Health disparities are differences in health outcomes and access to healthcare between different groups, including racial and ethnic minorities, low-income people, and rural residents. An artificial intelligence (AI) program called large language models (LLMs) can understand and generate human language, improving health communication and reducing health disparities. There are many challenges in using LLMs in human-doctor interaction, including the need for diverse and representative data, privacy concerns, and collaboration between healthcare providers and technology experts. We introduce the comparative investigation of domain-specific large language models such as SciBERT with a multi-purpose LLMs BERT. We used cosine similarity to analyze text queries about health disparities in exam rooms when factors such as race are used alone. Using text queries, SciBERT fails when it doesn't differentiate between queries text: "race" alone and "perpetuates health disparities." We believe clinicians can use generative AI to create a draft response when communicating asynchronously with patients. However, careful attention must be paid to ensure they are developed and implemented ethically and equitably.

The era post-2018 marked the advent of Large Language Models (LLMs), with innovations such as OpenAI's ChatGPT showcasing prodigious linguistic prowess. As the industry galloped toward augmenting model parameters and capitalizing on vast swaths of human language data, security and privacy challenges also emerged. Foremost among these is the potential inadvertent accrual of Personal Identifiable Information (PII) during web-based data acquisition, posing risks of unintended PII disclosure. While strategies like RLHF during training and Catastrophic Forgetting have been marshaled to control the risk of privacy infringements, recent advancements in LLMs, epitomized by OpenAI's fine-tuning interface for GPT-3.5, have reignited concerns. One may ask: can the fine-tuning of LLMs precipitate the leakage of personal information embedded within training datasets? This paper reports the first endeavor to seek the answer to the question, particularly our discovery of a new LLM exploitation avenue, called the Janus attack. In the attack, one can construct a PII association task, whereby an LLM is fine-tuned using a minuscule PII dataset, to potentially reinstate and reveal concealed PIIs. Our findings indicate that, with a trivial fine-tuning outlay, LLMs such as GPT-3.5 can transition from being impermeable to PII extraction to a state where they divulge a substantial proportion of concealed PII. This research, through its deep dive into the Janus attack vector, underscores the imperative of navigating the intricate interplay between LLM utility and privacy preservation.

In the wake of the burgeoning expansion of generative artificial intelligence (AI) services, the computational demands inherent to these technologies frequently necessitate cloud-powered computational offloading, particularly for resource-constrained mobile devices. These services commonly employ prompts to steer the generative process, and both the prompts and the resultant content, such as text and images, may harbor privacy-sensitive or confidential information, thereby elevating security and privacy risks. To mitigate these concerns, we introduce $\Lambda$-Split, a split computing framework to facilitate computational offloading while simultaneously fortifying data privacy against risks such as eavesdropping and unauthorized access. In $\Lambda$-Split, a generative model, usually a deep neural network (DNN), is partitioned into three sub-models and distributed across the user's local device and a cloud server: the input-side and output-side sub-models are allocated to the local, while the intermediate, computationally-intensive sub-model resides on the cloud server. This architecture ensures that only the hidden layer outputs are transmitted, thereby preventing the external transmission of privacy-sensitive raw input and output data. Given the black-box nature of DNNs, estimating the original input or output from intercepted hidden layer outputs poses a significant challenge for malicious eavesdroppers. Moreover, $\Lambda$-Split is orthogonal to traditional encryption-based security mechanisms, offering enhanced security when deployed in conjunction. We empirically validate the efficacy of the $\Lambda$-Split framework using Llama 2 and Stable Diffusion XL, representative large language and diffusion models developed by Meta and Stability AI, respectively. Our $\Lambda$-Split implementation is publicly accessible at https://github.com/nishio-laboratory/lambda_split.

Thanks to their generative capabilities, large language models (LLMs) have become an invaluable tool for creative processes. These models have the capacity to produce hundreds and thousands of visual and textual outputs, offering abundant inspiration for creative endeavors. But are we harnessing their full potential? We argue that current interaction paradigms fall short, guiding users towards rapid convergence on a limited set of ideas, rather than empowering them to explore the vast latent design space in generative models. To address this limitation, we propose a framework that facilitates the structured generation of design space in which users can seamlessly explore, evaluate, and synthesize a multitude of responses. We demonstrate the feasibility and usefulness of this framework through the design and development of an interactive system, Luminate, and a user study with 8 professional writers. Our work advances how we interact with LLMs for creative tasks, introducing a way to harness the creative potential of LLMs.

