Large Language Models (LLMs) have excelled at language understanding and generating human-level text. However, even with supervised training and human alignment, these LLMs are susceptible to adversarial attacks where malicious users can prompt the model to generate undesirable text. LLMs also inherently encode potential biases that can cause various harmful effects during interactions. Bias evaluation metrics lack standards as well as consensus and existing methods often rely on human-generated templates and annotations which are expensive and labor intensive. In this work, we train models to automatically create adversarial prompts to elicit biased responses from target LLMs. We present LLM- based bias evaluation metrics and also analyze several existing automatic evaluation methods and metrics. We analyze the various nuances of model responses, identify the strengths and weaknesses of model families, and assess where evaluation methods fall short. We compare these metrics to human evaluation and validate that the LLM-as-a-Judge metric aligns with human judgement on bias in response generation.

As Large Language Models (LLMs) become increasingly sophisticated and ubiquitous in natural language processing (NLP) applications, ensuring their robustness, trustworthiness, and alignment with human values has become a critical challenge. This paper presents a novel framework for contextual grounding in textual models, with a particular emphasis on the Context Representation stage. Our approach aims to enhance the reliability and ethical alignment of these models through a comprehensive, context-aware methodology. By explicitly capturing and representing relevant situational, cultural, and ethical contexts in a machine-readable format, we lay the foundation for anchoring a model's behavior within these contexts. Our approach leverages techniques from knowledge representation and reasoning, such as ontologies, semantic web technologies, and logic-based formalisms. We evaluate our framework on real-world textual datasets, demonstrating its effectiveness in improving model performance, fairness, and alignment with human expectations, while maintaining high accuracy. Furthermore, we discuss the other key components of the framework, including context-aware encoding, context-aware learning, interpretability and explainability, and continuous monitoring and adaptation. This research contributes to the growing body of work on responsible AI, offering a practical approach to developing more reliable, trustworthy, and ethically-aligned language models. Our findings have significant implications for the deployment of LLMs in sensitive domains such as healthcare, legal systems, and social services, where contextual understanding is paramount.

Natural language processing (NLP) has seen significant advancements with the advent of large language models (LLMs). However, substantial improvements are still needed for languages other than English, especially for specific domains like the applications of Mercosur Common Nomenclature (NCM), a Brazilian Harmonized System (HS). To address this gap, this study uses TeenyTineLLaMA, a foundational Portuguese LLM, as an LLM source to implement the NCM application processing. Additionally, a simplified Retrieval-Augmented Fine-Tuning (RAFT) technique, termed SLIM-RAFT, is proposed for task-specific fine-tuning of LLMs. This approach retains the chain-of-thought (CoT) methodology for prompt development in a more concise and streamlined manner, utilizing brief and focused documents for training. The proposed model demonstrates an efficient and cost-effective alternative for fine-tuning smaller LLMs, significantly outperforming TeenyTineLLaMA and ChatGPT-4 in the same task. Although the research focuses on NCM applications, the methodology can be easily adapted for HS applications worldwide.

Given the prevalence of large language models (LLMs) and the prohibitive cost of training these models from scratch, dynamically forgetting specific knowledge e.g., private or proprietary, without retraining the model has become an important capability. This paper proposes a novel method to achieve this objective called UNLEARN. The approach builds upon subspace methods to identify and specifically target the removal of knowledge without adversely affecting other knowledge in the LLM. Results demonstrate 96% of targeted knowledge can be forgotten while maintaining performance on other knowledge within 2.5% of the original model, significantly outperforming the discriminatory abilities of the previous state-of-the-art. A dual method called LEARN is also proposed for targeted knowledge addition. Results show LEARN can match the fine-tuning accuracy of Low-Rank Adaptation (LoRA) without adversely affecting similar tasks.

We outline the principles of classical assurance for computer-based systems that pose significant risks. We then consider application of these principles to systems that employ Artificial Intelligence (AI) and Machine Learning (ML). A key element in this "dependability" perspective is a requirement to have near-complete understanding of the behavior of critical components, and this is considered infeasible for AI and ML. Hence the dependability perspective aims to minimize trust in AI and ML elements by using "defense in depth" with a hierarchy of less complex systems, some of which may be highly assured conventionally engineered components, to "guard" them. This may be contrasted with the "trustworthy" perspective that seeks to apply assurance to the AI and ML elements themselves. In cyber-physical and many other systems, it is difficult to provide guards that do not depend on AI and ML to perceive their environment (e.g., other vehicles sharing the road with a self-driving car), so both perspectives are needed and there is a continuum or spectrum between them. We focus on architectures toward the dependability end of the continuum and invite others to consider additional points along the spectrum. For guards that require perception using AI and ML, we examine ways to minimize the trust placed in these elements; they include diversity, defense in depth, explanations, and micro-ODDs. We also examine methods to enforce acceptable behavior, given a model of the world. These include classical cyber-physical calculations and envelopes, and normative rules based on overarching principles, constitutions, ethics, or reputation. We apply our perspective to autonomous systems, AI systems for specific functions, generic AI such as Large Language Models, and to Artificial General Intelligence (AGI), and we propose current best practice and an agenda for research.

