Search

Begin New Search Topic Artificial Intelligence Political Geography Global Focus

51. Biosecurity in the Age of AI: What’s the Risk?

Author:
Janet Egan and Eric Rosenbach
Publication Date:
11-2023
Content Type:
Policy Brief
Institution:
Belfer Center for Science and International Affairs, Harvard University
Abstract:
“The biggest issue with AI is actually going to be … its use in biological conflict,” according to former Google CEO, Eric Schmidt.[1] And he’s not the only AI expert worried. In his testimony before the Senate Judiciary Committee Subcommittee on Privacy, Technology, and the Law, the CEO of Anthropic, Dario Amodei, warned that in just two to three years, AI has the potential to “greatly widen the range of actors with the technical capability to conduct a large-scale biological attack.”[2] OpenAI’s Sam Altman has called for regulation on AI models “that could help create novel biological agents.”[3]President Biden’s recent Executive Order on Safe, Secure and Trustworthy Development and Use of Artificial Intelligence explicitly tasks relevant agencies with assessing the ways in which AI can increase, and potentially help mitigate, biosecurity risks.[4]But what exactly has experts and officials so worried?
Topic:
Defense Policy, Science and Technology, Biosecurity, and Artificial Intelligence
Political Geography:
Global Focus

52. Generative AI and the Future of Work: A Reappraisal

Author:
Carl Benedikt Frey and Michael Osborne
Publication Date:
09-2023
Content Type:
Journal Article
Journal:
Brown Journal of World Affairs
Institution:
Brown Journal of World Affairs
Abstract:
Like now, the early 2010s was a time of excitement about new technological wonders. The world had bounced back from the worst financial crisis since the Great Depression and, despite residual financial malaise, America seemed to be on the cusp of a productivity boom.1 Artificial intelligence (AI), long regarded as an academic backwater, was finally bearing fruit. In 2011, a question-answering computer system known as IBM Watson beat the world champion in Jeopardy. Machine learning—a subfield of AI—further expanded the possibilities of what computers could do; in 2012, a machine learning team from Google trained a deep neural network that, without ever being told what a cat was, proved independently capable of recognizing cat videos on YouTube. Gone were the days when a computer programmer needed to write explicit rules to guide the actions of a machine in every contingency. Computers could now infer rules by tapping into the data trails left behind by increasing numbers of humans online. This phenomenon was most clearly visible with the development of autonomous vehicles. No programmer could be expected to foresee every situation a human driver might encounter in city traffic, and even less to capture these in a sequence of “If-Then-Do” commands. Instead, autonomous vehicles made progress by collecting vast amounts of data on drivers’ actions in city traffic to predict what humans would in any given situation.
Topic:
Science and Technology, Artificial Intelligence, Future of Work, and Generative Models
Political Geography:
Global Focus

53. Table for Twelve: Dinner with the Foremost Experts on AI

Author:
Mike Sexton, Fredrick Hernandez, and Jim Kessler
Publication Date:
12-2023
Content Type:
Special Report
Institution:
Third Way
Abstract:
We’re all trying to get smarter on AI. Well, suppose you could have dinner with some of the smartest experts on AI and get their take? Pull up a chair. We chose 11 AI thinkers—AI creators, defense experts, economists, academicians, tech capitalists, and sociologists—and condensed their viewpoints into 500 words with a link to go deeper. Some are true AI believers, others are deep skeptics, many fall in between. Why 11? So you could be number 12. Would you like red or white with your bots? This is the first product in a multi-year series to help educate decision-makers on the policy choices confronting Congress and the Administration on next generation Artificial Intelligence. Each summary represents our interpretation of their outlook based on published interviews and writings and have not been approved or edited by each individual.
Topic:
Defense Policy, National Security, Science and Technology, and Artificial Intelligence
Political Geography:
Global Focus

