The content that a recommender system (RS) shows to users influences them. Therefore, when choosing which recommender to deploy, one is implicitly also choosing to induce specific internal states in users. Even more, systems trained via long-horizon optimization will have direct incentives to manipulate users, e.g. shift their preferences so they are easier to satisfy. In this work we focus on induced preference shifts in users. We argue that - before deployment - system designers should: estimate the shifts a recommender would induce; evaluate whether such shifts would be undesirable; and even actively optimize to avoid problematic shifts. These steps involve two challenging ingredients: estimation requires anticipating how hypothetical policies would influence user preferences if deployed - we do this by using historical user interaction data to train predictive user model which implicitly contains their preference dynamics; evaluation and optimization additionally require metrics to assess whether such influences are manipulative or otherwise unwanted - we use the notion of "safe shifts", that define a trust region within which behavior is safe. In simulated experiments, we show that our learned preference dynamics model is effective in estimating user preferences and how they would respond to new recommenders. Additionally, we show that recommenders that optimize for staying in the trust region can avoid manipulative behaviors while still generating engagement.
Natural language is an intuitive and expressive way to communicate reward information to autonomous agents. It encompasses everything from concrete instructions to abstract descriptions of the world. Despite this, natural language is often challenging to learn from: it is difficult for machine learning methods to make appropriate inferences from such a wide range of input. This paper proposes a generalization of reward design as a unifying principle to ground linguistic communication: speakers choose utterances to maximize expected rewards from the listener's future behaviors. We first extend reward design to incorporate reasoning about unknown future states in a linear bandit setting. We then define a speaker model which chooses utterances according to this objective. Simulations show that short-horizon speakers (reasoning primarily about a single, known state) tend to use instructions, while long-horizon speakers (reasoning primarily about unknown, future states) tend to describe the reward function. We then define a pragmatic listener which performs inverse reward design by jointly inferring the speaker's latent horizon and rewards. Our findings suggest that this extension of reward design to linguistic communication, including the notion of a latent speaker horizon, is a promising direction for achieving more robust alignment outcomes from natural language supervision.
While modern policy optimization methods can do complex manipulation from sensory data, they struggle on problems with extended time horizons and multiple sub-goals. On the other hand, task and motion planning (TAMP) methods scale to long horizons but they are computationally expensive and need to precisely track world state. We propose a method that draws on the strength of both methods: we train a policy to imitate a TAMP solver's output. This produces a feed-forward policy that can accomplish multi-step tasks from sensory data. First, we build an asynchronous distributed TAMP solver that can produce supervision data fast enough for imitation learning. Then, we propose a hierarchical policy architecture that lets us use partially trained control policies to speed up the TAMP solver. In robotic manipulation tasks with 7-DoF joint control, the partially trained policies reduce the time needed for planning by a factor of up to 2.6. Among these tasks, we can learn a policy that solves the RoboSuite 4-object pick-place task 88% of the time from object pose observations and a policy that solves the RoboDesk 9-goal benchmark 79% of the time from RGB images (averaged across the 9 disparate tasks).
We describe cases where real recommender systems were modified in the service of various human values such as diversity, fairness, well-being, time well spent, and factual accuracy. From this we identify the current practice of values engineering: the creation of classifiers from human-created data with value-based labels. This has worked in practice for a variety of issues, but problems are addressed one at a time, and users and other stakeholders have seldom been involved. Instead, we look to AI alignment work for approaches that could learn complex values directly from stakeholders, and identify four major directions: useful measures of alignment, participatory design and operation, interactive value learning, and informed deliberative judgments.
AI systems often rely on two key components: a specified goal or reward function and an optimization algorithm to compute the optimal behavior for that goal. This approach is intended to provide value for a principal: the user on whose behalf the agent acts. The objectives given to these agents often refer to a partial specification of the principal's goals. We consider the cost of this incompleteness by analyzing a model of a principal and an agent in a resource constrained world where the $L$ attributes of the state correspond to different sources of utility for the principal. We assume that the reward function given to the agent only has support on $J < L$ attributes. The contributions of our paper are as follows: 1) we propose a novel model of an incomplete principal-agent problem from artificial intelligence; 2) we provide necessary and sufficient conditions under which indefinitely optimizing for any incomplete proxy objective leads to arbitrarily low overall utility; and 3) we show how modifying the setup to allow reward functions that reference the full state or allowing the principal to update the proxy objective over time can lead to higher utility solutions. The results in this paper argue that we should view the design of reward functions as an interactive and dynamic process and identifies a theoretical scenario where some degree of interactivity is desirable.
We introduce the concept of a multi-principal assistance game (MPAG), and circumvent an obstacle in social choice theory, Gibbard's theorem, by using a sufficiently collegial preference inference mechanism. In an MPAG, a single agent assists N human principals who may have widely different preferences. MPAGs generalize assistance games, also known as cooperative inverse reinforcement learning games. We analyze in particular a generalization of apprenticeship learning in which the humans first perform some work to obtain utility and demonstrate their preferences, and then the robot acts to further maximize the sum of human payoffs. We show in this setting that if the game is sufficiently collegial, i.e. if the humans are responsible for obtaining a sufficient fraction of the rewards through their own actions, then their preferences are straightforwardly revealed through their work. This revelation mechanism is non-dictatorial, does not limit the possible outcomes to two alternatives, and is dominant-strategy incentive-compatible.
Assistance games (also known as cooperative inverse reinforcement learning games) have been proposed as a model for beneficial AI, wherein a robotic agent must act on behalf of a human principal but is initially uncertain about the humans payoff function. This paper studies multi-principal assistance games, which cover the more general case in which the robot acts on behalf of N humans who may have widely differing payoffs. Impossibility theorems in social choice theory and voting theory can be applied to such games, suggesting that strategic behavior by the human principals may complicate the robots task in learning their payoffs. We analyze in particular a bandit apprentice game in which the humans act first to demonstrate their individual preferences for the arms and then the robot acts to maximize the sum of human payoffs. We explore the extent to which the cost of choosing suboptimal arms reduces the incentive to mislead, a form of natural mechanism design. In this context we propose a social choice method that uses shared control of a system to combine preference inference with social welfare optimization.
How can societies learn to enforce and comply with social norms? Here we investigate the learning dynamics and emergence of compliance and enforcement of social norms in a foraging game, implemented in a multi-agent reinforcement learning setting. In this spatiotemporally extended game, individuals are incentivized to implement complex berry-foraging policies and punish transgressions against social taboos covering specific berry types. We show that agents benefit when eating poisonous berries is taboo, meaning the behavior is punished by other agents, as this helps overcome a credit-assignment problem in discovering delayed health effects. Critically, however, we also show that introducing an additional taboo, which results in punishment for eating a harmless berry, improves the rate and stability with which agents learn to punish taboo violations and comply with taboos. Counterintuitively, our results show that an arbitrary taboo (a "silly rule") can enhance social learning dynamics and achieve better outcomes in the middle stages of learning. We discuss the results in the context of studying normativity as a group-level emergent phenomenon.
In artificial intelligence, we often specify tasks through a reward function. While this works well in some settings, many tasks are hard to specify this way. In deep reinforcement learning, for example, directly specifying a reward as a function of a high-dimensional observation is challenging. Instead, we present an interface for specifying tasks interactively using demonstrations. Our approach defines a set of increasingly complex policies. The interface allows the user to switch between these policies at fixed intervals to generate demonstrations of novel, more complex, tasks. We train new policies based on these demonstrations and repeat the process. We present a case study of our approach in the Lunar Lander domain, and show that this simple approach can quickly learn a successful landing policy and outperforms an existing comparison-based deep RL method.