This paper presents a compute-efficient approach to pre-training a Language Model - the "1.5-Pints" - in only 9 days, while outperforming state-of-the-art models as an instruction-following assistant. Based on MT-Bench (a benchmark that emulates human judgments), 1.5-Pints outperforms Apple's OpenELM and Microsoft's Phi. This is achieved by a carefully curated pre-training dataset of 57 billion tokens, using a mix of automated workflows and manual human review. The selection of the dataset prioritizes content that is considered expository and "textbook-like" to aid the model in reasoning and logical deduction, culminating in its overall ability as a strong and versatile AI model. In terms of the model architecture, we employed a modified Mistral tokenizer, alongside a Llama-2 architecture for wider compatibility. For training, we adopted the methodologies used by StableLM, TinyLlama, and Huggingface Zephyr.

The release of models like GPT [3] and Gemini [28] through platforms like ChatGPT and Bard have transformed Large Language Models (LLMs) into general-purpose tools that can be used by everyone. Although designed for next token prediction, LLMs have been shown to have emergent abilities and are able to perform a wide variety of tasks without task-specific training data [3, 20, 25, 30, 31]. Unfortunately, LLMs also frequently return wrong results, such as fictitious claims ("hallucinations") or conclusions that defy common sense or (naive qualitative) physics [13, 16, 27]. Such shortcoming may or may not be obvious but in any case impact trustworthiness. A recent famous example was a lawyer who submitted a legal brief generated by ChatGPT which contained many errors and false references [5, 6]. Asking the LLM for an explanation might help, but the explanation might contain errors again and does not necessarily reflect the process used to obtain its answer. Equipping and checking LLMs with trustworthy (logical) reasoning remains to be a current major problem [21, 22]. A general approach to address this problem equips LLMs with external functionality [8, 10, 13, 19, 21]. These equipped models are referred to as Augmented Language Models (ALMs).

Elon Musk is once again suing OpenAI and its chief executive, Sam Altman, resurrecting a legal battle against his former partners with a case that now claims they manipulated him into co-founding the artificial intelligence company. Months after abruptly withdrawing a similar lawsuit without explanation, Musk filed a new lawsuit on Monday in a northern California federal court. OpenAI denied the allegations in a statement to the Guardian, pointing to its previous blogposts about Musk's initial lawsuit earlier this year. Musk's latest complaint claims the case is a "textbook tale of altruism versus greed", repeating allegations in his previous suit that his former co-founders in OpenAI betrayed him by turning the company from a non-profit into a largely for-profit enterprise. "The perfidy and deceit is of Shakespearean proportions," it states.

Elon Musk has filed another lawsuit against OpenAI and the company's CEO Sam Altman, two months after withdrawing a previous one. Musk once again alleges that OpenAI breached its founding commitments by putting commercial concerns ahead of the public good. This time around, though, the suit has been filed in federal court rather than in a state court. That's because the new filing alleges that OpenAI violated federal racketeering laws by conspiring to defraud Musk, according to his lawyer, Marc Toberoff. "The previous suit lacked teeth -- and I don't believe in the tooth fairy," Toberoff told The New York Times. "This is a much more forceful lawsuit."

According to Bloomberg and The New York Times, Meta is in talks with the likes of Keegan-Michael Key, Awkwafina and Dame Judi Dench, among others, for its AI projects. The company apparently intends to incorporate their voices into a conversational generative AI-slash-digital assistant called MetaAI, which is rumored to be like Siri and Google Assistant, which could live within Facebook, Meta hardware, and all the other parts of the multimillion-dollar social network company. The actors' representatives are still negotiating for stricter limits, though SAG-AFTRA has reportedly agreed on terms with Meta. SAG-AFTRA, if you recall, fought for provisions to protect actors from the threat of job loss due to AI. Didn't Meta already do something like this? Yes. During its Connect event last year, the company also introduced a chatbot platform with 28 "characters" voiced by celebrities, including Snoop Dogg, Paris Hilton, Dwyane Wade and Kendall Jenner.