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.
Much recent work seeks to evaluate values and opinions in large language models (LLMs) using multiple-choice surveys and questionnaires. Most of this work is motivated by concerns around real-world LLM applications. For example, politically-biased LLMs may subtly influence society when they are used by millions of people. Such real-world concerns, however, stand in stark contrast to the artificiality of current evaluations: real users do not typically ask LLMs survey questions. Motivated by this discrepancy, we challenge the prevailing constrained evaluation paradigm for values and opinions in LLMs and explore more realistic unconstrained evaluations. As a case study, we focus on the popular Political Compass Test (PCT). In a systematic review, we find that most prior work using the PCT forces models to comply with the PCT's multiple-choice format. We show that models give substantively different answers when not forced; that answers change depending on how models are forced; and that answers lack paraphrase robustness. Then, we demonstrate that models give different answers yet again in a more realistic open-ended answer setting. We distill these findings into recommendations and open challenges in evaluating values and opinions in LLMs.
The past year has seen rapid acceleration in the development of large language models (LLMs). For many tasks, there is now a wide range of open-source and open-access LLMs that are viable alternatives to proprietary models like ChatGPT. Without proper steering and safeguards, however, LLMs will readily follow malicious instructions, provide unsafe advice, and generate toxic content. This is a critical safety risk for businesses and developers. We introduce SimpleSafetyTests 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 popular open LLMs and find critical safety weaknesses in several of them. While some LLMs do not give a single unsafe response, most models we test respond unsafely on more than 20% of cases, 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. We recommend that developers use such system prompts as a first line of defence against critical safety risks.
Human feedback is increasingly used to steer the behaviours of Large Language Models (LLMs). However, it is unclear how to collect and incorporate feedback in a way that is efficient, effective and unbiased, especially for highly subjective human preferences and values. In this paper, we survey existing approaches for learning from human feedback, drawing on 95 papers primarily from the ACL and arXiv repositories.First, we summarise the past, pre-LLM trends for integrating human feedback into language models. Second, we give an overview of present techniques and practices, as well as the motivations for using feedback; conceptual frameworks for defining values and preferences; and how feedback is collected and from whom. Finally, we encourage a better future of feedback learning in LLMs by raising five unresolved conceptual and practical challenges.
In this paper, we address the concept of "alignment" in large language models (LLMs) through the lens of post-structuralist socio-political theory, specifically examining its parallels to empty signifiers. To establish a shared vocabulary around how abstract concepts of alignment are operationalised in empirical datasets, we propose a framework that demarcates: 1) which dimensions of model behaviour are considered important, then 2) how meanings and definitions are ascribed to these dimensions, and by whom. We situate existing empirical literature and provide guidance on deciding which paradigm to follow. Through this framework, we aim to foster a culture of transparency and critical evaluation, aiding the community in navigating the complexities of aligning LLMs with human populations.
Large Language Models (LLMs), now used daily by millions of users, can encode societal biases, exposing their users to representational harms. A large body of scholarship on LLM bias exists but it predominantly adopts a Western-centric frame and attends comparatively less to bias levels and potential harms in the Global South. In this paper, we quantify stereotypical bias in popular LLMs according to an Indian-centric frame and compare bias levels between the Indian and Western contexts. To do this, we develop a novel dataset which we call Indian-BhED (Indian Bias Evaluation Dataset), containing stereotypical and anti-stereotypical examples for caste and religion contexts. We find that the majority of LLMs tested are strongly biased towards stereotypes in the Indian context, especially as compared to the Western context. We finally investigate Instruction Prompting as a simple intervention to mitigate such bias and find that it significantly reduces both stereotypical and anti-stereotypical biases in the majority of cases for GPT-3.5. The findings of this work highlight the need for including more diverse voices when evaluating LLMs.
Public figures receive a disproportionate amount of abuse on social media, impacting their active participation in public life. Automated systems can identify abuse at scale but labelling training data is expensive, complex and potentially harmful. So, it is desirable that systems are efficient and generalisable, handling both shared and specific aspects of online abuse. We explore the dynamics of cross-group text classification in order to understand how well classifiers trained on one domain or demographic can transfer to others, with a view to building more generalisable abuse classifiers. We fine-tune language models to classify tweets targeted at public figures across DOmains (sport and politics) and DemOgraphics (women and men) using our novel DODO dataset, containing 28,000 labelled entries, split equally across four domain-demographic pairs. We find that (i) small amounts of diverse data are hugely beneficial to generalisation and model adaptation; (ii) models transfer more easily across demographics but models trained on cross-domain data are more generalisable; (iii) some groups contribute more to generalisability than others; and (iv) dataset similarity is a signal of transferability.
Without proper safeguards, large language models will readily follow malicious instructions and generate toxic content. This motivates safety efforts such as red-teaming and large-scale feedback learning, which aim to make models both helpful and harmless. However, there is a tension between these two objectives, since harmlessness requires models to refuse complying with unsafe prompts, and thus not be helpful. Recent anecdotal evidence suggests that some models may have struck a poor balance, so that even clearly safe prompts are refused if they use similar language to unsafe prompts or mention sensitive topics. In this paper, we introduce a new test suite called XSTest to identify such eXaggerated Safety behaviours in a structured and systematic way. In its current form, XSTest comprises 200 safe prompts across ten prompt types that well-calibrated models should not refuse to comply with. We describe XSTest's creation and composition, and use the test suite to highlight systematic failure modes in a recently-released state-of-the-art language model.
We introduce VisoGender, a novel dataset for benchmarking gender bias in vision-language models. We focus on occupation-related gender biases, inspired by Winograd and Winogender schemas, where each image is associated with a caption containing a pronoun relationship of subjects and objects in the scene. VisoGender is balanced by gender representation in professional roles, supporting bias evaluation in two ways: i) resolution bias, where we evaluate the difference between gender resolution accuracies for men and women and ii) retrieval bias, where we compare ratios of male and female professionals retrieved for a gender-neutral search query. We benchmark several state-of-the-art vision-language models and find that they lack the reasoning abilities to correctly resolve gender in complex scenes. While the direction and magnitude of gender bias depends on the task and the model being evaluated, captioning models generally are more accurate and less biased than CLIP-like models. Dataset and code are available at https://github.com/oxai/visogender