Generative Artificial Intelligence Threats to Information Integrity and Potential Policy Responses

Published Apr 16, 2024

by Todd C. Helmus, Bilva Chandra

更多

This paper highlights the ecosystem of generative artificial intelligence (AI) threats to information integrity and democracy and the potential policy responses to mitigate the nexus of those evolving threats. The authors focus on the information environment and how generative AI—such as large language models or AI-generated images and audio—is able to accelerate existing harms on the internet and beyond. The policy options that could address these complex problems are vast, varying from much-needed social media reforms to using federal agencies to create sweeping standards for AI-generated content. The authors provide an overview of the risks that generative AI presents to democratic systems, as well as tangible and detailed whole-of-government and societal solutions to mitigate these risks at scale.

收起

A Framework for Assessing the Costs and Benefits of Digital Engineering

A Systems Approach

Published Mar 13, 2024

更多

by N. Peter Whitehead, Thomas Light, Adrian Luna, Jim Mignano

RAND researchers worked to understand the costs and benefits of digital engineering in the U.S. Department of Defense (DoD) and develop a decision support framework for digital engineering activities in weapon system programs. To prepare, the authors reviewed the literature and interviewed stakeholders to understand the current state of digital engineering practice and prior efforts to assess the costs and benefits of digital engineering and model-based systems engineering. They then developed decision support frameworks incorporating (1) established DoD cost-benefit analysis approaches and (2) established systems engineering decision methodologies. Along the way, the authors noted critical issues with rigor and risks in the practice of DoD digital engineering and added that aspect to the study.

This research suggests that cost-benefit decision support for digital engineering is possible at any stage of a weapon system program life cycle if program data have been collected accordingly or if goal-based systems engineering principles are leveraged. Calculating definitive costs and benefits of digital engineering is imperfect because no analyst will have access to an identical weapon system program developed without digital engineering — the counterfactual scenario.

Key Findings

Absent a consensus about what constitutes digital engineering, scholars and practitioners lack a common foundation on which to conduct analysis of costs and benefits.

Tracking digital engineering investments and distinguishing them from other costs (e.g., systems engineering, multipurpose IT investments) does not appear to be a standard practice.

Basing program decisions for digital engineering resource allocation on a systems thinking approach might produce optimal program outcomes.

Much effort has been devoted to making the case for digital engineering by projecting possible benefits without, at the same time, evaluating costs.

Some authors — and some services — are looking to a digital engineering maturity model as a path to improving outcomes without any cost data, causality map, or risk analysis to support that conclusion.

Recommendations

Develop consistency and goal-focused consensus in what digital engineering is (and is not).

Establish clear digital engineering program goals and system boundaries to eliminate pretending and to eliminate applying software principles and standards to complex systems.

Collect program goal-derived data to support assessment of digital engineering costs and benefits.

Eliminate aspirational and general guidance (top-down direction) for implementing digital engineering in such terms as to the maximum extent practical.

Promote an environment in which digital engineering is a useful tool among many others to be used in good systems engineering practice and weapon system project management.

Objectively analyze enterprise digital engineering tools in their operational context.

Promote digital engineering practice and culture where learning from mistakes is as important as achieving success.

Establish policy whereby understanding and mitigating risks is a key facet of digital engineering and MBSE practice.

Establish a roadmap for identifying and achieving absorptive capacity needs across DoD and the respective services in conjunction with the respective digital transformations.

Develop a framework for establishing the hierarchical goals for leveraging data and intellectual property in a weapon system program.

Establish DoD weapon system program life cycle decision and milestone gate criteria for effectively leveraging modeling and simulation based on established systems engineering practice.

Objectively study and develop recommendations for the long-term life cycles of multi-level classification systems, data, and infrastructures as required.

收起

Towards AUKUS Collaboration on Responsible Military Artificial Intelligence

Co-Design and Co-Development of AI Among the United States, the UK and Australia

Published Feb 6, 2024

更多

by Austin Wyatt, James Ryseff, Elisa Yoshiara, Benjamin Boudreaux, Marigold Black, James Black

As the United States and its close allies, such as the United Kingdom (UK) and Australia, face rising threats from China, a key part of their strategy has been to rely on allies and partners to outpace their competitor. However, operating alongside allies and partners also comes with operational challenges and complexities. The authors investigate the challenges and barriers that could inhibit or prevent the co-development of artificial intelligence (AI) between the United States and its closest allies and partners and make recommendations for overcoming those barriers.

