Decoding Biases: Automated Methods and LLM Judges for Gender Bias Detection in Language Models

Large Language Models (LLMs) have excelled at language understanding and generating human-level text. However, even with supervised training and human alignment, these LLMs are susceptible to adversarial attacks where malicious users can prompt the model to generate undesirable text. LLMs also inherently encode potential biases that can cause various harmful effects during interactions. Bias evaluation metrics lack standards as well as consensus and existing methods often rely on human-generated templates and annotations which are expensive and labor intensive. In this work, we train models to automatically create adversarial prompts to elicit biased responses from target LLMs. We present LLM- based bias evaluation metrics and also analyze several existing automatic evaluation methods and metrics. We analyze the various nuances of model responses, identify the strengths and weaknesses of model families, and assess where evaluation methods fall short. We compare these metrics to human evaluation and validate that the LLM-as-a-Judge metric aligns with human judgement on bias in response generation.

Learning from Red Teaming: Gender Bias Provocation and Mitigation in Large Language Models

Recently, researchers have made considerable improvements in dialogue systems with the progress of large language models (LLMs) such as ChatGPT and GPT-4. These LLM-based chatbots encode the potential biases while retaining disparities that can harm humans during interactions. The traditional biases investigation methods often rely on human-written test cases. However, these test cases are usually expensive and limited. In this work, we propose a first-of-its-kind method that automatically generates test cases to detect LLMs' potential gender bias. We apply our method to three well-known LLMs and find that the generated test cases effectively identify the presence of biases. To address the biases identified, we propose a mitigation strategy that uses the generated test cases as demonstrations for in-context learning to circumvent the need for parameter fine-tuning. The experimental results show that LLMs generate fairer responses with the proposed approach.

Position Matters! Empirical Study of Order Effect in Knowledge-grounded Dialogue

With the power of large pretrained language models, various research works have integrated knowledge into dialogue systems. The traditional techniques treat knowledge as part of the input sequence for the dialogue system, prepending a set of knowledge statements in front of dialogue history. However, such a mechanism forces knowledge sets to be concatenated in an ordered manner, making models implicitly pay imbalanced attention to the sets during training. In this paper, we first investigate how the order of the knowledge set can influence autoregressive dialogue systems' responses. We conduct experiments on two commonly used dialogue datasets with two types of transformer-based models and find that models view the input knowledge unequally. To this end, we propose a simple and novel technique to alleviate the order effect by modifying the position embeddings of knowledge input in these models. With the proposed position embedding method, the experimental results show that each knowledge statement is uniformly considered to generate responses.

Controllable Response Generation for Assistive Use-cases

Conversational agents have become an integral part of the general population for simple task enabling situations. However, these systems are yet to have any social impact on the diverse and minority population, for example, helping people with neurological disorders, for example ALS, and people with speech, language and social communication disorders. Language model technology can play a huge role to help these users carry out daily communication and social interactions. To enable this population, we build a dialog system that can be controlled by users using cues or keywords. We build models that can suggest relevant cues in the dialog response context which is used to control response generation and can speed up communication. We also introduce a keyword loss to lexically constrain the model output. We show both qualitatively and quantitatively that our models can effectively induce the keyword into the model response without degrading the quality of response. In the context of usage of such systems for people with degenerative disorders, we present human evaluation of our cue or keyword predictor and the controllable dialog system and show that our models perform significantly better than models without control. Our study shows that keyword control on end to end response generation models is powerful and can enable and empower users with degenerative disorders to carry out their day to day communication.

Audio-Visual Understanding of Passenger Intents for In-Cabin Conversational Agents

Building multimodal dialogue understanding capabilities situated in the in-cabin context is crucial to enhance passenger comfort in autonomous vehicle (AV) interaction systems. To this end, understanding passenger intents from spoken interactions and vehicle vision systems is a crucial component for developing contextual and visually grounded conversational agents for AV. Towards this goal, we explore AMIE (Automated-vehicle Multimodal In-cabin Experience), the in-cabin agent responsible for handling multimodal passenger-vehicle interactions. In this work, we discuss the benefits of a multimodal understanding of in-cabin utterances by incorporating verbal/language input together with the non-verbal/acoustic and visual clues from inside and outside the vehicle. Our experimental results outperformed text-only baselines as we achieved improved performances for intent detection with a multimodal approach.

Low Rank Fusion based Transformers for Multimodal Sequences

Our senses individually work in a coordinated fashion to express our emotional intentions. In this work, we experiment with modeling modality-specific sensory signals to attend to our latent multimodal emotional intentions and vice versa expressed via low-rank multimodal fusion and multimodal transformers. The low-rank factorization of multimodal fusion amongst the modalities helps represent approximate multiplicative latent signal interactions. Motivated by the work of~\cite{tsai2019MULT} and~\cite{Liu_2018}, we present our transformer-based cross-fusion architecture without any over-parameterization of the model. The low-rank fusion helps represent the latent signal interactions while the modality-specific attention helps focus on relevant parts of the signal. We present two methods for the Multimodal Sentiment and Emotion Recognition results on CMU-MOSEI, CMU-MOSI, and IEMOCAP datasets and show that our models have lesser parameters, train faster and perform comparably to many larger fusion-based architectures.

