Theory of Mind (ToM) is a critical component of intelligence but its assessment remains the subject of heated debates. Prior research applied human ToM assessments to natural language processing models using either human-created standardized tests or rule-based templates. However, these methods primarily focus on simplistic reasoning and require further validation. Here, we leverage dynamic epistemic logic to isolate a particular component of ToM and to generate controlled problems. We also introduce new verbalization techniques to express these problems in English natural language. Our findings indicate that some language model scaling (from 70M to 6B and 350M to 174B) does not consistently yield results better than random chance. While GPT-4 demonstrates superior epistemic reasoning capabilities, there is still room for improvement. Our code and datasets are publicly available (https://huggingface.co/datasets/sileod/mindgames , https://github.com/sileod/llm-theory-of-mind )

When a technology is labelled AI, the vendor must provide information that makes it clear how AI is used as a differentiator and what problems it solves that can't be solved by other technologies, explained Jim Hare, a research VP at Gartner, who focuses on analytics and data science. "You have to go in with the assumption that it isn't AI, and the vendor has to prove otherwise," Hare said. "It's like the big data era -- where all the vendors say they have big data -- but on steroids." "What I'm seeing is that anything typically called machine learning is now being labelled AI, when in reality it is weak or narrow AI, and it solves a specific problem," he said. IT buyers must hold the vendor accountable for its claims by asking how it defines AI and requesting information about what's under the hood, Hare said.

When a technology is labelled AI, the vendor must provide information that makes it clear how AI is used as a differentiator and what problems it solves that can't be solved by other technologies, explained Jim Hare, a research VP at Gartner, who focuses on analytics and data science. "What I'm seeing is that anything typically called machine learning is now being labelled AI, when in reality it is weak or narrow AI, and it solves a specific problem," he said. Also, Hare urges IT buyers to demand a demonstration of artificial intelligence products using their own data to see them in action solving a business problem they have. The second problem Gartner highlights is that machine learning can address many of the problems businesses need to solve.

Analysts predict that by 2020, artificial intelligence technologies will be in almost every new software and service release. And if they're not actually in them, technology vendors will probably use smoke and mirrors marketing tactics to make users believe they are. Many tech vendors already shoehorn the AI label into the marketing of every new piece of software they develop, and it's causing confusion in the market. To muddle things further, major software vendors accuse their competitors of egregious mislabeling, even when the products in question truly do include artificial intelligence technologies. AI mischaracterization is one of the three major problems in the AI market, as highlighted by Gartner recently.

Just as there exists varied uses for computational models of narrative, there exists a wide variety of languages aimed at representing stories. A number of them have historic roots in automated generation, for which these languages have to be limited in order to make the generation process computationally feasible. Other are focused on story understanding, with close ties to natural language making many reasoning processes computationally intractable. In this paper, we discuss the trade-off between expressivity and computational complexity of the reasoning process and argue that Impulse, a temporal, modal logic provides more expressivity than languages historically associated with story generation, while still affording reasoning capabilities. We show that these properties enable certain aspects of narrative discourse generation by using two examples from different genres, and claim that this generalizes to a broader class of problems.

Typically, translation algorithms from a symbolic to a connectionist representation and vice-versa are employed to provide either (i) a neural implementation of a logic, (ii) a logical characterisation of a neural system, or (iii) a hybrid learning system that brings together features from connectionism and symbolic artificial intelligence (Holldobler, 1993). Until recently, neural-symbolic systems were not able to fully represent, reason and learn expressive languages other than propositional and fragments of first-order logic (Cloete & Zurada, 2000). However, in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), a new approach to knowledge representation and reasoning in neural-symbolic systems based on neural networks ensembles has been introduced. This new approach shows that modal logics can be effectively represented in artificial neural networks. In this paper, following the approach introduced in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), we move one step further and show that temporal logics can be effectively represented in artificial neural o Artur Garcez is partly supported by the Nuffield Foundation. Luis Lamb is partly supported by CNPq. The authors would like to thank the referees for their comments.

Typically, translation algorithms from a symbolic to a connectionist representation and vice-versa are employed to provide either (i) a neural implementation of a logic, (ii) a logical characterisation of a neural system, or (iii) a hybrid learning system that brings together features from connectionism and symbolic artificial intelligence (Holldobler, 1993). Until recently, neural-symbolic systems were not able to fully represent, reason and learn expressive languages other than propositional and fragments of first-order logic (Cloete & Zurada, 2000). However, in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), a new approach to knowledge representation and reasoning in neural-symbolic systems based on neural networks ensembles has been introduced. This new approach shows that modal logics can be effectively represented in artificial neural networks. In this paper, following the approach introduced in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), we move one step further and show that temporal logics can be effectively represented in artificial neural o Artur Garcez is partly supported by the Nuffield Foundation. Luis Lamb is partly supported by CNPq. The authors would like to thank the referees for their comments.

Typically, translation algorithms from a symbolic to a connectionist representation and vice-versa are employed to provide either (i) a neural implementation of a logic, (ii) a logical characterisation of a neural system, or (iii) a hybrid learning system that brings together features from connectionism and symbolic artificial intelligence (Holldobler, 1993). Until recently, neural-symbolic systems were not able to fully represent, reason and learn expressive languages other than propositional and fragments of first-order logic (Cloete & Zurada, 2000). However, in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), a new approach to knowledge representation and reasoning in neural-symbolic systems based on neural networks ensembles has been introduced. This new approach shows that modal logics can be effectively represented in artificial neural networks. In this paper, following the approach introduced in (d'Avila Garcez et al., 2002b; d'Avila Garcez et al., 2002c; d'Avila Garcez et al., 2003), we move one step further and show that temporal logics can be effectively represented in artificial neural o Artur Garcez is partly supported by the Nuffield Foundation. Luis Lamb is partly supported by CNPq. The authors would like to thank the referees for their comments.