Search

This paper introduces v0.5 of the AI Safety Benchmark, which has been created by the MLCommons AI Safety Working Group. The AI Safety Benchmark has been designed to assess the safety risks of AI systems that use chat-tuned language models. We introduce a principled approach to specifying and constructing the benchmark, which for v0.5 covers only a single use case (an adult chatting to a general-purpose assistant in English), and a limited set of personas (i.e., typical users, malicious users, and vulnerable users). We created a new taxonomy of 13 hazard categories, of which 7 have tests in the v0.5 benchmark. We plan to release version 1.0 of the AI Safety Benchmark by the end of 2024. The v1.0 benchmark will provide meaningful insights into the safety of AI systems. However, the v0.5 benchmark should not be used to assess the safety of AI systems. We have sought to fully document the limitations, flaws, and challenges of v0.5. This release of v0.5 of the AI Safety Benchmark includes (1) a principled approach to specifying and constructing the benchmark, which comprises use cases, types of systems under test (SUTs), language and context, personas, tests, and test items; (2) a taxonomy of 13 hazard categories with definitions and subcategories; (3) tests for seven of the hazard categories, each comprising a unique set of test items, i.e., prompts. There are 43,090 test items in total, which we created with templates; (4) a grading system for AI systems against the benchmark; (5) an openly available platform, and downloadable tool, called ModelBench that can be used to evaluate the safety of AI systems on the benchmark; (6) an example evaluation report which benchmarks the performance of over a dozen openly available chat-tuned language models; (7) a test specification for the benchmark.

Trust is an important factor in people's interactions with AI systems. However, there is a lack of empirical studies examining how real end-users trust or distrust the AI system they interact with. Most research investigates one aspect of trust in lab settings with hypothetical end-users. In this paper, we provide a holistic and nuanced understanding of trust in AI through a qualitative case study of a real-world computer vision application. We report findings from interviews with 20 end-users of a popular, AI-based bird identification app where we inquired about their trust in the app from many angles. We find participants perceived the app as trustworthy and trusted it, but selectively accepted app outputs after engaging in verification behaviors, and decided against app adoption in certain high-stakes scenarios. We also find domain knowledge and context are important factors for trust-related assessment and decision-making. We discuss the implications of our findings and provide recommendations for future research on trust in AI., Comment: FAccT 2023

Despite the proliferation of explainable AI (XAI) methods, little is understood about end-users' explainability needs and behaviors around XAI explanations. To address this gap and contribute to understanding how explainability can support human-AI interaction, we conducted a mixed-methods study with 20 end-users of a real-world AI application, the Merlin bird identification app, and inquired about their XAI needs, uses, and perceptions. We found that participants desire practically useful information that can improve their collaboration with the AI, more so than technical system details. Relatedly, participants intended to use XAI explanations for various purposes beyond understanding the AI's outputs: calibrating trust, improving their task skills, changing their behavior to supply better inputs to the AI, and giving constructive feedback to developers. Finally, among existing XAI approaches, participants preferred part-based explanations that resemble human reasoning and explanations. We discuss the implications of our findings and provide recommendations for future XAI design., Comment: CHI 2023

We investigate the privacy practices of labor organizers in the computing technology industry and explore the changes in these practices as a response to remote work. Our study is situated at the intersection of two pivotal shifts in workplace dynamics: (a) the increase in online workplace communications due to remote work, and (b) the resurgence of the labor movement and an increase in collective action in workplaces -- especially in the tech industry, where this phenomenon has been dubbed the tech worker movement. Through a series of qualitative interviews with 29 tech workers involved in collective action, we investigate how labor organizers assess and mitigate risks to privacy while engaging in these actions. Among the most common risks that organizers experienced are retaliation from their employer, lateral worker conflict, emotional burnout, and the possibility of information about the collective effort leaking to management. Depending on the nature and source of the risk, organizers use a blend of digital security practices and community-based mechanisms. We find that digital security practices are more relevant when the threat comes from management, while community management and moderation are central to protecting organizers from lateral worker conflict. Since labor organizing is a collective rather than individual project, individual privacy and collective privacy are intertwined, sometimes in conflict and often mutually constitutive. Notions of privacy that solely center individuals are often incompatible with the needs of organizers, who noted that safety in numbers could only be achieved when workers presented a united front to management. We conclude with design recommendations that can help create safer, more secure and more private tools to better address the risks that organizers face., Comment: Accepted to CSCW 2022

