Recent advancements in large language models (LLMs) have facilitated the development of chatbots with sophisticated conversational capabilities. However, LLMs exhibit frequent inaccurate responses to queries, hindering applications in educational settings. In this paper, we investigate the effectiveness of integrating a knowledge base (KB) with LLM intelligent tutors to increase response reliability. To achieve this, we design a scaleable KB that affords educational supervisors seamless integration of lesson curricula, which is automatically processed by the intelligent tutoring system. We then detail an evaluation, where student participants were presented with questions about the artificial intelligence curriculum to respond to. GPT-4 intelligent tutors with varying hierarchies of KB access and human domain experts then assessed these responses. Lastly, students cross-examined the intelligent tutors' responses to the domain experts' and ranked their various pedagogical abilities. Results suggest that, although these intelligent tutors still demonstrate a lower accuracy compared to domain experts, the accuracy of the intelligent tutors increases when access to a KB is granted. We also observe that the intelligent tutors with KB access exhibit better pedagogical abilities to speak like a teacher and understand students than those of domain experts, while their ability to help students remains lagging behind domain experts.

Large language models are quickly gaining momentum, yet are found to demonstrate gender bias in their responses. In this paper, we conducted a content analysis of social media discussions to gauge public perceptions of gender bias in LLMs which are trained in different cultural contexts, i.e., ChatGPT, a US-based LLM, or Ernie, a China-based LLM. People shared both observations of gender bias in their personal use and scientific findings about gender bias in LLMs. A difference between the two LLMs was seen -- ChatGPT was more often found to carry implicit gender bias, e.g., associating men and women with different profession titles, while explicit gender bias was found in Ernie's responses, e.g., overly promoting women's pursuit of marriage over career. Based on the findings, we reflect on the impact of culture on gender bias and propose governance recommendations to regulate gender bias in LLMs.

During the last stage of RLHF, a large language model is aligned to human intents via PPO training, a process that generally requires large-scale computational resources. In this technical report, we empirically investigate an efficient implementation of RLHF using low-rank adaptation (LoRA), which allows us to align the LLaMA 7B checkpoint on the Alpaca dataset (Taori et al., 2023) using only two A100 GPUs instead of the eight required for full model fine-tuning. Despite tuning only 0.2% of LLaMA 7B's parameters, our implementation achieves better performance than the publicly-released AlpacaFarm checkpoint (Dubois et al., 2023) with full model fine-tuning. Next, we analyze several configurations of our LoRA-based PPO implementation, varying the form of the KL regularization term in the training objective. We find that (1) removing this penalty term does not harm performance on the AlpacaFarm evaluation set under our LoRA setup; (2) other regularizers, such as Jensen-Shannon divergence, lead to improved performance; and (3) while PPO training negatively impacts the factuality of model-generated responses, training with LoRA largely mitigates this effect. We release our code and pretrained checkpoints to facilitate future research on more efficient RLHF.

This paper delves into the transformative power of Generative AI-driven storytelling in the realm of marketing. Generative AI, distinct from traditional machine learning, offers the capability to craft narratives that resonate with consumers on a deeply personal level. Through real-world examples from industry leaders like Google, Netflix and Stitch Fix, we elucidate how this technology shapes marketing strategies, personalizes consumer experiences, and navigates the challenges it presents. The paper also explores future directions and recommendations for generative AI-driven storytelling, including prospective applications such as real-time personalized storytelling, immersive storytelling experiences, and social media storytelling. By shedding light on the potential and impact of generative AI-driven storytelling in marketing, this paper contributes to the understanding of this cutting-edge approach and its transformative power in the field of marketing.

LLMs are increasingly powerful and widely used to assist users in a variety of tasks. This use risks the introduction of LLM biases to consequential decisions such as job hiring, human performance evaluation, and criminal sentencing. Bias in NLP systems along the lines of gender and ethnicity has been widely studied, especially for specific stereotypes (e.g., Asians are good at math). In this paper, we investigate bias along less studied, but still consequential, dimensions, such as age and beauty, measuring subtler correlated decisions that LLMs (specially autoregressive language models) make between social groups and unrelated positive and negative attributes. We ask whether LLMs hold wide-reaching biases of positive or negative sentiment for specific social groups similar to the ``what is beautiful is good'' bias found in people in experimental psychology. We introduce a template-generated dataset of sentence completion tasks that asks the model to select the most appropriate attribute to complete an evaluative statement about a person described as a member of a specific social group. We also reverse the completion task to select the social group based on an attribute. Finally, we report the correlations that we find for multiple cutting-edge LLMs. This dataset can be used as a benchmark to evaluate progress in more generalized biases and the templating technique can be used to expand the benchmark with minimal additional human annotation.

