develop the AI models. The development of AI technology is still undemocratic and affected by the same systemic biases present in our societies. Additionally, there are several ethical issues around what kind of AI can or should be used and to what extent. This debate is perhaps most relevant with the AI-based decision systems of Lethal Autonomous Weapons Systems (LAWS) [12].
As AI becomes more pervasive and an integral part of intelligence activities, intelligence agencies will find even more creative ways of using this technology to enhance their capabilities to maximise the value of their intelligence products. This phenomenon will be even more evident within All-Source analysis. AI may eventually become the magic bullet for intelligence agencies to subdue the hydra that is the massive stream of data flowing from many different sources. However, this technology still has some way to go.
Sources
Peter Maas, “Inside NSA, Officials Privately Criticize “Collect It All” Surveillance”. Accessed on 17th July 2021, https://theintercept.com/2015/05/28/nsa-officials-privately-criticize-collect-it-all-surveillance/
Glenn Greenwald, The Crux of the NSA Story in One Phrase: 'Collect It All'. Accessed on 19th July 2021, https://www.theguardian.com/commentisfree/2013/jul/15/crux-nsa-collect-it-all
A Guide to Using Artificial Intelligence in the Public Sector, GOV.UK, Accessed on 7th December 2020, https://www.gov.uk/government/publications/understanding-artificial-intelligence/a-guide-to-using-artificial-intelligence-in-the-public-sector
All-Source Analysis, National Initiative for Cybersecurity Careers and Studies, Accessed on 7th December 2020, https://niccs.cisa.gov/workforce-development/cyber-security-workforce-framework/all-source-analysis.
Thomas Fingar, “A Guide to All-Source Analysis”, the Intelligencer: Journal of U.S. Intelligence Studies, 19(1), 2012: pp 63-66
Mark M. Lowenthal, Intelligence: From Secrets to Policy, Fourth Edition, (Washington DC, CQ Press, 2003)
U.S. Office of the Director of National Intelligence (ODNI) Strategy Document, the AIM Initiative: A Strategy for Augmenting Intelligence Using Machines, Foreword by Dan Coats, 2019
IBID
Artificial Intelligence to Sort through ISR Data Glut, National Defence, Accessed on 28th November 2020, https://www.nationaldefensemagazine.org/articles/2018/1/16/artificial-intelligence-to--sort-through-isr-data-glut
Inside In-Q-Tel’s Latest Tech Scouting Find, Washington Technology, Accessed on 7th December 2020, https://washingtontechnology.com/articles/2019/06/05/inqtel-forge-ai-investment.aspx
14 Cutting Edge Firms Funded by the CIA, Business Insider, Accessed on 5th December 2020, https://www.businessinsider.com/companies-funded-by-cia-2016-9?r=US&IR=T
Satoru Mori, “US Technological Competition with China: The Military, Industrial and Digital Network Dimensions,” Asia-Pacific Review 26 (1), 2019: pp 77–120
