ChatGPT is a large language model developed by OpenAI as a conversational agent. ChatGPT was trained on data generated by humans and by receiving human feedback. This training process results in a bias toward humans' traits and preferences. In this paper, we stress multiple biases of ChatGPT, and show that its responses demonstrate many human traits. We Begin by showing a very high correlation between the frequency of digits generated by ChatGPT and humans' favorite numbers, with the most frequent digit generated by ChatGPT, matching humans' most favorable number, 7. We continue by showing that ChatGPT's responses in several social experiments are much closer to those of humans' than to those of fully rational agents. Finally, we show that several cognitive biases, known in humans, are also present in ChatGPT's responses.

One of the ways to make reinforcement learning (RL) more ef- ficient is by utilizing human advice. Because human advice is expensive, the central question in advice-based reinforcement learning is, how to decide in which states the agent should ask for advice. To approach this challenge, various advice strate- gies have been proposed. Although all of these strategies dis- tribute advice more efficiently than naive strategies (such as choosing random states), they rely solely on the agent’s inter- nal representation of the task (the action-value function, the policy, etc.) and therefore, are rather inefficient when this rep- resentation is not accurate, in particular, in the early stages of the learning process. To address this weakness, we propose an approach to advice-based RL, in which the human’s role is not limited to giving advice in chosen states, but also includes hint- ing apriori (before the learning procedure) which sub-domains of the state space require more advice. Specifically, we sug- gest different ways to improve any given advice strategy by utilizing the concept of critical states: states in which it is very important to choose the correct action. Finally, we present ex- periments in 2 environments that validate the efficiency of our approach.

We tackle the problem of an agent interacting with humans in a general-sum environment, i.e., a non-zero sum, non-fully cooperative setting, where the agent's goal is to increase its own utility. We show that when data is limited, building an accurate human model is very challenging, and that a reinforcement learning agent, which is based on this data, does not perform well in practice. Therefore, we propose that the agent should try maximizing a linear combination of the human's utility and its own utility rather than simply trying to maximize only its own utility.

This paper explores the hypothesis of the historical evolution of Semitic morphology from biconsonantal (2C) etymons, to triconsonantal (3C) roots, which make up the majority of words in Biblical Hebrew as well as in other Semitic languages such as modern Hebrew, Arabic, etc. The rules for reducing the 3C roots to their 2C etymons are provided in detail. We use BHSA, a manually annotated corpus of the Hebrew Bible, and Word2Vec, a method for converting words to a vector representing their semantic meaning, to study the hypothesis of evolution from 2C etymons to 3C roots in biblical Hebrew. Namely, we show that words in Hebrew with different roots, that might have originated from the same 2C etymons form a denser cluster than random sets of words of the same size. These differences are statistically significant and strongly support the hypothesis of evolution from 2C etymons to 3C roots.

The Shapley value is one of the most important normative division schemes in cooperative game theory, satisfying basic axioms. However, some allocation according to the Shapley value may seem unfair to humans. In this paper, we develop an automatic method that generates intuitive explanations for a Shapley-based payoff allocation, which utilizes the basic axioms. Given any coalitional game, our method decomposes it to sub-games, for which it is easy to generate verbal explanations, and shows that the given game is composed of the sub-games. Since the payoff allocation for each sub-game is perceived as fair, the Shapley-based payoff allocation for the given game should seem fair as well. We run an experiment with 630 human participants and show that when applying our method, humans perceive the Shapley-based payoff allocation as more fair than the Shapley-based payoff allocation without any explanation or with explanations generated by other methods.

We develop an automatic method that, given a contrastive query from the user, generates contrastive explanations based on items' features and users' preferences. That is, once receiving a recommendation, the users have the option to ask the system why it did not recommend a specific different item. Our method enables a recommendation system to reply with a meaningful and convincing personalized explanation. For example, the recommendation system may recommend a Samsung S22 phone. The user may ask the system why it did not recommend the Xiaomi 12. Based on the user's preferences, all other users' preferences, and the specific phones in question, our method might infer that a good camera is particularly important to the user, and thus, say that the Samsung S22 includes a better camera. We show that humans are more convinced that the recommended item is better than the contrastive item when using our contrastive explanations.