Research Papers

Conceptualising Contestability: Perspectives on Contesting Algorithmic Decisions

Henrietta Lyons, Eduardo Velloso, Tim Miller

As the use of algorithmic systems in high-stakes decision-making increases, the ability to contest algorithmic decisions is being recognised as an important safeguard for individuals. Yet, there is little guidance on what `contestability'--the ability to contest decisions--in relation to algorithmic decision-making requires. Recent research presents different conceptualisations of contestability in algorithmic decision-making. We contribute to this growing body of work by describing and analysing the perspectives of people and organisations who made submissions in response to Australia's proposed `AI Ethics Framework', the first framework of its kind to include `contestability' as a core ethical principle. Our findings reveal that while the nature of contestability is disputed, it is seen as a way to protect individuals, and it resembles contestability in relation to human decision-making. We reflect on and discuss the implications of these findings.

Although AI is transforming the world, there are serious concerns about its ability to behave and make decisions responsibly. Many ethical regulations, principles, and frameworks for responsible AI have been issued recently. However, they are high level and difficult to put into practice. On the other hand, most AI researchers focus on algorithmic solutions, while the responsible AI challenges actually crosscut the entire engineering lifecycle and components of AI systems. To close the gap in operationalizing responsible AI, this paper aims to develop a roadmap on software engineering for responsible AI. The roadmap focuses on (i) establishing multi-level governance for responsible AI systems, (ii) setting up the development processes incorporating process-oriented practices for responsible AI systems, and (iii) building responsible-AI-by-design into AI systems through system-level architectural style, patterns and techniques.

This paper examines the systemic risks posed by incremental advancements in artificial intelligence, developing the concept of `gradual disempowerment', in contrast to the abrupt takeover scenarios commonly discussed in AI safety. We analyze how even incremental improvements in AI capabilities can undermine human influence over large-scale systems that society depends on, including the economy, culture, and nation-states. As AI increasingly replaces human labor and cognition in these domains, it can weaken both explicit human control mechanisms (like voting and consumer choice) and the implicit alignments with human interests that often arise from societal systems' reliance on human participation to function. Furthermore, to the extent that these systems incentivise outcomes that do not line up with human preferences, AIs may optimize for those outcomes more aggressively. These effects may be mutually reinforcing across different domains: economic power shapes cultural narratives and political decisions, while cultural shifts alter economic and political behavior. We argue that this dynamic could lead to an effectively irreversible loss of human influence over crucial societal systems, precipitating an existential catastrophe through the permanent disempowerment of humanity. This suggests the need for both technical research and governance approaches that specifically address the risk of incremental erosion of human influence across interconnected societal systems.

Ethics in the Age of AI: An Analysis of AI Practitioners' Awareness and Challenges

Aastha Pant, Rashina Hoda, Simone V. Spiegler +2

Ethics in AI has become a debated topic of public and expert discourse in recent years. But what do people who build AI - AI practitioners - have to say about their understanding of AI ethics and the challenges associated with incorporating it in the AI-based systems they develop? Understanding AI practitioners' views on AI ethics is important as they are the ones closest to the AI systems and can bring about changes and improvements. We conducted a survey aimed at understanding AI practitioners' awareness of AI ethics and their challenges in incorporating ethics. Based on 100 AI practitioners' responses, our findings indicate that majority of AI practitioners had a reasonable familiarity with the concept of AI ethics, primarily due to workplace rules and policies. Privacy protection and security was the ethical principle that majority of them were aware of. Formal education/training was considered somewhat helpful in preparing practitioners to incorporate AI ethics. The challenges that AI practitioners faced in the development of ethical AI-based systems included (i) general challenges, (ii) technology-related challenges and (iii) human-related challenges. We also identified areas needing further investigation and provided recommendations to assist AI practitioners and companies in incorporating ethics into AI development.

Investigating Practices and Opportunities for Cross-functional Collaboration around AI Fairness in Industry Practice

Wesley Hanwen Deng, Nur Yildirim, Monica Chang +3

An emerging body of research indicates that ineffective cross-functional collaboration -- the interdisciplinary work done by industry practitioners across roles -- represents a major barrier to addressing issues of fairness in AI design and development. In this research, we sought to better understand practitioners' current practices and tactics to enact cross-functional collaboration for AI fairness, in order to identify opportunities to support more effective collaboration. We conducted a series of interviews and design workshops with 23 industry practitioners spanning various roles from 17 companies. We found that practitioners engaged in bridging work to overcome frictions in understanding, contextualization, and evaluation around AI fairness across roles. In addition, in organizational contexts with a lack of resources and incentives for fairness work, practitioners often piggybacked on existing requirements (e.g., for privacy assessments) and AI development norms (e.g., the use of quantitative evaluation metrics), although they worry that these tactics may be fundamentally compromised. Finally, we draw attention to the invisible labor that practitioners take on as part of this bridging and piggybacking work to enact interdisciplinary collaboration for fairness. We close by discussing opportunities for both FAccT researchers and AI practitioners to better support cross-functional collaboration for fairness in the design and development of AI systems.