54. Mind the Gap: The Challenges of Assurance for Artificial Intelligence

Author:
Isaac Gazendam and Philip Dawson
Publication Date:
11-2023
Content Type:
Working Paper
Institution:
Center for International Security and Cooperation (CISAC), Stanford University
Abstract:
Artificial intelligence (AI) capabilities have progressed rapidly since defeating human players at board games. From voice assistants to chatbots, image filters to multi-modal content generation, AI is accelerating individual, commercial, and scientific purposes. The development of increasingly advanced AI systems, each with unique strengths and vulnerabilities, has also made assessing and managing AI risk a daunting prospect. Controls do exist. A variety of stakeholders have made strides in proposing tools and methods to govern AI systems. A common approach applied across these attempts is based on assurance, a collection of approaches and techniques that aim to measure the quality, reliability, and trustworthiness of AI products, services, organizations, or professionals. To some extent, governments are relying on assurance-based approaches to govern AI – from assessments and audit to certification – regardless of whether they have plans to develop AI regulations. Yet the AI assurance ecosystem remains nascent. Frameworks for AI assessments and audits are still evolving, as are supporting standards and accreditation criteria for third-party auditors, reviewers and evaluators. In some cases, it is now possible to apply certain assurance mechanisms to the governance of narrower applications of AI and related organizational processes. However, assurance-based approaches to more advanced systems based on large language models (LLMs) or generative models remain in their infancy.
Topic:
Science and Technology, Governance, Regulation, Artificial Intelligence, and Large Language Models (LLMs)
Political Geography:
Global Focus

55. Peace of Mind: Cognitive Warfare and the Governance of Subversion in the 21st Century

Author:
Jean-Marc Rickli, Frederico Mantellassi, and Gwyn Glasser
Publication Date:
08-2023
Content Type:
Policy Brief
Institution:
The Geneva Centre for Security Policy
Abstract:
The 21st century is marked by unprecedented and exponential technological advances. These advances are changing the ways in which states influence, coerce, subvert and wage war, and have democratised access to global means of influence for non-state actors – and even individuals. This has accelerated the prevalence of so-called hybrid forms of warfare. An “age of unpeace” appears to have established itself, with a pervasive sense of both non-peace and non-war. Influence campaigns utilising digitally spread disinformation and instrumentalising social media technologies have proliferated, enhancing ways to target the human mind. Current and future developments in artificial intelligence (AI), cognitive sciences, neurotechnologies, and other related fields will further increase the risks of mass manipulation and lead to the possibility of the militarisation of the mind as the battlefield of the future. The emergence of the cognitive domain as the sixth domain of warfare will lead to the increased conduct of cognitive warfare, which is likely to raise the profile and efficiency of non-kinetic means of subversion over kinetic means of coercion, while benefitting from a lack of international governance. If international governance remains static while these tools advance at great speeds, the international system will lack the frameworks, tools, and understanding needed to govern the means of 21st century subversion. This policy brief explores the emergence of cognitive warfare – which aims at controlling what and how an adversary thinks – and the rise on the international stage of non-kinetic means of subversion enabled by emerging technologies. It proposes the establishment of governance frameworks to regulate the use of emerging technologies for purposes of cognitive warfare with the ultimate aim of subversion. It promotes the concept of “subversion control” in order to prevent the militarisation of the mind. It further recommends the regulation of enabling emerging technologies, such as neurotechnologies and AI, while promoting the enhancement of “societal resilience”, especially in democracies, whose open and digitalised information environments make them structurally more vulnerable to the practices of cognitive warfare. Lastly, the brief seeks to promote more research into the concept of cognitive warfare itself, w
Topic:
Science and Technology, Artificial Intelligence, and Cognitive Warfare
Political Geography:
Global Focus

56. AI-Related Risk: The Merits of an ESG-Based Approach to Oversight

Author:
Mardi Witzel and Niraj Bhargava
Publication Date:
08-2023
Content Type:
Working Paper
Institution:
Centre for International Governance Innovation (CIGI)
Abstract:
Artificial intelligence (AI) technology has been found to generate value for many firms; however, it also has unintended and undesirable consequences. The reality of AI-related risk has led to the development of AI governance frameworks and calls for greater oversight of the use of AI. The merits of an ESG (environmental, social, governance)-based approach to oversight of AI-related risk are considered in this paper, with a focus on the current trajectory of international sustainability standards development. Despite their differences, AI governance and ESG reporting both seek to address risk in the broadest sense, with proactive and transparent approaches to its management and mitigation. A preliminary set of recommendations for incorporating material AI-related risk into ESG reporting, covering both general or context-setting disclosures and industry-specific disclosures, is provided.
Topic:
Science and Technology, Governance, and Artificial Intelligence
Political Geography:
Global Focus

