The Winograd Schema Challenge—a set of twin sentences involving pronoun reference disambiguation that seem to require the use of commonsense knowledge—was proposed by Hector Levesque in 2011. By 2019, a number of AI systems, based on large pre-trained transformer-based language models and fine-tuned on these kinds of problems, achieved better than 90% accuracy. In this paper, we review the history of the Winograd Schema Challenge and discuss the lasting contributions of the flurry of research that has taken place on the WSC in the last decade. We discuss the significance of various datasets developed for WSC, and the research community's deeper understanding of the role of surrogate tasks in assessing the intelligence of an AI system.

Commonsense reasoning is in principle a central problem in artificial intelligence, but it is a very difficult one. One approach that has been pursued since the earliest days of the field has been to encode commonsense knowledge as statements in a logic-based representation language and to implement commonsense reasoning as some form of logical inference. This paper surveys the use of logic-based representations of commonsense knowledge in artificial intelligence research.

Actually, the personal anecdotes get a little wearisome.After reading about how Michael Trick used an algorithm to choose a wife, and how

We prove that the class of rational polyhedra and the class of topologically regular regions definable in an o-minimal structure over the reals are each elementarily equivalent to the class of polyhedra for topological languages.

We consider the expressive power of the first-order structure Ω, C where Ω is either of two of different domains of extended regions in Euclidean space, and C(x,y) is the topological relation "Region x is in contact with region y."We prove two main theorems:• Let P[Q] be the domain of bounded, non-empty, rational polyhedra in two-or threedimensional Euclidean space.A relation Γ over P[Q] is definable in the structure P[Q], C if and only if Γ is arithmetic and invariant under rational PL-homeomorphisms of the space to itself.We also extend this result to a number of other domains, including the domain of all polyhedra and the domain of semi-algebraic regions.• Let R be the space of bounded, non-empty, closed regular regions. in n-dimensional Euclidean space.Any analytical relation over lower-dimensional (i.e.empty interior) compact point sets that is invariant under homeomorphism is implicitly definable in the structure R, C .

We carry out a comparative study of the expressive power of different ontologies of matter in terms of the ease with which simple physical knowledge can be represented. In particular, we consider five ontologies of models of matter: particle models, fields, two ontologies for continuous material, and a hybrid model. We evaluate these in terms of how easily eleven benchmark physical laws and scenarios can be represented.

We categorize in recursion-theoretic terms the expressivity of a number of first-order languages that allow quantification over regions in Euclidean space. Specifically we show the following: (1) Let be any class of closed regions in Euclidean space that includes all simple polygons. Let C(x, y) be the relation, ‘region x is connected to region y’ and let Convex(x) be the property, ‘region x is convex’. Then any relation over that is analytical and invariant under affine transformations is first-order definable in the structure 〈⁠, C, Convex〉. (2) Let be as in (1), and let Closer(x, y, z) be the relation ‘region x is closer to y than to z.’ Then any relation over that is analytical and invariant under orthogonal transformations is first-order definable in the structure 〈⁠, Closer〉. (3) Let be the class of finite unions of intervals in the real line. Then any relation over that is analytical and invariant under linear transformations is first-order definable in the structure 〈⁠, Closer〉. (4) If the class of regions is restricted to be polygons with rational vertices, then results analogous to (1–3) hold, substituting ‘arithmetical relation’ for ‘analytical relation’.

In the Library of Congress cataloguing system, the call number for Where Mathematics Comes From, by George Lakoff and Raphael Nuñez, is QA141.I5.L37 2000, in the mathematics section. At the NYU Boo...

This paper presents a theory expressed in first-order logic for describing and supporting inference about action, knowledge, planning, and communication, in an egalitarian multi-agent setting. The underlying ontology of the theory uses a situationbased temporal model and a possible-worlds model of knowledge. It supports plans and communications of a very general kind, both informative communications and requests. Communications may refer to states of the world or states of knowledge in the past, present, or future. We demonstrate that the theory is powerful enough to represent several interesting multi-agent planning problems and to justify their solutions. We have proven that the theory of knowledge, communication, and planning is consistent with a broad range of physical theories, despite the existence of a number of potential paradoxes.

Order of magnitude reasoning - reasoning by rough comparisons of the sizes of quantities - is often called `back of the envelope calculation', with the implication that the calculations are quick though approximate. This paper exhibits an interesting class of constraint sets in which order of magnitude reasoning is demonstrably fast. Specifically, we present a polynomial-time algorithm that can solve a set of constraints of the form `Points a and b are much closer together than points c and d.' We prove that this algorithm can be applied if `much closer together' is interpreted either as referring to an infinite difference in scale or as referring to a finite difference in scale, as long as the difference in scale is greater than the number of variables in the constraint set. We also prove that the first-order theory over such constraints is decidable.

For an agent to be able to rely on a plan, he must know both that he is physically capable of carrying out the physical actions involved, and that he knows enough to carry out the plan. In this paper, we advance and discuss new definitions of ‘knowing enough to carry out a plan’, for the case of a single agent carrying out a sequence of primitive actions one at a time. We consider both determinate and indeterminate plans. We show how these definitions can be expressed in a formal logic, using a situation calculus model of time and a possible worlds model of knowledge. The definitions strictly subsume previous theories for the single-agent case without concurrent actions. We illustrate the power of the definition by showing that it supports results of the following kinds: Positive verification: showing that a plan is feasible. Negative verification: showing that a plan is infeasible. Monotonicity: the more an agent knows, the more plans are executable. Reduction for omniscient agent: for an omniscient agent, a plan is epistemically feasible if and only if it is physically feasible. Simple recursive rules that are sufficient conditions for the feasibility of a plan described as a sequence or a conditional combination of subplans.

My friend Dan Weld is so kind in his review that my only possible response is to

Constraint propagation is often used in AI systems to perform inference about quantities. This paper studies one particular kind of constraint propagation, where quantities are labelled with signs or with intervals, and these labels are propagated through recorded constraints. We review the uses of such inference schemes in AI systems of various kinds, and evaluate their strengths and weaknesses. In particular, we determine the completeness and running time of constraint propagation for various kinds of labels and constraints.

Several algorithms are presented for the nonpreemptlve assignment of n independent tasks to m unrelated processors One algorithm requires polynomial Ume in n and m and IS at most 2x/~ times worse than optimal in the worst case This is the best polynomial-time algorithm known for scheduling such sets of tasks.An algorithm with slightly better worst case performance requires polynomial time in n but exponential ume in m This 1s the best algorithm known that requires time O(nlogn) for every fixed value of m KEY WORDS AND PHRASES nonpreemptlve schedules, worst case fimshlng time, performance ratio, unrelated processors, largest processing time CR CATEGORIES' 4 32, 4 35, 5 25, 5 39 IntroductzonThis paper presents a number of polynomial-time algorithms for the scheduling of a set of n independent tasks on m processors of different speeds.The processors are unrelated in the sense that there is no notion of a fast processor always requiring less time than a slow processor, irrespective of the task being executed.Rather, the time required for the execution of a task on a processor is a function of both the task and the processor.This models the situation in which general-purpose processors have specialized capabilities that permit them to execute certain tasks more efficiently than others.An example of this might be in a distributed system where the time requirement of a task on a processor may depend on communication costs.The decision problem of determining whether a given set of tasks can be assigned with finishing time smaller than a given bound is NP-complete [6].As a result, it seems unlikely that an algorithm can be found which runs m polynomial time and always produces the optimal assignment [3].It is thus worthwhile to investigate approximation algorithms.