Key Findings

The strengths of each partner must be leveraged while preserving the perspective and end-value of the product to junior partners.

The three members of AUKUS and many other countries are still evaluating how AI can be integrated into military applications, as well as how it might be integrated responsibly.

There are many challenges, such as international and domestic regulations, to enabling cooperation between militaries across traditionally stringent barriers.

Recommendations

AUKUS states should prioritize and streamline approaches to ethically and legally responsible use of AI.

AUKUS states should create new multinational research and development centres to continue to develop the capability for collaborative research.

AUKUS states should integrate data and AI into multinational exercises to develop military AI effectively and practically.

AUKUS states should collaboratively develop compatible operational concepts for the use of military AI.

AUKUS states should explore mechanisms for greater innovation and ecosystem integration by allowing for cross-sector collaboration and the free movement of talent between states.

AUKUS states should lessen the burden of data-sharing and intellectual property rights regulations to promote collaboration between states.

收起

Attributing Biological Weapons Use

Strengthening Department of Defense Capabilities to Investigate Deliberate Biological Incidents

Published Feb 6, 2024

更多

by Tracey Rissman, Annette Prieto

The White House has given the U.S. Department of Defense (DoD) a lead role in U.S. efforts to strengthen the United Nations Secretary-General's Mechanism for Investigation of Alleged Use of Chemical and Biological Weapons (UNSGM). In 2022, a White House report highlighted the importance of determining the facts related to the attribution of alleged use of biological weapons (BW), including toxin weapons. That 2022 report aimed to outline the U.S. government's approach to counter the full range of possibly catastrophic biological incidents, whether natural, accidental, or deliberate, and outlines goals and objectives for strengthening the biodefense enterprise. It also identifies priorities and target areas for each mission objective during a biological incident.

In this report, the authors examine issues related to the attribution of BW use and identify areas in which DoD could enhance its capabilities to (1) support U.S. investigative capabilities into alleged uses of biological and toxin weapons and (2) strengthen international efforts, specifically United Nations mechanisms, to hold state and nonstate actors accountable for BW use.

Key Findings

Attribution of BW is important for reasons beyond holding an actor(s) responsible

Developing and maintaining robust capabilities and processes in conducting investigations of BW use can provide a deterrent against their use, improve understanding of adversaries' BW-related capabilities and doctrine, and inform improvements to biodefense capabilities.

Attribution is difficult but essential to support international efforts to counter BW

A range of technical capabilities is needed to first identify that a biological incident occurred and then determine the origin or establish attribution.

Various national and international biosurveillance systems are in place to monitor for biological agents and to report in real time whether a biological incident has occurred.

Laboratory personnel need technologies to analyze physiological, environmental, and other samples sufficiently to identify characteristics that could help identify the biological agent's origin.

Microbial forensics involves using epidemiology and microbiologic technology to characterize biological agents and identify their origins, whether natural or synthetic.

High-quality evidence could help distinguish natural or accidental biological incidents from deliberate ones

Determining the cause of a biological incident is complicated because some biological agents have natural, common sources and can cause natural or accidental outbreaks of disease.

The proper collection, handling, analysis, and reporting of evidence facilitate attribution.

Investigation teams should be trained on proper collection, handling, analysis, and reporting of evidence and should have the necessary equipment to do so.

Recommendations

DoD should enhance and build flexibility and redundancy, as appropriate, into its processes and capabilities to investigate and attribute biological incidents.

DoD should continue to work with allies to build their capabilities to collect information and samples related to biological incidents, especially those that occur in locations where DoD has fewer allies or less access.

DoD should exercise and update processes to conduct sampling-related activities, including collecting samples, handling and transporting samples, communicating and collaborating with other U.S. government departments and agencies, protecting patient information, communicating and collaborating with allies, analyzing samples, and reporting and sharing results of sample analysis.

DoD should improve and exercise procedures to maintain and document chain of custody for samples collected from biological incidents.

DoD should exercise and update processes to collect intelligence information, including interviewing victims, witnesses, and medical personnel.

DoD should enhance efforts to train U.S. personnel and partners on methodologies to collect and analyze evidence during investigations of alleged BW use and on processes to share and report results.

