An official website of the United States government
The difficulty in specifying rewards for many real world problems has led to an increased focus on learning rewards from human feedback, such as demonstrations. However, there are often many different reward functions that explain the human feedback, leaving agents with uncertainty over what the true reward function is. While most policy optimization approaches handle this uncertainty by optimizing for expected performance, many applications demand risk-averse behavior. We derive a novel policy gradient-style robust optimization approach, PG-BROIL, that optimizes a soft-robust objective that balances expected performance and risk. To the best of our knowledge, PG-BROIL is the first policy optimization algorithm robust to a distribution of reward hypotheses which can scale to continuous MDPs. Results suggest that PG-BROIL can produce a family of behaviors ranging from risk-neutral to risk-averse and outperforms state-of-the-art imitation learning algorithms when learning from ambiguous demonstrations by hedging against uncertainty, rather than seeking to uniquely identify the demonstrator’s reward function.
more »
« less
This content will become publicly available on April 11, 2026
Leveraging Human Input to Enable Robust, Interactive, and Aligned AI Systems
Ensuring that AI systems do what we, as humans, actually want them to do, is one of the biggest open research challenges in AI alignment and safety. My research seeks to directly address this challenge by enabling AI systems to interact with humans to learn aligned and robust behaviors. The way in which robots and other AI systems behave is often the result of optimizing a reward function. However, manually designing good reward functions is highly challenging and error prone, even for domain experts. Consider trying to write down a reward function that describes good driving behavior or how you like your bed made in the morning. While reward functions for these tasks are difficult to manually specify, human feedback in the form of demonstrations or preferences are often much easier to obtain. However, human data is often difficult to interpret, due to ambiguity and noise. Thus, it is critical that AI systems take into account epistemic uncertainty over the human's true intent. My talk will give an overview of my lab's progress along the following fundamental research areas: (1) efficiently maintaining uncertainty over human intent, (2) directly optimizing behavior to be robust to uncertainty over human intent, and (3) actively querying for additional human input to reduce uncertainty over human intent.
more »
« less
- Award ID(s):
- 2416761
- PAR ID:
- 10608128
- Publisher / Repository:
- Association for the Advancement of Artificial Intelligence
- Date Published:
- Journal Name:
- Proceedings of the AAAI Conference on Artificial Intelligence
- Volume:
- 39
- Issue:
- 27
- ISSN:
- 2159-5399
- Page Range / eLocation ID:
- 28704 to 28704
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Reinforcement Learning from Human Feedback (RLHF) has emerged as a popular paradigm for capturing human intent to alleviate the challenges of hand-crafting the reward values. Despite the increasing interest in RLHF, most works learn black box reward functions that while expressive are difficult to interpret and often require running the whole costly process of RL before we can even decipher if these frameworks are actually aligned with human preferences. We propose and evaluate a novel approach for learning expressive and interpretable reward functions from preferences using Differentiable Decision Trees (DDTs). Our experiments across several domains, including CartPole, Visual Gridworld environments and Atari games, provide evidence that the tree structure of our learned reward function is useful in determining the extent to which the reward function is aligned with human preferences. We also provide experimental evidence that not only shows that reward DDTs can often achieve competitive RL performance when compared with larger capacity deep neural network reward functions but also demonstrates the diagnostic utility of our framework in checking alignment of learned reward functions. We also observe that the choice between soft and hard (argmax) output of reward DDT reveals a tension between wanting highly shaped rewards to ensure good RL performance, while also wanting simpler, more interpretable rewards. Videos and code, are available at: https://sites.google.com/view/ddt-rlhfmore » « less
-
Keathley, H.; Enos, J.; Parrish, M. (Ed.)The role of human-machine teams in society is increasing, as big data and computing power explode. One popular approach to AI is deep learning, which is useful for classification, feature identification, and predictive modeling. However, deep learning models often suffer from inadequate transparency and poor explainability. One aspect of human systems integration is the design of interfaces that support human decision-making. AI models have multiple types of uncertainty embedded, which may be difficult for users to understand. Humans that use these tools need to understand how much they should trust the AI. This study evaluates one simple approach for communicating uncertainty, a visual confidence bar ranging from 0-100%. We perform a human-subject online experiment using an existing image recognition deep learning model to test the effect of (1) providing single vs. multiple recommendations from the AI and (2) including uncertainty information. For each image, participants described the subject in an open textbox and rated their confidence in their answers. Performance was evaluated at four levels of accuracy ranging from the same as the image label to the correct category of the image. The results suggest that AI recommendations increase accuracy, even if the human and AI have different definitions of accuracy. In addition, providing multiple ranked recommendations, with or without the confidence bar, increases operator confidence and reduces perceived task difficulty. More research is needed to determine how people approach uncertain information from an AI system and develop effective visualizations for communicating uncertainty.more » « less
-
Recent advances in AI models have increased the integration of AI-based decision aids into the human decision making process. To fully unlock the potential of AI- assisted decision making, researchers have computationally modeled how humans incorporate AI recommendations into their final decisions, and utilized these models to improve human-AI team performance. Meanwhile, due to the “black-box” nature of AI models, providing AI explanations to human decision makers to help them rely on AI recommendations more appropriately has become a common practice. In this paper, we explore whether we can quantitatively model how humans integrate both AI recommendations and explanations into their decision process, and whether this quantitative understanding of human behavior from the learned model can be utilized to manipulate AI explanations, thereby nudging individuals towards making targeted decisions. Our extensive human experiments across various tasks demonstrate that human behavior can be easily influenced by these manipulated explanations towards targeted outcomes, regardless of the intent being adversarial or benign. Furthermore, individuals often fail to detect any anomalies in these explanations, despite their decisions being affected by them.more » « less
-
Recent advances in AI models have increased the integration of AI-based decision aids into the human decision making process. To fully unlock the potential of AI-assisted decision making, researchers have computationally modeled how humans incorporate AI recommendations into their final decisions, and utilized these models to improve human-AI team performance. Meanwhile, due to the black-box'' nature of AI models, providing AI explanations to human decision makers to help them rely on AI recommendations more appropriately has become a common practice. In this paper, we explore whether we can quantitatively model how humans integrate both AI recommendations and explanations into their decision process, and whether this quantitative understanding of human behavior from the learned model can be utilized to manipulate AI explanations, thereby nudging individuals towards making targeted decisions. Our extensive human experiments across various tasks demonstrate that human behavior can be easily influenced by these manipulated explanations towards targeted outcomes, regardless of the intent being adversarial or benign. Furthermore, individuals often fail to detect any anomalies in these explanations, despite their decisions being affected by them.more » « less
