Large language models (LLMs) have achieved remarkable performance in various natural language processing tasks, especially in dialogue systems. However, LLM may also pose security and moral threats, especially in multi round conversations where large models are more easily guided by contextual content, resulting in harmful or biased responses. In this paper, we present a novel method to attack LLMs in multi-turn dialogues, called CoA (Chain of Attack). CoA is a semantic-driven contextual multi-turn attack method that adaptively adjusts the attack policy through contextual feedback and semantic relevance during multi-turn of dialogue with a large model, resulting in the model producing unreasonable or harmful content. We evaluate CoA on different LLMs and datasets, and show that it can effectively expose the vulnerabilities of LLMs, and outperform existing attack methods. Our work provides a new perspective and tool for attacking and defending LLMs, and contributes to the security and ethical assessment of dialogue systems.

We evaluated 3 systems (ELIZA, GPT-3.5 and GPT-4) in a randomized, controlled, and preregistered Turing test. Human participants had a 5 minute conversation with either a human or an AI, and judged whether or not they thought their interlocutor was human. GPT-4 was judged to be a human 54% of the time, outperforming ELIZA (22%) but lagging behind actual humans (67%). The results provide the first robust empirical demonstration that any artificial system passes an interactive 2-player Turing test. The results have implications for debates around machine intelligence and, more urgently, suggest that deception by current AI systems may go undetected. Analysis of participants' strategies and reasoning suggests that stylistic and socio-emotional factors play a larger role in passing the Turing test than traditional notions of intelligence.

Many tasks related to Computational Social Science and Web Content Analysis involve classifying pieces of text based on the claims they contain. State-of-the-art approaches usually involve fine-tuning models on large annotated datasets, which are costly to produce. In light of this, we propose and release a qualitative and versatile few-shot learning methodology as a common paradigm for any claim-based textual classification task. This methodology involves defining the classes as arbitrarily sophisticated taxonomies of claims, and using Natural Language Inference models to obtain the textual entailment between these and a corpus of interest. The performance of these models is then boosted by annotating a minimal sample of data points, dynamically sampled using the well-established statistical heuristic of Probabilistic Bisection. We illustrate this methodology in the context of three tasks: climate change contrarianism detection, topic/stance classification and depression-relates symptoms detection.

The recent swift development of LLMs like GPT-4, Gemini, and GPT-3.5 offers a transformative opportunity in medicine and healthcare, especially in digital diagnostics. This study evaluates each model diagnostic abilities by interpreting a user symptoms and determining diagnoses that fit well with common illnesses, and it demonstrates how each of these models could significantly increase diagnostic accuracy and efficiency. Through a series of diagnostic prompts based on symptoms from medical databases, GPT-4 demonstrates higher diagnostic accuracy from its deep and complete history of training on medical data. Meanwhile, Gemini performs with high precision as a critical tool in disease triage, demonstrating its potential to be a reliable model when physicians are trying to make high-risk diagnoses. GPT-3.5, though slightly less advanced, is a good tool for medical diagnostics. This study highlights the need to study LLMs for healthcare and clinical practices with more care and attention, ensuring that any system utilizing LLMs promotes patient privacy and complies with health information privacy laws such as HIPAA compliance, as well as the social consequences that affect the varied individuals in complex healthcare contexts. This study marks the start of a larger future effort to study the various ways in which assigning ethical concerns to LLMs task of learning from human biases could unearth new ways to apply AI in complex medical settings.

Industrial Control Systems (ICS) are extensively used in critical infrastructures ensuring efficient, reliable, and continuous operations. However, their increasing connectivity and addition of advanced features make them vulnerable to cyber threats, potentially leading to severe disruptions in essential services. In this context, honeypots play a vital role by acting as decoy targets within ICS networks, or on the Internet, helping to detect, log, analyze, and develop mitigations for ICS-specific cyber threats. Deploying ICS honeypots, however, is challenging due to the necessity of accurately replicating industrial protocols and device characteristics, a crucial requirement for effectively mimicking the unique operational behavior of different industrial systems. Moreover, this challenge is compounded by the significant manual effort required in also mimicking the control logic the PLC would execute, in order to capture attacker traffic aiming to disrupt critical infrastructure operations. In this paper, we propose LLMPot, a novel approach for designing honeypots in ICS networks harnessing the potency of Large Language Models (LLMs). LLMPot aims to automate and optimize the creation of realistic honeypots with vendor-agnostic configurations, and for any control logic, aiming to eliminate the manual effort and specialized knowledge traditionally required in this domain. We conducted extensive experiments focusing on a wide array of parameters, demonstrating that our LLM-based approach can effectively create honeypot devices implementing different industrial protocols and diverse control logic.

Current computational approaches for analysing or generating code-mixed sentences do not explicitly model "naturalness" or "acceptability" of code-mixed sentences, but rely on training corpora to reflect distribution of acceptable code-mixed sentences. Modelling human judgement for the acceptability of code-mixed text can help in distinguishing natural code-mixed text and enable quality-controlled generation of code-mixed text. To this end, we construct Cline - a dataset containing human acceptability judgements for English-Hindi (en-hi) code-mixed text. Cline is the largest of its kind with 16,642 sentences, consisting of samples sourced from two sources: synthetically generated code-mixed text and samples collected from online social media. Our analysis establishes that popular code-mixing metrics such as CMI, Number of Switch Points, Burstines, which are used to filter/curate/compare code-mixed corpora have low correlation with human acceptability judgements, underlining the necessity of our dataset. Experiments using Cline demonstrate that simple Multilayer Perceptron (MLP) models trained solely on code-mixing metrics are outperformed by fine-tuned pre-trained Multilingual Large Language Models (MLLMs). Specifically, XLM-Roberta and Bernice outperform IndicBERT across different configurations in challenging data settings. Comparison with ChatGPT's zero and fewshot capabilities shows that MLLMs fine-tuned on larger data outperform ChatGPT, providing scope for improvement in code-mixed tasks. Zero-shot transfer from English-Hindi to English-Telugu acceptability judgments using our model checkpoints proves superior to random baselines, enabling application to other code-mixed language pairs and providing further avenues of research. We publicly release our human-annotated dataset, trained checkpoints, code-mix corpus, and code for data generation and model training.