57. Artificial Intelligence and Arms Control

Author:
Paul Scharre and Megan Lamberth
Publication Date:
10-2022
Content Type:
Special Report
Institution:
Center for a New American Security (CNAS)
Abstract:
Advances in artificial intelligence (AI) pose immense opportunity for militaries around the world. With this rising potential for AI-enabled military systems, some activists are sounding the alarm, calling for restrictions or outright bans on some AI-enabled weapon systems.1 Conversely, skeptics of AI arms control argue that as a general-purpose technology developed in the civilian context, AI will be exceptionally hard to control.2 AI is an enabling technology with countless nonmilitary applications; this factor differentiates it from many other military technologies, such as landmines or missiles.3 Because of its widespread availability, an absolute ban on all military applications of AI is likely infeasible. There is, however, a potential for prohibiting or regulating specific use cases. The international community has, at times, banned or regulated weapons with varying degrees of success. In some cases, such as the ban on permanently blinding lasers, arms control has worked remarkably well to date. In other cases, however, such as attempted limits on unrestricted submarine warfare or aerial bombardment of cities, states failed to achieve lasting restraint in war. States’ motivations for controlling or regulating weapons vary. States may seek to limit the diffusion of a weapon that is particularly disruptive to political or social stability, contributes to excessive civilian casualties, or causes inhumane injury to combatants. This paper examines the potential for arms control for military applications of AI by exploring historical cases of attempted arms control, analyzing both successes and failures. The first part of the paper explores existing academic literature related to why some arms control measures succeed while others fail. The paper then proposes several criteria that influence the success of arms control.4 Finally, it analyzes the potential for AI arms control and suggests next steps for policymakers. Detailed historical cases of attempted arms control—from ancient prohibitions to modern agreements—can be found in appendix A in the pdf available for download. For a summary table of historical attempts at arms control, see appendix B. History teaches us that policymakers, scholars, and members of civil society can take concrete steps today to improve the chances of successful AI arms control in the future. These include taking policy actions to shape the way the technology evolves and increasing dialogue at all levels to better understand how AI applications may be used in warfare. Any AI arms control will be challenging. There may be cases, however, where arms control is possible under the right conditions, and small steps today could help lay the groundwork for future successes.
Topic:
Arms Control and Proliferation, Artificial Intelligence, and Military
Political Geography:
Global Focus

58. AI Startups and the Fight Against Mis/Disinformation Online: An Update

Author:
Anya Schiffrin, Hiba Beg, Pablo Eyzaguirre, Zachey Kliger, Tianyu Mao, Aditi Rukhaiyar, Kristen Saldarini, and Ojani Walthrust
Publication Date:
07-2022
Content Type:
Working Paper
Institution:
German Marshall Fund of the United States (GMFUS)
Abstract:
The events following Russia’s invasion of Ukraine have shown again the power of online mis/disinformation. As it continues to grow and spread, there are new and continuing attempts to address this problem. These include supply-side and demand-side fixes (including media-literacy programs, fact-checking and, in Europe, new regulations) but few of these have scaled. This paper looks at one kind of supply-side attempt to tackle the prevalence of online mis/disinformation: the market for tech-based solutions that use some form of artificial intelligence (AI) machine/deep learning for content moderation, media integrity, and verification. This paper presents the findings of interviews of 20 niche firms that use AI to identify online mis/disinformation, many of which were previously surveyed for a 2019 paper on the role of AI startups in the fight against disinformation. These companies did not release their revenue figures but it seems that the market for their services is smaller than many entrepreneurs had originally hoped, and that Google and Facebook are not relying on such firms for help in identifying online mis/disinformation. The cost of the services provided by these startups and the desire to keep things in-house and protect their activities from outside scrutiny are part of why the tech giants do not rely on small startups for help with screening online mis/disinformation. This may be why funding for these startups does not seem to have grown significantly and so more than half of them are now focusing on the business-to-business market, selling mis/disinformation mitigation services to, for example, insurance companies, large public entities, and governments, among others. There also appears to be a limited market for business-to-consumer solutions for detecting mis/disinformation. However, continuing advances in AI as well as forthcoming regulation by the European Union and the United Kingdom will continue to spur innovation, which may stimulate demand. University initiatives, academics, journalism organizations, and cybersecurity experts are all also trying to come up with ways to identify and control the spread of mis/disinformation. Ultimately, despite its advances, technology alone will not solve the online mis/disinformation problem. Giant social media platforms have few financial incentives to crack down on this—quite the opposite, in fact. To push social media platforms to act against online mis/disinformation and illegal speech, regulation must deftly address the issue while preserving freedom of expression. There is a further problem in the form of the political polarization that has intensified in the United States and other parts of the world. Fixing this is likely to be well beyond the role of business and technologists.
Topic:
Science and Technology, Cybersecurity, Democracy, Artificial Intelligence, Disinformation, Misinformation, and Startup
Political Geography:
Global Focus