ChatGPT-4 with Code Interpreter can be used to solve introductory college-level vector calculus and electromagnetism problems Tanuj Kumar (tanuj.kumar@wisc.edu) Executive summary: We evaluated three modes of ChatGPT -- 3.5, 4, and 4 with Code Interpreter -- on a set of college-level engineering-math and electromagnetism problems, such as those often given to sophomore electrical engineering majors. We selected a set of 13 problems without first testing them with ChatGPT, and had ChatGPT solve them multiple times, using a fresh instance (chat) of ChatGPT each time. The problems range from elementary to medium-level. We were strict in our evaluation of ChatGPT's performance, marking a solution as incorrect if even a small part of the solution was wrong. Our major conclusions are: ChatGPT-4 with Code Interpreter (ChatGPT-4/CI), recently renamed Advanced Data Analysis, was able to satisfactorily solve most problems we tested most of the time. Qualitatively, one could give ChatGPT-4/CI a solid passing grade in introductory engineering math and electromagnetics.

We introduce bipol, a new metric with explainability, for estimating social bias in text data. Harmful bias is prevalent in many online sources of data that are used for training machine learning (ML) models. In a step to address this challenge we create a novel metric that involves a two-step process: corpus-level evaluation based on model classification and sentence-level evaluation based on (sensitive) term frequency (TF). After creating new models to detect bias along multiple axes using SotA architectures, we evaluate two popular NLP datasets (COPA and SQUAD). As additional contribution, we created a large dataset (with almost 2 million labelled samples) for training models in bias detection and make it publicly available. We also make public our codes.

Non-stationary online learning has drawn much attention in recent years. In particular, dynamic regret and adaptive regret are proposed as two principled performance measures for online convex optimization in non-stationary environments. To optimize them, a two-layer online ensemble is usually deployed due to the inherent uncertainty of the non-stationarity, in which a group of base-learners are maintained and a meta-algorithm is employed to track the best one on the fly. However, the two-layer structure raises the concern about the computational complexity -- those methods typically maintain $\mathcal{O}(\log T)$ base-learners simultaneously for a $T$-round online game and thus perform multiple projections onto the feasible domain per round, which becomes the computational bottleneck when the domain is complicated. In this paper, we present efficient methods for optimizing dynamic regret and adaptive regret, which reduce the number of projections per round from $\mathcal{O}(\log T)$ to $1$. Moreover, our obtained algorithms require only one gradient query and one function evaluation at each round. Our technique hinges on the reduction mechanism developed in parameter-free online learning and requires non-trivial twists on non-stationary online methods. Empirical studies verify our theoretical findings.

Tim Ballaret once dreamed of becoming a stockbroker but ultimately found fulfillment helping high school students in south Los Angeles understand the relevance of math and science to their daily lives. But making engaging class materials is time-consuming, so this spring he started experimenting with generative AI tools. Recommendations by friends and influential teachers on social media led Ballaret to try MagicSchool, a tool for K-12 educators powered by OpenAI's text generation algorithms. He used it for tasks like creating math word problems that match his students' interests, like Taylor Swift and Minecraft, but the real test came when he used MagicSchool this summer to outline a year's worth of lesson plans for a new applied science and engineering class. "Taking back my summer helped me be more refreshed for a new school year," he says.

Large Language Models (LLMs) represent the recent success of deep learning in achieving remarkable human-like predictive performance. It has become a mainstream strategy to leverage fine-tuning to adapt LLMs for various real-world applications due to the prohibitive expenses associated with LLM training. The learning rate is one of the most important hyperparameters in LLM fine-tuning with direct impacts on both fine-tuning efficiency and fine-tuned LLM quality. Existing learning rate policies are primarily designed for training traditional deep neural networks (DNNs), which may not work well for LLM fine-tuning. We reassess the research challenges and opportunities of learning rate tuning in the coming era of Large Language Models. This paper makes three original contributions. First, we revisit existing learning rate policies to analyze the critical challenges of learning rate tuning in the era of LLMs. Second, we present LRBench++ to benchmark learning rate policies and facilitate learning rate tuning for both traditional DNNs and LLMs. Third, our experimental analysis with LRBench++ demonstrates the key differences between LLM fine-tuning and traditional DNN training and validates our analysis.