Exploring Context, Attention and Audio Features for Audio Visual Scene-Aware Dialog

We are witnessing a confluence of vision, speech and dialog system technologies that are enabling the IVAs to learn audio-visual groundings of utterances and have conversations with users about the objects, activities and events surrounding them. Recent progress in visual grounding techniques and Audio Understanding are enabling machines to understand shared semantic concepts and listen to the various sensory events in the environment. With audio and visual grounding methods, end-to-end multimodal SDS are trained to meaningfully communicate with us in natural language about the real dynamic audio-visual sensory world around us. In this work, we explore the role of `topics' as the context of the conversation along with multimodal attention into such an end-to-end audio-visual scene-aware dialog system architecture. We also incorporate an end-to-end audio classification ConvNet, AclNet, into our models. We develop and test our approaches on the Audio Visual Scene-Aware Dialog (AVSD) dataset released as a part of the DSTC7. We present the analysis of our experiments and show that some of our model variations outperform the baseline system released for AVSD.

Leveraging Topics and Audio Features with Multimodal Attention for Audio Visual Scene-Aware Dialog

With the recent advancements in Artificial Intelligence (AI), Intelligent Virtual Assistants (IVA) such as Alexa, Google Home, etc., have become a ubiquitous part of many homes. Currently, such IVAs are mostly audio-based, but going forward, we are witnessing a confluence of vision, speech and dialog system technologies that are enabling the IVAs to learn audio-visual groundings of utterances. This will enable agents to have conversations with users about the objects, activities and events surrounding them. In this work, we present three main architectural explorations for the Audio Visual Scene-Aware Dialog (AVSD): 1) investigating `topics' of the dialog as an important contextual feature for the conversation, 2) exploring several multimodal attention mechanisms during response generation, 3) incorporating an end-to-end audio classification ConvNet, AclNet, into our architecture. We discuss detailed analysis of the experimental results and show that our model variations outperform the baseline system presented for the AVSD task.

Modeling Intent, Dialog Policies and Response Adaptation for Goal-Oriented Interactions

Building a machine learning driven spoken dialog system for goal-oriented interactions involves careful design of intents and data collection along with development of intent recognition models and dialog policy learning algorithms. The models should be robust enough to handle various user distractions during the interaction flow and should steer the user back into an engaging interaction for successful completion of the interaction. In this work, we have designed a goal-oriented interaction system where children can engage with agents for a series of interactions involving `Meet \& Greet' and `Simon Says' game play. We have explored various feature extractors and models for improved intent recognition and looked at leveraging previous user and system interactions in novel ways with attention models. We have also looked at dialog adaptation methods for entrained response selection. Our bootstrapped models from limited training data perform better than many baseline approaches we have looked at for intent recognition and dialog action prediction.

Towards Multimodal Understanding of Passenger-Vehicle Interactions in Autonomous Vehicles: Intent/Slot Recognition Utilizing Audio-Visual Data

Understanding passenger intents from spoken interactions and car's vision (both inside and outside the vehicle) are important building blocks towards developing contextual dialog systems for natural interactions in autonomous vehicles (AV). In this study, we continued exploring AMIE (Automated-vehicle Multimodal In-cabin Experience), the in-cabin agent responsible for handling certain multimodal passenger-vehicle interactions. When the passengers give instructions to AMIE, the agent should parse such commands properly considering available three modalities (language/text, audio, video) and trigger the appropriate functionality of the AV system. We had collected a multimodal in-cabin dataset with multi-turn dialogues between the passengers and AMIE using a Wizard-of-Oz scheme via realistic scavenger hunt game. In our previous explorations, we experimented with various RNN-based models to detect utterance-level intents (set destination, change route, go faster, go slower, stop, park, pull over, drop off, open door, and others) along with intent keywords and relevant slots (location, position/direction, object, gesture/gaze, time-guidance, person) associated with the action to be performed in our AV scenarios. In this recent work, we propose to discuss the benefits of multimodal understanding of in-cabin utterances by incorporating verbal/language input (text and speech embeddings) together with the non-verbal/acoustic and visual input from inside and outside the vehicle (i.e., passenger gestures and gaze from in-cabin video stream, referred objects outside of the vehicle from the road view camera stream). Our experimental results outperformed text-only baselines and with multimodality, we achieved improved performances for utterance-level intent detection and slot filling.