https://plato.stanford.edu/archives/spr2024/entries/logic-ai/ >[!info] Definition >*Artificial Intelligence* (referred to hereafter by its nickname, “*AI*”) is the subfield of Computer Science devoted to developing programs that enable computers to display behavior that can (broadly) be characterized as intelligent. - John McCarthy’s plan was to use ideas from philosophical logic to formalize commonsense reasoning. - new setting: building working, large-scale computational models of rational agency. ## [1\. Logic in Artificial Intelligence](https://plato.stanford.edu/entries/logic-ai//#LogiArtiInte) ### [1.1 The Role of Logic in Artificial Intelligence](https://plato.stanford.edu/entries/logic-ai//#RoleLogiArtiInte) - Logic can, for instance, provide a specification for a programming language, aiming at soundness and completeness. - More loosely - logical ideas can inform parts of the software development process. - Logical theory informs applications, and applications challenge logical theory and can lead to theoretical innovations. - Core components: - declarative representations - their retrieval and maintenance - the reasoning systems they service - Knowledge representation - declarative information/declarative representations ### [1.2 Philosophical Logic](https://plato.stanford.edu/entries/logic-ai//#PhilLogi) - a desire to apply the methods of mathematical logic to nonmathematical domains ### [1.3 Logic in AI and Philosophical Logic](https://plato.stanford.edu/entries/logic-ai//#LogiAIPhilLogi) - logical AI is philosophical logic constrained by an interest in large-scale formalization and in feasible, implementable reasoning. - logic deals with reasoning—and relatively little of the reasoning we do is mathematical, while almost all of the mathematical reasoning done by nonmathematicians is mere calculation. ## [2\. John McCarthy and Commonsense Logicism](https://plato.stanford.edu/entries/logic-ai//#JohnMcCaCommLogi) ### [2.1 Logic and AI](https://plato.stanford.edu/entries/logic-ai//#LogiAI) - McCarthy felt that even if AI implementations do not straightforwardly use logical reasoning techniques like theorem proving, a logical formalization will help to understand the reasoning problem itself. - The claim is that without a logical account of the reasoning domain, it will not be possible to implement the reasoning itself. ### [2.2 The Formalization of Common Sense](https://plato.stanford.edu/entries/logic-ai//#FormCommSens) - McCarthy’s long-term objective was to formalize _commonsense reasoning_, the prescientific reasoning that engages human thinking about everyday problems. ## [3\. Nonmonotonic Reasoning and Nonmonotonic Logics](https://plato.stanford.edu/entries/logic-ai//#NonmReasNonmLogi) ### [3.1 Nonmonotonicity](https://plato.stanford.edu/entries/logic-ai//#Nonm) >[!info] Definition >A logic is *monotonic* if its consequence relation has the property that if a set of formulas $T$ implies a consequence $B$ then a larger set $T \cup \{A\}$ will also imply $B$. >A logic is *nonmonotonic* if its consequence relation lacks this property. - While a mathematical proof must cover every contingency, practical reasoning routinely closes its eyes to some possibilities. ### [3.2 Beginnings](https://plato.stanford.edu/entries/logic-ai//#Begi) Minsky's challenges to the logical approach which helped drive nonmonotonic logic: - building large-scale representations - reasoning efficiently - representing control knowledge - providing for the flexible revision of defeasible beliefs. Applications that drove nonmonotonic logic: - belief revision - closed-world reasoning - planning. #### 3.2.1 Belief Revision >[!info] Definition (Doyle) >*Truth Maintenance System.