Medical experts may use Artificial Intelligence (AI) systems with greater trust if these are supported by contextual explanations that let the practitioner connect system inferences to their context of use. However, their importance in improving model usage and understanding has not been extensively studied. Hence, we consider a comorbidity risk prediction scenario and focus on contexts regarding the patients clinical state, AI predictions about their risk of complications, and algorithmic explanations supporting the predictions. We explore how relevant information for such dimensions can be extracted from Medical guidelines to answer typical questions from clinical practitioners. We identify this as a question answering (QA) task and employ several state-of-the-art LLMs to present contexts around risk prediction model inferences and evaluate their acceptability. Finally, we study the benefits of contextual explanations by building an end-to-end AI pipeline including data cohorting, AI risk modeling, post-hoc model explanations, and prototyped a visual dashboard to present the combined insights from different context dimensions and data sources, while predicting and identifying the drivers of risk of Chronic Kidney Disease - a common type-2 diabetes comorbidity. All of these steps were performed in engagement with medical experts, including a final evaluation of the dashboard results by an expert medical panel. We show that LLMs, in particular BERT and SciBERT, can be readily deployed to extract some relevant explanations to support clinical usage. To understand the value-add of the contextual explanations, the expert panel evaluated these regarding actionable insights in the relevant clinical setting. Overall, our paper is one of the first end-to-end analyses identifying the feasibility and benefits of contextual explanations in a real-world clinical use case.

Recent advances in instruction-following large language models (LLMs) have led to dramatic improvements in a range of NLP tasks. Unfortunately, we find that the same improved capabilities amplify the dual-use risks for malicious purposes of these models. Dual-use is difficult to prevent as instruction-following capabilities now enable standard attacks from computer security. The capabilities of these instruction-following LLMs provide strong economic incentives for dual-use by malicious actors. In particular, we show that instruction-following LLMs can produce targeted malicious content, including hate speech and scams, bypassing in-the-wild defenses implemented by LLM API vendors. Our analysis shows that this content can be generated economically and at cost likely lower than with human effort alone. Together, our findings suggest that LLMs will increasingly attract more sophisticated adversaries and attacks, and addressing these attacks may require new approaches to mitigations.

Text-based dialogues are now widely used to solve real-world problems. In cases where solution strategies are already known, they can sometimes be codified into workflows and used to guide humans or artificial agents through the task of helping clients. We introduce a new problem formulation that we call Workflow Discovery (WD) in which we are interested in the situation where a formal workflow may not yet exist. Still, we wish to discover the set of actions that have been taken to resolve a particular problem. We also examine a sequence-to-sequence (Seq2Seq) approach for this novel task. We present experiments where we extract workflows from dialogues in the Action-Based Conversations Dataset (ABCD). Since the ABCD dialogues follow known workflows to guide agents, we can evaluate our ability to extract such workflows using ground truth sequences of actions. We propose and evaluate an approach that conditions models on the set of possible actions, and we show that using this strategy, we can improve WD performance. Our conditioning approach also improves zero-shot and few-shot WD performance when transferring learned models to unseen domains within and across datasets. Further, on ABCD a modified variant of our Seq2Seq method achieves state-of-the-art performance on related but different problems of Action State Tracking (AST) and Cascading Dialogue Success (CDS) across many evaluation metrics.

Despite cross-lingual generalization demonstrated by pre-trained multilingual models, the translate-train paradigm of transferring English datasets across multiple languages remains to be a key mechanism for training task-specific multilingual models. However, for many low-resource languages, the availability of a reliable translation service entails significant amounts of costly human-annotated translation pairs. Further, translation services may continue to be brittle due to domain mismatch between task-specific input text and general-purpose text used for training translation models. For multilingual semantic parsing, we demonstrate the effectiveness and flexibility offered by large language models (LLMs) for translating English datasets into several languages via few-shot prompting. Through extensive comparisons on two public datasets, MTOP and MASSIVE, spanning 50 languages and several domains, we show that our method of translating data using LLMs outperforms a strong translate-train baseline on 41 out of 50 languages. We study the key design choices that enable more effective multilingual data translation via prompted LLMs.

If your work involves analyzing and reporting on data, then it's understandable that you might feel a bit concerned by the rapid advances being made by artificial intelligence (AI). In particular, the viral ChatGPT app has captured the imagination of the general public in recent months, acting as a powerful demonstration of what AI is already capable of. For some, it may also seem like a warning about what might be in store for the future. Will ChatGPT Put Data Analysts Out of Work? Undoubtedly, one of the strengths of AI is its ability to make sense of large amounts of data – searching out patterns and putting it into reports, documents, and formats that humans can easily understand.

Since the official announcement in November 2022, there has been an enormous amount of buzz and excitement about OpenAI's ChatGPT. Industry experts are publishing articles about it, social networks are filled with comments about it, and local, national, and global news organisations are reporting stories about it. From students using ChatGPT to complete assignments for class to me getting a little help from ChatGPT to write my latest'Virtual Viewpoint' column, it certainly seems like everyone is testing it out. As a specialist within the conversational AI space, Creative Virtual is excited about what ChatGPT and the technology behind it bring to our industry. We've been having lots of discussions with our customers and partners, as well as internally, about how this can deliver value to businesses using our V-Person solutions.

Check out all the on-demand sessions from the Intelligent Security Summit here. Aidan Gomez, cofounder and CEO of Cohere AI, admits that the company, which offers developers and businesses access to natural language processing (NLP) powered by large language models (LLMs), is "crazy under the radar." Given the quality of the company's foundation models, which many say are competitive with the best from Google, OpenAI and others, that shouldn't be the case, he told VentureBeat. But Cohere, he emphasizes, has been "squarely focused on the enterprise and how we can add value there." In any case, the Toronto-based Cohere, founded in 2019 by Gomez, Ivan Zhang and Nick Frosst, may not remain unnoticed for long.

Microsoft CEO Satya Nadella is keen for the software maker to be seen as a leader in AI, and counter any response from rival Google. In fact, he's so eager to get Microsoft pushing in this direction that he arrived early to the company's Bing AI event on Tuesday morning. While the event was planned for 10AM PT, Nadella wanted to start 20 minutes early. That left event organizers scrambling to start earlier than expected, with the event kicking off five minutes earlier than planned and Nadella appearing onstage two minutes before the original 10AM PT start time.

Technological breakthroughs in the U.S. never fail to inspire challengers, followers and opportunists in China. It's ChatGPT's turn to capture the imagination of the world's largest internet population. On WeChat, ChatGPT's "trending index," an indicator of a keyword's popularity on the social network, rose 155 folds within the last 30 days. It's fascinating to watch how OpenAI's powerful language model sparks great interest among the country's tech giants, startups and ordinary people, not least because it offers a lens to understand the state of the AI race between two superpowers. Unlike many other major Western internet platforms, the ChatGPT site isn't blocked in China, yet.

First of all, you have to remember the relationship with OpenAI and our cooperation with OpenAI has many facets. The most important thing is what we've done over the last four years to actually build out the core infrastructure on which OpenAI is built: these large models, the training infrastructure -- and the infrastructure doesn't look like regular cloud infrastructure. We had to evolve Azure to have specialized AI infrastructure on which OpenAI is built. And by the way, Inception and Character.ai There will be many others who will use Azure infrastructure.