DoD should continue to invest in biotechnology, including microbial forensic technology. Officials should develop, improve, and refine capabilities to model biological incidents.

DoD should consider investing in field-deployable technology or ways of quickly transporting such analytical capabilities to the field to facilitate rapid response to biological incidents.

DoD should maintain transparency about DoD efforts to work with international partners on biosecurity and biodefense efforts.

DoD should continue countering false narratives that it is supporting offensive BW-related work worldwide.

收起

Planning for the Aftermath

Assessing Options for U.S. Strategy Toward Russia After the Ukraine War

Published Feb 9, 2024

更多

by Samuel Charap, Miranda Priebe

U.S. policy choices made at the conclusions of past wars have had enduring consequences. Although there is no end in sight to the Russia-Ukraine war at the time of this writing in late 2023, U.S. policymakers should begin considering postwar Russia strategy now.

The authors of this report address this issue through an alternative futures analysis. Planning for after the war is complicated by significant uncertainty: Neither the conflict's trajectory nor the international climate at the war's end can be predicted with confidence. Given this uncertainty, the implications of U.S. strategic choices must be explored in different contexts. The authors develop two ideal-type postwar worlds, as defined by the outcome of the war and the character of the broader international environment: a less favorable world and a more favorable world. They also propose two ideal-type options for postwar U.S. strategy toward Russia, a hardline approach and a less hardline approach. Then, they consider how each strategy would play out in each world over the course of the decade after the war ends, yielding four alternative futures. They draw on the history of U.S.-Russia relations and the literatures on rivalries, interstate conflict, and alliances to assess the implications of each future for U.S. interests.

Key Findings

Wartime choices could shape the postwar world

Not taking these longer-term factors into consideration could lead to missed opportunities to shape the postwar environment.

The choices that will have to be made in the immediate aftermath of the war can have ripple effects on many long-term U.S. interests

These effects are not straightforward. Policymakers thus need time — long before the war ends — to consider these choices.

The United States may be able to influence the conflict outcome to promote its long-term postwar interests

The United States cannot determine the outcome of the war on its own; its decisions will never have the same impact as those of the two combatants. But Washington does have policy options to try to affect the trajectory of the conflict.

A longer, more violent war would lock in adverse consequences for U.S. interests

For example, a longer war could significantly undermine Ukraine's postwar recovery.

U.S. policy during and after the war can reduce the risk of Russia-Ukraine conflict recurrence

The United States has instruments to increase the likelihood that any peace endures.

Ramping up military pressure on Russia in Europe after the war could pose more risks than benefits

The war has weakened Russia and shown that the North Atlantic Treaty Organization has a strong deterrent against Russian attacks on allies. Further forward deployments and other measures are likely unnecessary to deter opportunistic Russian aggression, but may make war by misperception about U.S. intentions more likely.

收起

The Day After

Postwar U.S. Strategy Toward Russia

Published Feb 9, 2024

更多

by Miranda Priebe, Samuel Charap

收起

Evaluation of Alternative Price and Credit Policies for Spare Parts

Addressing Price Changes and Reliance on Credits

Published Feb 13, 2024

更多

by Ellen M. Pint, Adam C. Resnick, Marygail K. Brauner, Aisha Najera Chesler, Kenneth J. Girardini, Erin N. Leidy, Candice Miller

Under current policies, Army units purchase depot-level reparable parts (DLRs) from the Army Working Capital Fund (AWCF) and receive credits for returning unserviceable DLR carcasses. During fiscal year 2021, Army leaders decided to stop issuing credits to help the AWCF recover from a cash shortage caused by declining sales during the coronavirus disease 2019 pandemic. However, units continued to return DLR carcasses to the supply system even though they no longer had a financial incentive to do so. This raised the question of whether the AWCF's price and credit policies could be modified to reduce reliance on credits and the financial uncertainty associated with price changes on back-ordered repair parts.

U.S. Army Forces Command (FORSCOM) asked the RAND Arroyo Center to evaluate alternative price and credit policies, including single price, under which units would pay the standard price net of the credit for DLRs, and born-on pricing, which would set the price at the time the unit orders the part instead of updating it when the part is issued. In coordination with the sponsor, the research team developed additional courses of action and a set of evaluation criteria, including effects on unit buying power, financial management workload, DLR demands and returns, financial stability of the AWCF, and Army reserve components and other services. Researchers also examined required approvals, changes to enterprise resource planning (ERP) systems, and changes to budgeting and funds distribution.

