Building datasets for dialogue tasks is expensive and time-consuming, requiring recruitment, training, and data collection from study participants. In response, much recent work has sought to use large language models (LLMs) to simulate both human-human and human-LLM interactions, as they have been shown to generate convincingly human-like text in many settings. However, how well do LLM-based simulations reflect real human dialogue? In this work, we answer this question by generating a large-scale dataset of 100,000 paired LLM-LLM and human-LLM dialogues from the WildChat dataset and quantifying how well the LLM simulations align with their human counterparts. Overall, we find relatively low alignment between simulations and human interactions, with systematic differences in multiple textual properties, including style and conversational dynamics. Further, we find that models perform similarly in simulating English, Chinese, and Russian dialogues. Our results also suggest that LLMs only simulate a narrow range of the overall distribution of human dialogue, as they perform better on the subset of humans who write similarly to the LLM’s own style.

Large Language Models are expressive tools that enable complex tasks of text understanding within Computational Social Science. Their versatility, while beneficial, poses a barrier for establishing standardized best practices within the field. To bring clarity on the values of different strategies, we present an overview of the performance of modern LLM-based classification methods on a benchmark of 23 social knowledge tasks. Our results point to three best practices: prioritize models with larger vocabulary and pre-training corpora; avoid simple zero-shot in favor of AI-enhanced prompting; fine-tune on task-specific data, and consider more complex forms instruction-tuning on multiple datasets only when only training data is more abundant.

In the realm of Computational Social Science (CSS), practitioners often navigate complex, low-resource domains and face the costly and time-intensive challenges of acquiring and annotating data. We aim to establish a set of guidelines to address such challenges, comparing the use of human-labeled data with synthetically generated data from GPT-4 and Llama-2 in ten distinct CSS classification tasks of varying complexity. Additionally, we examine the impact of training data sizes on performance. Our findings reveal that models trained on human-labeled data consistently exhibit superior or comparable performance compared to their synthetically augmented counterparts. Nevertheless, synthetic augmentation proves beneficial, particularly in improving performance on rare classes within multi-class tasks. Furthermore, we leverage GPT-4 and Llama-2 for zero-shot classification and find that, while they generally display strong performance, they often fall short when compared to specialized classifiers trained on moderately sized training sets.

With the COVID-19 pandemic raging world-wide since the beginning of the 2020 decade, the need for monitoring systems to track relevant information on social media is vitally important. This paper describes our submission to the WNUT-2020 Task 2: Identification of informative COVID-19 English Tweets. We investigate the effectiveness for a variety of classification models, and found that domain-specific pre-trained BERT models lead to the best performance. On top of this, we attempt a variety of ensembling strategies, but these attempts did not lead to further improvements. Our final best model, the standalone CT-BERT model, proved to be highly competitive, leading to a shared first place in the shared task. Our results emphasize the importance of domain and task-related pre-training.