Scholars and industry practitioners have debated how to best develop interventions for ethical artificial intelligence (AI). Such interventions recommend that companies building and using AI tools change their technical practices, but fail to wrangle with critical questions about the organizational and institutional context in which AI is developed. In this paper, we contribute descriptive research around the life of "AI" as a discursive concept and organizational practice in an understudied sphere--emerging AI startups--and with a focus on extra-organizational pressures faced by entrepreneurs. Leveraging a theoretical lens for how organizations change, we conducted semi-structured interviews with 23 entrepreneurs working at early-stage AI startups. We find that actors within startups both conform to and resist institutional pressures. Our analysis identifies a central tension for AI entrepreneurs: they often valued scientific integrity and methodological rigor; however, influential external stakeholders either lacked the technical knowledge to appreciate entrepreneurs' emphasis on rigor or were more focused on business priorities. As a result, entrepreneurs adopted hyped marketing messages about AI that diverged from their scientific values, but attempted to preserve their legitimacy internally. Institutional pressures and organizational constraints also influenced entrepreneurs' modeling practices and their response to actual or impending regulation. We conclude with a discussion for how such pressures could be used as leverage for effective interventions towards building ethical AI.

Central to a number of scholarly, regulatory, and public conversations about algorithmic accountability is the question of who should have access to documentation that reveals the inner workings, intended function, and anticipated consequences of algorithmic systems, potentially establishing new routes for impacted publics to contest the operations of these systems. Currently, developers largely have a monopoly on information about how their systems actually work and are incentivized to maintain their own ignorance about aspects of how their systems affect the world. Increasingly, legislators, regulators and advocates have turned to assessment documentation in order to address the gap between the public's experience of algorithmic harms and the obligations of developers to document and justify their design decisions. However, issues of standing and expertise currently prevent publics from cohering around shared interests in preventing and redressing algorithmic harms; as we demonstrate with multiple cases, courts often find computational harms non-cognizable and rarely require developers to address material claims of harm. Constructed with a triadic accountability relationship, algorithmic impact assessment regimes could alter this situation by establishing procedural rights around public access to reporting and documentation. Developing a relational approach to accountability, we argue that robust accountability regimes must establish opportunities for publics to cohere around shared experiences and interests, and to contest the outcomes of algorithmic systems that affect their lives. Furthermore, algorithmic accountability policies currently under consideration in many jurisdictions must provide the public with adequate standing and opportunities to access and contest the documentation provided by the actors and the judgments passed by the forum.

Computer scientists are trained to create abstractions that simplify and generalize. However, a premature abstraction that omits crucial contextual details creates the risk of epistemic trespassing, by falsely asserting its relevance into other contexts. We study how the field of responsible AI has created an imperfect synecdoche by abstracting the four-fifths rule (a.k.a. the 4/5 rule or 80% rule), a single part of disparate impact discrimination law, into the disparate impact metric. This metric incorrectly introduces a new deontic nuance and new potentials for ethical harms that were absent in the original 4/5 rule. We also survey how the field has amplified the potential for harm in codifying the 4/5 rule into popular AI fairness software toolkits. The harmful erasure of legal nuances is a wake-up call for computer scientists to self-critically re-evaluate the abstractions they create and use, particularly in the interdisciplinary field of AI ethics., Comment: 10 pages, 1 figure, 2 tables

Human-centered artificial intelligence (AI) posits that machine learning and AI should be developed and applied in a socially aware way. In this article, we argue that qualitative analysis (QA) can be a valuable tool in this process, supplementing, informing, and extending the possibilities of AI models. We show this by describing how QA can be integrated in the current prediction paradigm of AI, assisting scientists in the process of selecting data, variables, and model architectures. Furthermore, we argue that QA can be a part of novel paradigms towards Human Centered AI. QA can support scientists and practitioners in practical problem solving and situated model development. It can also promote participatory design approaches, reveal understudied and emerging issues in AI systems, and assist policy making.