59. AI Audit-Washing and Accountability

Author:
Ellen P. Goodman
Publication Date:
11-2022
Content Type:
Working Paper
Institution:
German Marshall Fund of the United States (GMFUS)
Abstract:
We are still some distance from a worldwide robot takeover, but artificial intelligence (AI)—the training of computer systems with large data sets to make decisions and solve problems—is revolutionizing the way governments and societies function. AI has enormous potential: accelerating innovation, unlocking new value from data, and increasing productivity by freeing us from mundane tasks. AI can draw new inferences from health data to foster breakthroughs in cancer screening or improve climate modeling and early-warning systems for extreme weather or emergency situations. As we seek solutions to today’s vexing problems—climate disruption, social inequality, health crises—AI will be central. Its centrality requires that stakeholders exercise greater governance over AI and hold AI systems accountable for their potential harms, including discriminatory impact, opacity, error, insecurity, privacy violations, and disempowerment. In this context, calls for audits to assess the impact of algorithmic decision-making systems and expose and mitigate related harms are proliferating1 , accompanied by the rise of an algorithmic auditing industry and legal codification. These are welcome developments. Audits can provide a flexible co-regulatory solution, allowing necessary innovation in AI while increasing transparency and accountability. AI is a crucial element of the growing tech competition between authoritarian and democratic states—and ensuring that AI is accountable and trusted is a key part of ensuring democratic advantage. Clear standards for trustworthy AI will help the United States remain a center of innovation and shape technology to democratic values. The “algorithmic audit” nevertheless remains ill-defined and inexact, whether concerning social media platforms or AI systems generally. The risk is significant that inadequate audits will obscure problems with algorithmic systems and create a permission structure around poorly designed or implemented AI. A poorly designed or executed audit is at best meaningless and at worst even excuses harms that the audits claim to mitigate. Inadequate audits or those without clear standards provide false assurance of compliance with norms and laws, “audit-washing” problematic or illegal practices. Like green-washing and ethics-washing before, the audited entity can claim credit without doing the work. To address these risks, this paper identifies the core questions that need answering to make algorithmic audits a reliable AI accountability mechanism. The “who” of audits includes the person or organization conducting the audit, with clearly defined qualifications, conditions for data access, and guardrails for internal audits. The “what” includes the type and scope of audit, including its position within a larger sociotechnical system. The “why” covers audit objectives, whether narrow legal standards or broader ethical goals, essential for audit comparison. Finally, the “how” includes a clear articulation of audit standards, an important baseline for the development of audit certification mechanisms and to guard against audit-washing. Algorithmic audits have the potential to transform the way technology works in the 21st century, much as financial audits transformed the way businesses operated in the 20th century. They will take different forms, either within a sector or across sectors, especially for systems which pose the highest risk. But as algorithmic audits are encoded into law or adopted voluntarily as part of corporate social responsibility, the audit industry must arrive at shared understandings and expectations of audit goals and procedures. This paper provides such an outline so that truly meaningful algorithmic audits can take their deserved place in AI governance frameworks.
Topic:
Science and Technology, Cybersecurity, Democracy, Regulation, Accountability, and Artificial Intelligence
Political Geography:
Global Focus

60. A gender perspective on artificial intelligence and jobs: The vicious cycle of digital inequality

Author:
Estrella Gomez-Herrera and Sabine Köszegi
Publication Date:
08-2022
Content Type:
Working Paper
Institution:
Bruegel
Abstract:
The worldwide artificial intelligence market is expected to increase enormously in the next few years. Because of AI’s immense potential, virtually all industries will be affected by the implementation of AI systems, resulting in the digitalisation and automation of work processes. This will cause disruptive shifts in labour markets, in terms of the number and profiles of jobs in industries as well as worker skill requirements. We take a gender perspective and analyse how gender stereotypes and gendered work segregation on the one hand, and digitalisation and automation (as a consequence of AI implementation) on the other hand, are entangled and result in a vicious cycle of digital gender inequality. We provide insights into the gender-specific impact of AI technologies, which is relevant for the mitigation of the potential risk of the creation of social inequality and exclusion. We show that existing empirical evidence already indicates that AI will not increase gender equality but will somewhat further exacerbate the gender inequality in labour markets, ranging from further horizontal and vertical occupational gender segregation to an increase in the gender pay gap. We summarise policy guidance and measures to decrease gender inequality in the future.
Topic:
Inequality, Digital Economy, Artificial Intelligence, Gender, and Work
Political Geography:
Global Focus