Predominately in explainable artificial intelligence (XAI) research, the Shapley value (SV) is applied to determine feature attributions for any black box model. Shapley interaction indices extend the SV to define any-order feature interactions. Defining a unique Shapley interaction index is an open research question and, so far, three definitions have been proposed, which differ by their choice of axioms. Moreover, each definition requires a specific approximation technique. Here, we propose SHAPley Interaction Quantification (SHAP-IQ), an efficient sampling-based approximator to compute Shapley interactions for arbitrary cardinal interaction indices (CII), i.e. interaction indices that satisfy the linearity, symmetry and dummy axiom. SHAP-IQ is based on a novel representation and, in contrast to existing methods, we provide theoretical guarantees for its approximation quality, as well as estimates for the variance of the point estimates. For the special case of SV, our approach reveals a novel representation of the SV and corresponds to Unbiased KernelSHAP with a greatly simplified calculation. We illustrate the computational efficiency and effectiveness by explaining language, image classification and high-dimensional synthetic models.

Introducing v0.5 of the AI Safety Benchmark from MLCommons

Bertie Vidgen, Adarsh Agrawal, Ahmed M. Ahmed +97

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.

Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence (XAI), a fast-growing research area that is so far lacking in firm theoretical foundations. Building on work in logic, probability, and causality, we establish the central role of necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We provide a sound and complete algorithm for computing explanatory factors with respect to a given context, and demonstrate its flexibility and competitive performance against state of the art alternatives on various tasks.

Artificial intelligence (AI) is a technology which is increasingly being utilised in society and the economy worldwide, and its implementation is planned to become more prevalent in coming years. AI is increasingly being embedded in our lives, supplementing our pervasive use of digital technologies. But this is being accompanied by disquiet over problematic and dangerous implementations of AI, or indeed, even AI itself deciding to do dangerous and problematic actions, especially in fields such as the military, medicine and criminal justice. These developments have led to concerns about whether and how AI systems adhere, and will adhere to ethical standards. These concerns have stimulated a global conversation on AI ethics, and have resulted in various actors from different countries and sectors issuing ethics and governance initiatives and guidelines for AI. Such developments form the basis for our research in this report, combining our international and interdisciplinary expertise to give an insight into what is happening in Australia, China, Europe, India and the US.

SimpleSafetyTests: a Test Suite for Identifying Critical Safety Risks in Large Language Models

Bertie Vidgen, Nino Scherrer, Hannah Rose Kirk +4

The past year has seen rapid acceleration in the development of large language models (LLMs). However, without proper steering and safeguards, LLMs will readily follow malicious instructions, provide unsafe advice, and generate toxic content. We introduce SimpleSafetyTests (SST) as a new test suite for rapidly and systematically identifying such critical safety risks. The test suite comprises 100 test prompts across five harm areas that LLMs, for the vast majority of applications, should refuse to comply with. We test 11 open-access and open-source LLMs and four closed-source LLMs, and find critical safety weaknesses. While some of the models do not give a single unsafe response, most give unsafe responses to more than 20% of the prompts, with over 50% unsafe responses in the extreme. Prepending a safety-emphasising system prompt substantially reduces the occurrence of unsafe responses, but does not completely stop them from happening. Trained annotators labelled every model response to SST (n = 3,000). We use these annotations to evaluate five AI safety filters (which assess whether a models' response is unsafe given a prompt) as a way of automatically evaluating models' performance on SST. The filters' performance varies considerably. There are also differences across the five harm areas, and on the unsafe versus safe responses. The widely-used Perspective API has 72% accuracy and a newly-created zero-shot prompt to OpenAI's GPT-4 performs best with 89% accuracy. Content Warning: This paper contains prompts and responses that relate to child abuse, suicide, self-harm and eating disorders, scams and fraud, illegal items, and physical harm.

Rethinking Model Evaluation as Narrowing the Socio-Technical Gap

Q. Vera Liao, Ziang Xiao

The recent development of generative large language models (LLMs) poses new challenges for model evaluation that the research community and industry have been grappling with. While the versatile capabilities of these models ignite much excitement, they also inevitably make a leap toward homogenization: powering a wide range of applications with a single, often referred to as ``general-purpose'', model. In this position paper, we argue that model evaluation practices must take on a critical task to cope with the challenges and responsibilities brought by this homogenization: providing valid assessments for whether and how much human needs in diverse downstream use cases can be satisfied by the given model (\textit{socio-technical gap}). By drawing on lessons about improving research realism from the social sciences, human-computer interaction (HCI), and the interdisciplinary field of explainable AI (XAI), we urge the community to develop evaluation methods based on real-world contexts and human requirements, and embrace diverse evaluation methods with an acknowledgment of trade-offs between realisms and pragmatic costs to conduct the evaluation. By mapping HCI and current NLG evaluation methods, we identify opportunities for evaluation methods for LLMs to narrow the socio-technical gap and pose open questions.

This paper presents an empirical study on the weights of neural networks, where we interpret each model as a point in a high-dimensional space -- the neural weight space. To explore the complex structure of this space, we sample from a diverse selection of training variations (dataset, optimization procedure, architecture, etc.) of neural network classifiers, and train a large number of models to represent the weight space. Then, we use a machine learning approach for analyzing and extracting information from this space. Most centrally, we train a number of novel deep meta-classifiers with the objective of classifying different properties of the training setup by identifying their footprints in the weight space. Thus, the meta-classifiers probe for patterns induced by hyper-parameters, so that we can quantify how much, where, and when these are encoded through the optimization process. This provides a novel and complementary view for explainable AI, and we show how meta-classifiers can reveal a great deal of information about the training setup and optimization, by only considering a small subset of randomly selected consecutive weights. To promote further research on the weight space, we release the neural weight space (NWS) dataset -- a collection of 320K weight snapshots from 16K individually trained deep neural networks.