Key Findings

There was no evidence that units held back returns of DLR carcasses while credits were shut off, suggesting that single price is a feasible alternative.

Absorbing price changes for all transactions that pass through the AWCF would require a small increase in the supply management surcharge but would eliminate 95 percent of FORSCOM's unmatched transactions and 98 percent of the associated costs. However, this policy is less effective for other Army commands where fewer transactions pass through the AWCF, particularly for the Army National Guard and Army Reserve.

Raising the threshold for automatic reconciliation of transactions with price changes up to $750 would cover 90 percent of transactions but would still require financial managers to ensure that prior-year funding is available to pay for them.

Recommendations

Implement single price or exchange pricing (which charges units a delta bill if they fail to return DLR carcasses) to address reliance on credits. Single price would be easier to implement in existing ERP systems, but exchange pricing addresses possible concerns about retaining financial incentives to return DLR carcasses.

In the short term, raise the threshold for automatic processing of unmatched transactions to address price changes; in the longer term, implement ERP changes to absorb price changes for all transactions that pass through the AWCF.

收起

Has Trust in the U.S. Intelligence Community Eroded?

Examining the Relationship Between Policymakers and Intelligence Providers

Published Feb 13, 2024

更多

by Christopher Dictus, Yuliya Shokh, Isabelle Nazha, Marek N. Posard, Richard S. Girven, Sina Beaghley, Anthony Vassalo

Over the past several years, media reports and articles by policy and U.S. Intelligence Community (IC) professionals have suggested that the IC is held in increasingly lower regard by some decisionmakers—and also by some in the general public—and that predictions made by IC professionals have had variable success in influencing decisionmakers. The authors of this report explored whether and to what degree trust in intelligence predictions and national estimates has degraded over time and what internal and external factors might be drivers of any perceived or real changes in the relationship between policymakers and the IC.

The degree of perceived bias in intelligence estimates is highly dependent on the presidential administration in power. Also, policymakers most frequently introduce bias in intelligence assessments from a desire to minimize the appearance of dissent, while the IC tends to introduce bias through self-censorship. The research team observed tensions in the relationship between IC professionals and policymakers: The IC has an incentive to elicit positive feedback from policymakers, and there are limited benefits (with regard to both the careers of intelligence professionals and their agencies' budgets) in receiving negative feedback from policymakers. These tensions could create friction between (1) providing objective information and analysis to policymakers and (2) serving policymakers as customers of intelligence.

Key Findings

The degree of perceived bias in intelligence estimates is highly dependent on the presidential administration in power

Some administrations bias and politicize intelligence less, some more. Moreover, each President has a different relationship with the IC.

The administration in power is the primary driver of both external and internal bias. The President selects and commands those who seek to introduce external bias. However, it is equally important to note that when the IC self-censors, it is doing so because of how the President governs.

The most common reason for bias from policymakers is the desire to minimize the appearance of dissent

Policymakers most commonly seek to introduce bias in intelligence estimates to reduce the appearance of dissent. Administration officials do not want to have to explain to the President, the Congress, or the American people why a part of the executive branch disagrees with a chosen policy.

The most common reason for bias internal to the IC is self-censorship

IC analysts and managers have consciously changed or watered down assessments. This self-censorship was largely driven by a desire to stay relevant or an effort to avoid retaliation from policymakers.

There is an inherent tension in the IC-policymaker relationship

The IC has an incentive to elicit positive feedback. There are no benefits, either to careers or budgets, for negative feedback. This creates friction between the mission of providing policymakers objective information and serving the policymakers as customers.

Recommendations

It is imperative that uniform standards establish what constitutes improper bias and what does not. Although the IC conducts training on the types of bias that can affect analyses, IC analytic standards discuss bias in only the most general way.

To adequately investigate claims of improperly biased intelligence, inspectors general should be insulated from the command structure and should have the ability to institute real and enforceable punishments for those who commit offenses.

It is the responsibility of the President and the Senate to ensure that the right individuals serve at the highest levels of the IC. The language of U.S. law could be strengthened by laying out what kind of expertise is needed to effectively manage the intelligence agencies of the United States.

收起