Robustness and Interpretability of Foundation Models

对研究背景的简要描述，包括潜在的影响：大型语言模型和基础模型在各种应用中被广泛采用，例如编程、网络搜索、编写书籍、辅助医生、客户服务、图形设计等。这项研究解决了生成式人工智能开发和部署中的重要挑战，例如健壮性、公平性、隐私、可解释性和安全使用。目的和目标：目标包括：(I)对自回归模型的形式化和基准测试攻击，(Ii)新的安全评估方法，(Iii)对系统提示泄漏的攻击，(Iv)认证专业领域，(V)检查跨语言的公平性，(Vi)统一稳健性和隐私，(Vii)认证校准和集合，(Vii)解释安全/不安全的行为，以及(Viii)发展前缀和后缀调整的理论。研究方法的新奇之处：研究将引入新的想法，例如(I)编入预算的对抗性提示，(Ii)聊天式攻击，(Iii)基于分布分歧的目标，(Iv)针对差异隐私的集合级别Lipschitzness，以及(V)校准错误感知培训。对EPSRC的战略和研究领域的挑战：人工智能参与的公司或合作者：无

Brief description of the context of the research including potential impact: Large language models and foundation models are seeing widespread adoption in various applications, e.g., programming, web search, writing books, assisting medical practitioners, customer service, graphic design, etc. This research addresses important challenges in the development and deployment of generative AI, such as robustness, fairness, privacy, interpretability, and safe use.Aims and Objectives: Objectives include (i) formalizing and benchmarking attacks on autoregressive models, (ii) new evaluation methods for safety, (iii) attacks for system prompt leakage, (iv) certifying domains of expertise, (v) examining fairness across languages, (vi) unifying robustness and privacy, (vii) certifying calibration and ensembles, (vii) interpreting safe/unsafe behaviour, and (viii) developing a theory of prefix and suffix tuning.Novelty of the research methodology: The research would introduce novel ideas such as (i) budgeted adversarial prompts, (ii) chat-style attacks, (iii) distribution divergence-based objectives, (iv) generalizing set-level Lipschitzness for differential privacy, and (v) calibration error-aware training.Alignment to EPSRC's strategies and research areas: Artificial intelligenceAny companies or collaborators involved: None