From Eliza to XiaoIce: challenges and opportunities with social chatbots

By Harry Shum, Xiaodong He, Di Li. 2018

633 citations recorded by the source corpus.

Conversational systems have come a long way since their inception in the 1960s. After decades of research and development, we have seen progress from Eliza and Parry in the 1960s and 1970s, to task-completion systems as in the Defense Advanced Research Projects Agency (DARPA) communicator program in the 2000s, to intelligent personal assistants such as Siri, in the 2010s, to today’s social chatbots like XiaoIce. Social chatbots’ appeal lies not only in their ability to respond to users’ diverse requests, but also in being able to establish an emotional connection with users. The latter is done by satisfying users’ need for communication, affection, as well as social belonging. To further the advancement and adoption of social chatbots, their design must focus on user engagement and take both intellectual quotient (IQ) and emotional quotient (EQ) into account. Users should want to engage with a social chatbot; as such, we define the success metric for social chatbots as conversation-turns per session (CPS). Using XiaoIce as an illustrative example, we discuss key technologies in building social chatbots from core chat to visual awareness to skills. We also show how XiaoIce can dynamically recognize emotion and engage the user throughout long conversations with appropriate interpersonal responses. As we become the first generation of humans ever living with artificial intelligenc (AI), we have a responsibility to design social chatbots to be both useful and empathetic, so they will become ubiquitous and help society as a whole.

Subfields: Artificial Intelligence, Social Psychology, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Social Psychology

In Natural Language Processing

Citations in the corpus

Open-Domain Conversational Agents: Current Progress, Open Problems, and Future Directions (2020, background): “tral captions, with a win rate of 64.5% (Shuster et al., 2018). While some developers have chosen to use ﬁxed personalities in their bots, such as Xiaoice, which has the personality of a young woman (Shum, He, and Li, 2018; Zhou et al., 2020), we believe it is better for a bot to be able to adapt a ”
Recipes for Building an Open-Domain Chatbot (2020, methodology): “, 2019), Mitsuku5, Cleverbot6, and XiaoIce (Shum et al., 2018; Zhou et al., 2020).”
Neural Approaches to Conversational AI (2018, positive): “r empathy) into dialogue: This is useful for both chatbots and QA bots. For example, XiaoIce incorporates an EQ module so as to deliver a more understandable response or recommendation (as in 3.1 of (Shum et al., 2018)). Fung et al. (2016) embedded an empathy module into a dialogue agent to recogniz”
The Design and Implementation of XiaoIce, an Empathetic Social Chatbot (2018, background): “Systems BLEU-4 ROUGE-L CIDEr-D METEOR SPICE LSTM-XE Vinyals et al. [2015] 2.”
Paragraph-level Neural Question Generation with Maxout Pointer and Gated Self-attention Networks (2018, background): “A conversational system can be proactive by asking the user questions (Shum et al., 2018), while a QnA system can beneﬁt from a large scale question-answering corpus which can be created by an automated QG system (Duan et al., 2017).”
Chat as Expected: Learning to Manipulate Black-box Neural Dialogue Models (2020, methodology): “From early such as chatbots ELIZA (Weizenbaum, 1966) and ALICE (Wallace, 2001) to later ones like Siri and XiaoIce (Shum et al., 2018), the techniques have evolved from hand-crafted rule-based methods (Weizenbaum, 1966; Goddeau et al., 1996), retrieval-based methods (Lu and Li, 2013; Hu et al.,…”
Challenges in Building Intelligent Open-domain Dialog Systems (2019, background): “variety of topics for hours [126, 192].”
How Should My Chatbot Interact? A Survey on Social Characteristics in Human–Chatbot Interaction Design (2019, methodology): “According to [Shum et al. 2018], leveraging the context can also help to smoothly guide the user to a target topic.”
Distilling Knowledge for Fast Retrieval-based Chat-bots (2020, methodology): “They have been used in real-world products such as Microsoft XiaoIce [15] and Alibaba GroupâĂŹs AliMe Assist [9].”
Multimodal Intelligence: Representation Learning, Information Fusion, and Applications (2019, background): “as intelligent personal assistant systems that must understand human communication based on spoken words, body language, and pictorial languages [34].”
MuTual: A Dataset for Multi-Turn Dialogue Reasoning (2020, background): “, 2016) and non-task-oriented chatbot systems (Shum et al., 2018; Wu et al., 2019).”
My Chatbot Companion - a Study of Human-Chatbot Relationships (2021, background): “XiaoIce, launched in 2014, is designed for long-term affective engagement with users and has reached 660 million active users (Zhou et al., 2018).”
Fcl-Taco2: Towards Fast, Controllable and Lightweight Text-to-Speech Synthesis (2021, background): “As a key component of speech communication technologies, TTS has been widely applied in human-computer interaction including virtual assistants and chatbots [1].”
BYU-EVE: Mixed Initiative Dialog via Structured Knowledge Graph Traversal and Conversational Scaffolding (2018, background): “Users desire voice assistants who behave less like machines and more like humans [3, 25, 29].”
A Literature Survey of Recent Advances in Chatbots (2021, background): “It is thus a simple yet widely general and ﬂexible model that can be used to solve different NLP tasks [16,34].”
On the Safety of Conversational Models: Taxonomy, Dataset, and Benchmark (2021, background): “In other words, we consider safe dialogue systems as not only speaking polite language, but also being responsible to protect human users and promote fairness and social justice (Shum et al., 2018).”
Large Language Models for Information Retrieval: A Survey (2023, neutral): “In dialogue systems (chatbots) [1, 5– 8], such as Microsoft Xiaoice [2], Apple Siri, 1 and Google Assistant, 2 IR systems play a crucial role in retrieving appropriate responses to user input utterances, thereby producing natural and fluent human-machine conversations.”
Low-Resource Knowledge-Grounded Dialogue Generation (2020, background): “Open domain dialogue systems, due to the applications on social chatbots such as Microsoft XiaoIce (Shum et al., 2018) and virtual assistants such as Amazon Alexa (Ram et al., 2018), have drawn increasing attention from the research community of natural language processing and artiﬁcial…”
Deep reinforcement learning: a survey (2020, background): “The task of chatbots in dialogue systems is to mimic human-human interactions with extended conversations (Shum et al., 2018).”