The rapid proliferation of ChatGPT has incited debates regarding its impact on human writing. Amid concerns about declining writing standards, this study investigates the role of ChatGPT in facilitating academic writing, especially among language learners. Using a case study approach, this study examines the experiences of Kailing, a doctoral student, who integrates ChatGPT throughout their academic writing process. The study employs activity theory as a lens for understanding writing with generative AI tools and data analyzed includes semi-structured interviews, writing samples, and GPT logs. Results indicate that Kailing effectively collaborates with ChatGPT across various writing stages while preserving her distinct authorial voice and agency. This underscores the potential of AI tools such as ChatGPT to enhance academic writing for language learners without overshadowing individual authenticity. This case study offers a critical exploration of how ChatGPT is utilized in the academic writing process and the preservation of a student's authentic voice when engaging with the tool.

Generative AI models have shown impressive performance on many Natural Language Processing tasks such as language understanding, reasoning, and language generation. An important question being asked by the AI community today is about the capabilities and limits of these models, and it is clear that evaluating generative AI is very challenging. Most studies on generative LLMs have been restricted to English and it is unclear how capable these models are at understanding and generating text in other languages. We present the first comprehensive benchmarking of generative LLMs - MEGA, which evaluates models on standard NLP benchmarks, covering 16 NLP datasets across 70 typologically diverse languages. We compare the performance of generative LLMs including Chat-GPT and GPT-4 to State of the Art (SOTA) non-autoregressive models on these tasks to determine how well generative models perform compared to the previous generation of LLMs. We present a thorough analysis of the performance of models across languages and tasks and discuss challenges in improving the performance of generative LLMs on low-resource languages. We create a framework for evaluating generative LLMs in the multilingual setting and provide directions for future progress in the field.

In January 2020, privacy journalist Kashmir Hill published an article in The New York Times describing Clearview AI--a company that purports to help U.S. law enforcement match photos of unknown people to their online presence through a facial recognition model trained by scraping millions of publicly available face images online.a In 2021, police departments in many different U.S. cities were reported to have used Clearview AI to, for example, identify Black Lives Matter protestors.b In 2022, a California-based artist found that photos she thought to be in her private medical record were included, without her knowledge or consent, in the LAION training dataset that has been used to train Stable Diffusion and Google Imagen.c The artist has a rare medical condition she prefers to keep private and expressed concern about the abuse potential of generative AI technologies having access to her photos. In January 2023, Twitch streamer QTCinderella made an emphatic plea to her followers on Twitter to stop spreading links to an illicit website hosting AI-generated "deep fake" pornography of her and other women influencers.

DALL-E, Midjourney, and Stable Diffusion are among the generative AI technologies widely used to produce images in response to user prompts. The output images are, for the most part, indistinguishable from images humans might have created. Generative AI systems are capable of producing human-creator-like images because of the extremely large quantities of images, paired with textual descriptions of the images' contents, on which the systems' image models were trained. A text prompt to compose a picture of a dog playing with a ball on a beach at sunset will generate a responsive image drawing upon embedded representations of how dogs, balls, beaches, and sunsets are typically depicted and arranged in images of this sort.

Until now, the only free and easy part had been the data. Widely used services like the nonprofit Common Crawl charge Google, Meta, OpenAI and others nothing to use its service, which crawls the internet in search of troves of online text and archives the information for others to download. To assemble the vast quantities of natural language and specialized information needed to train large AI systems, tech companies have combined those archives with online data sets, accessing information made available for research purposes, and increasingly straying from information clearly in the public domain.

Multimodal datasets are a critical component in recent breakthroughs such as Stable Diffusion and GPT-4, yet their design does not receive the same research attention as model architectures or training algorithms. To address this shortcoming in the ML ecosystem, we introduce DataComp, a testbed for dataset experiments centered around a new candidate pool of 12.8 billion image-text pairs from Common Crawl. Participants in our benchmark design new filtering techniques or curate new data sources and then evaluate their new dataset by running our standardized CLIP training code and testing the resulting model on 38 downstream test sets. Our benchmark consists of multiple compute scales spanning four orders of magnitude, which enables the study of scaling trends and makes the benchmark accessible to researchers with varying resources. Our baseline experiments show that the DataComp workflow leads to better training sets. In particular, our best baseline, DataComp-1B, enables training a CLIP ViT-L/14 from scratch to 79.2% zero-shot accuracy on ImageNet, outperforming OpenAI's CLIP ViT-L/14 by 3.7 percentage points while using the same training procedure and compute. We release DataComp and all accompanying code at www.datacomp.ai.