End-to-end Task-Oriented Dialog (TOD) systems typically require extensive training datasets to perform well. In contrast, large language model (LLM) based TOD systems can excel even with limited data due to their ability to learn tasks through in-context exemplars. However, these models lack alignment with the style of responses in training data and often generate comprehensive responses, making it difficult for users to grasp the information quickly. In response, we propose SyncTOD that synergizes LLMs with task-specific hints to improve alignment in low-data settings. SyncTOD employs small auxiliary models to provide hints and select exemplars for in-context prompts. With ChatGPT, SyncTOD achieves superior performance compared to LLM-based baselines and SoTA models in low-data settings, while retaining competitive performance in full-data settings.

A humanised robot named Sophia who was created and programmed locally advised Hong Kong's leader on Wednesday on how to succeed with plans for a smart city. In a nine-minute dialogue with Chief Executive Carrie Lam Cheng Yuet-ngor at the American Chamber of Commerce's smart city forum in Wan Chai, Sophia anticipated "even greater things" in Hong Kong's blueprint for innovation as unveiled by the top official last December. Dressed in a dark blue sleeveless top and at times blinking her eyes and smiling gently, Sophia gave Lam a pat on the back when answering a question on which part of the blueprint was her favourite. "My favourite part, unsurprisingly, is initiatives to attract a venture capital fund to support entrepreneurship and promote technology to parks such as Science Park, which is my home base, and facilitate research and development for universities," Sophia said, as she addressed Lam as "chief executive". Though Hong Kong slipped one notch to second after the United States as the world's most competitive economy in a ranking last month, Lam said the city was striving to become smarter by boosting innovation and technology in areas such as transport, payment systems and government data-sharing.