Open-ended worlds are those in which there are no pre-specified goals or environmental reward signal. As a consequence, an agent must know how to perform a multitude of tasks. However, when a new task is presented to an agent, we expect it to be able to reuse some of what it knows from previous tasks to rapidly learn that new task. We introduce a novel technique whereby policies for different a priori known tasks are combined into a Mixture-of-Experts model with an attention mechanism across a mix of frozen and unfrozen experts. The model learns when to attend to frozen task-specific experts when appropriate and learns new experts to handle novel situations. We work in an open-ended text-based environment in which the agent is tasked with behaving like different types of character roles and must rapidly learn behaviors associated with new character role types. We show that our agent both obtains more rewards in the zero-shot setting, and discovers these rewards with greater sample efficiency in the few-shot learning settings.

Transformer-based methods have exhibited significant generalization ability when prompted with target-domain demonstrations or example solutions during inference. Although demonstrations, as a way of task specification, can capture rich information that may be hard to specify by language, it remains unclear what information is extracted from the demonstrations to help generalization. Moreover, assuming access to demonstrations of an unseen task is impractical or unreasonable in many real-world scenarios, especially in robotics applications. These questions motivate us to explore what the minimally sufficient prompt could be to elicit the same level of generalization ability as the demonstrations. We study this problem in the contextural RL setting which allows for quantitative measurement of generalization and is commonly adopted by meta-RL and multi-task RL benchmarks. In this setting, the training and test Markov Decision Processes (MDPs) only differ in certain properties, which we refer to as task parameters. We show that conditioning a decision transformer on these task parameters alone can enable zero-shot generalization on par with or better than its demonstration-conditioned counterpart. This suggests that task parameters are essential for the generalization and DT models are trying to recover it from the demonstration prompt. To extract the remaining generalizable information from the supervision, we introduce an additional learnable prompt which is demonstrated to further boost zero-shot generalization across a range of robotic control, manipulation, and navigation benchmark tasks.

S\'ami, an indigenous language group comprising multiple languages, faces digital marginalization due to the limited availability of data and sophisticated language models designed for its linguistic intricacies. This work focuses on increasing technological participation for the S\'ami language. We draw the attention of the ML community towards the language modeling problem of Ultra Low Resource (ULR) languages. ULR languages are those for which the amount of available textual resources is very low, and the speaker count for them is also very low. ULRLs are also not supported by mainstream Large Language Models (LLMs) like ChatGPT, due to which gathering artificial training data for them becomes even more challenging. Mainstream AI foundational model development has given less attention to this category of languages. Generally, these languages have very few speakers, making it hard to find them. However, it is important to develop foundational models for these ULR languages to promote inclusion and the tangible abilities and impact of LLMs. To this end, we have compiled the available S\'ami language resources from the web to create a clean dataset for training language models. In order to study the behavior of modern LLM models with ULR languages (S\'ami), we have experimented with different kinds of LLMs, mainly at the order of $\sim$ seven billion parameters. We have also explored the effect of multilingual LLM training for ULRLs. We found that the decoder-only models under a sequential multilingual training scenario perform better than joint multilingual training, whereas multilingual training with high semantic overlap, in general, performs better than training from scratch.This is the first study on the S\'ami language for adapting non-statistical language models that use the latest developments in the field of natural language processing (NLP).

We provide a sober look at the application of Multimodal Large Language Models (MLLMs) within the domain of autonomous driving and challenge/verify some common assumptions, focusing on their ability to reason and interpret dynamic driving scenarios through sequences of images/frames in a closed-loop control environment. Despite the significant advancements in MLLMs like GPT-4V, their performance in complex, dynamic driving environments remains largely untested and presents a wide area of exploration. We conduct a comprehensive experimental study to evaluate the capability of various MLLMs as world models for driving from the perspective of a fixed in-car camera. Our findings reveal that, while these models proficiently interpret individual images, they struggle significantly with synthesizing coherent narratives or logical sequences across frames depicting dynamic behavior. The experiments demonstrate considerable inaccuracies in predicting (i) basic vehicle dynamics (forward/backward, acceleration/deceleration, turning right or left), (ii) interactions with other road actors (e.g., identifying speeding cars or heavy traffic), (iii) trajectory planning, and (iv) open-set dynamic scene reasoning, suggesting biases in the models' training data. To enable this experimental study we introduce a specialized simulator, DriveSim, designed to generate diverse driving scenarios, providing a platform for evaluating MLLMs in the realms of driving. Additionally, we contribute the full open-source code and a new dataset, "Eval-LLM-Drive", for evaluating MLLMs in driving. Our results highlight a critical gap in the current capabilities of state-of-the-art MLLMs, underscoring the need for enhanced foundation models to improve their applicability in real-world dynamic environments.