* An algorithm providing a mechanism for updating the “beliefs” of a knowledge repository. - The idea is to keep track of the support of beliefs, and to use the record of these support dependencies when it is necessary to revise beliefs. - In a TMS, part of the support for a belief can consist in the _absence_ of some other belief. This introduces nonmonotonicity. For instance, it provides for defaults: beliefs that are induced by the absence of contrary beliefs. #### 3.2.2 Closed-world reasoning >[!info] Definition >The _closed-world assumption_: as far as simple claims (i.e. positive or negative literals) are concerned, the system assumes that it knows all that there is to be known. - This is another case of inference from the absence of a proof. A negative is proved, in effect, by the failure of a systematic attempt to prove the positive. #### 3.2.3 Planning >[!info] Definition >*Causal inertia* - variables are unchanged by the performance of an action unless there is a special reason to think that they will change. >*The frame problem:* the challenge of formalizing causal inertia in a temporal framework. - When making plans, we assume causal inertia. - This suggests that nonmonotonic temporal formalisms should apply usefully to reasoning about action and change, and in particular might address the frame problem. ### [3.3 The Earliest Formalisms](https://plato.stanford.edu/entries/logic-ai//#EarlForm) Three influential approaches to nonmonotonic logic: - _circumscription_ (McCarthy): restricting attention to certain types of models - _modal approaches_ (Doyle & McDermott): addition of modal operator $L$, informally interpreted as ‘provable’. - _default logic_ (Reiter): addition of "default rules" of the form "in the presence of $A_{1}\dots A_{n}$ and the absence of $B_{1} \dots B_{m}$ conclude $Cquot; The modal approach and default logic were proved to be "equivalent". ### [3.4 Later Work in Nonmonotonic Logic](https://plato.stanford.edu/entries/logic-ai//#LateWorkNonmLogi) ## [4\. Reasoning about Action and Change](https://plato.stanford.edu/entries/logic-ai//#ReasAbouActiChan) ### [4.1 Priorian Tense Logic](https://plato.stanford.edu/entries/logic-ai//#PrioTensLogi) ### [4.2 Planning Problems and the Situation Calculus](https://plato.stanford.edu/entries/logic-ai//#PlanProbSituCalc) ### [4.3 Formalizing Microworlds](https://plato.stanford.edu/entries/logic-ai//#FormMicr) ### [4.4 Prediction and the Frame Problem](https://plato.stanford.edu/entries/logic-ai//#PredFramProb) ### [4.5 Nonmonotonic Treatments of Inertia and a Package of Problems](https://plato.stanford.edu/entries/logic-ai//#NonmTreaInerPackProb) ### [4.6 Some Emergent Frameworks](https://plato.stanford.edu/entries/logic-ai//#SomeEmerFram) ## [5\. Causal Reasoning](https://plato.stanford.edu/entries/logic-ai//#CausReas) ## [6\. Spatial Reasoning](https://plato.stanford.edu/entries/logic-ai//#SpatReas) ## [7\. Reasoning about Knowledge](https://plato.stanford.edu/entries/logic-ai//#ReasAbouKnow) ## [8\. Towards a Formalization of Common Sense](https://plato.stanford.edu/entries/logic-ai//#TowaFormCommSens) ## [9\. Taxonomic Representation and Reasoning](https://plato.stanford.edu/entries/logic-ai//#TaxoReprReas) ### [9.1 Concept-Based Classification](https://plato.stanford.edu/entries/logic-ai//#ConcBaseClas) ### [9.2 Nonmonotonic Inheritance](https://plato.stanford.edu/entries/logic-ai//#NonmInhe) ## [10\. Contextual Reasoning](https://plato.stanford.edu/entries/logic-ai//#ContReas) ## [11\. Prospects for a Logical Theory of Practical Reason](https://plato.stanford.edu/entries/logic-ai//#ProsForLogiTheoPracReas) ## [12\. Readings](https://plato.stanford.edu/entries/logic-ai//#Read) ## [Bibliography](https://plato.stanford.edu/entries/logic-ai//#Bib) ## [Academic Tools](https://plato.stanford.edu/entries/logic-ai//#Aca) ## [Other Internet Resources](https://plato.stanford.edu/entries/logic-ai//#Oth) ## [Related Entries](https://plato.stanford.edu/entries/logic-ai//#Rel) ___