Existing task-oriented conversational search systems heavily rely on domain ontologies with pre-defined slots and candidate value sets. In practical applications, these prerequisites are hard to meet, due to the emerging new user requirements and ever-changing scenarios. To mitigate these issues for better interaction performance, there are efforts working towards detecting out-of-vocabulary values or discovering new slots under unsupervised or semi-supervised learning paradigm. However, overemphasizing on the conversation data patterns alone induces these methods to yield noisy and arbitrary slot results. To facilitate the pragmatic utility, real-world systems tend to provide a stringent amount of human labelling quota, which offers an authoritative way to obtain accurate and meaningful slot assignments. Nonetheless, it also brings forward the high requirement of utilizing such quota efficiently. Hence, we formulate a general new slot discovery task in an information extraction fashion and incorporate it into an active learning framework to realize human-in-the-loop learning. Specifically, we leverage existing language tools to extract value candidates where the corresponding labels are further leveraged as weak supervision signals. Based on these, we propose a bi-criteria selection scheme which incorporates two major strategies, namely, uncertainty-based sampling and diversity-based sampling to efficiently identify terms of interest. We conduct extensive experiments on several public datasets and compare with a bunch of competitive baselines to demonstrate the effectiveness of our method. We have made the code and data used in this paper publicly available.

The recent release of AI technology that generates new text has raised serious questions among the research community. For one, "Can ChatGPT be named an author of a research paper?" The resounding answer from arXiv leaders and advisors is, "No." A computer program cannot, for example, take responsibility for the contents of a paper. Nor can it agree to arXiv's terms and conditions.

Large language models (LMs) provide novel opportunities for interface design. Large language models have undoubtedly advanced to the point where they may be compared to a genuine writer. The models do an excellent job of comprehending the subject matter. Recent LMs (such as GPT-2 and GPT-3) can create a wide range of prose and conversations with unrivaled fluency. These models may be fine-tuned to become more skilled at specific activities, such as email composition or health consultations.

Artificial intelligence (AI), like robotics, has long been seen as "future technologies". However, much as with robots, we can now affirm that AI is not just science fiction, but much more than that. AI is very much alive in our personal and professional lives, and it is swiftly catching up to mobile devices in terms of popularity. There is not a single activity in our daily activities, the use of AI is not impacting us. From Alexa, Siri to self-driving cars, AI is stepping up to assist us just like a human would.

From winning Jeopardy in 2011 to helping write a sad song last year, IBM's Watson cognitive computing platform is all over popular culture. Press releases fly out about Watson producing a movie trailer, powering a Macy's shopping app, even controlling lights on an internet-connected dress--along with more serious applications like working on cancer treatments. It seems, from IBM's hype, that Watson can do everything. But Bernie Meyerson, IBM's chief innovation officer, wants to dial back the hype in some ways, calling Watson "just the first step on a very, very long road." Watson can be helpful in a lot of industries, such as medicine, which are awash in data, but it can't replace people, he says.