Causal Parrots: Large Language Models May Talk Causality But Are Not Causal
6
Citation
0
Reference
10
Related Paper
Citation Trend
Abstract:
Some argue scale is all what is needed to achieve AI, covering even causal models. We make it clear that large language models (LLMs) cannot be causal and give reason onto why sometimes we might feel otherwise. To this end, we define and exemplify a new subgroup of Structural Causal Model (SCM) that we call meta SCM which encode causal facts about other SCM within their variables. We conjecture that in the cases where LLM succeed in doing causal inference, underlying was a respective meta SCM that exposed correlations between causal facts in natural language on whose data the LLM was ultimately trained. If our hypothesis holds true, then this would imply that LLMs are like parrots in that they simply recite the causal knowledge embedded in the data. Our empirical analysis provides favoring evidence that current LLMs are even weak `causal parrots.'Keywords:
Causality
Causal model
Causal analysis
Causal reasoning
Causal structure
Causal model
Causal structure
Causal analysis
Causal reasoning
Causality
Causal decision theory
Contingency table
Cite
Citations (619)
It is known that existed causal reasoning methods will have problems to propagate the causal behaviors when there are some conflicts. To solve this problem, the authors provide a causal relation formulation, which introduces the concepts of ‘time’ and ‘effect’. Then a process for causal reasoning is provided: constructs the causal relation graph, then propagates the causal behaviors of the reason variables (exogenous variables), and uses qualitative algebra to solve conflicting cases. It is an extension of Iwasaki's causal reasoning method.
Causal reasoning
Causal model
Qualitative reasoning
Causal structure
Causal analysis
Analytic reasoning
Cite
Citations (0)
Attempts to characterize people’s causal knowledge of a domain in terms of causal network structures miss a key level of abstraction: the laws that allow people to formulate meaningful causal network hypotheses, and thereby learn and reason about novel causal systems so e ectively. We outline a preliminary framework for modeling causal laws in terms of generative grammars for causal networks. We then present an experiment showing that causal grammars can be learned rapidly in a novel domain and used to support one-shot inferences about the unobserved causal properties of new objects. Finally, we give a Bayesian analysis explaining how causal grammars may be induced from the limited data available in our experiments.
Causal model
Causal structure
Causal analysis
Abstraction
Causal reasoning
Causality
Causal decision theory
Cite
Citations (38)
Abstract Inferring the effect of interventions within complex systems is a fundamental problem of statistics. A widely studied approach uses structural causal models that postulate noisy functional relations among a set of interacting variables. The underlying causal structure is then naturally represented by a directed graph whose edges indicate direct causal dependencies. In a recent line of work, additional assumptions on the causal models have been shown to render this causal graph identifiable from observational data alone. One example is the assumption of linear causal relations with equal error variances that we will take up in this work. When the graph structure is known, classical methods may be used for calculating estimates and confidence intervals for causal-effects. However, in many applications, expert knowledge that provides an a priori valid causal structure is not available. Lacking alternatives, a commonly used two-step approach first learns a graph and then treats the graph as known in inference. This, however, yields confidence intervals that are overly optimistic and fail to account for the data-driven model choice. We argue that to draw reliable conclusions, it is necessary to incorporate the remaining uncertainty about the underlying causal structure in confidence statements about causal-effects. To address this issue, we present a framework based on test inversion that allows us to give confidence regions for total causal-effects that capture both sources of uncertainty: causal structure and numerical size of non-zero effects.
Causal structure
Causal model
Cite
Citations (0)
Based on the structural causal model, this study derived a causal graph that shows the causal relationship between the factors predicting the teaching competency of lower secondary school teachers in South Korea, the UK(England), and Finland. Also, it compared and analyzed the causal path to each country’s teaching competency. To this end, the data of lower secondary school teachers and principals, who participated in TALIS 2018, in Korea, the UK(England), and Finland were analyzed. First, the top 20 factors that predict teaching competency by each country were extracted by applying the mixed-effect random forest technique in consideration of the multi-layer structure of the data. Then, the causal graphs were derived by applying the causal discovery algorithm based on a structural causal model with the extracted predictors. As a result, there were common factors and discrimination factors in the top 20 predictors extracted from each national data, and the causal paths to teaching competency were compared and analyzed in the context of each country based on the causal graph by country. In addition, in the field of education, the possibility of using causal inference based on structural causal models was discussed, and the limitations and implications of this study were presented.
Causal model
Causal analysis
Causal reasoning
Cite
Citations (0)
Abstract Causal beliefs and reasoning are deeply embedded in many parts of our cognition. We are clearly ‘causal cognizers’, as we easily and automatically (try to) learn the causal structure of the world, use causal knowledge to make decisions and predictions, generate explanations using our beliefs about the causal structure of the world, and use causal knowledge in many other ways. Because causal cognition is so ubiquitous, psychological research into it is itself an enormous topic, and literally hundreds of people have devoted entire careers to the study of it. Causal cognition can be divided into two rough categories: causal learning and causal reasoning. The former encompasses the processes by which we learn about causal relations in the world at both the type and token levels; the latter refers to the ways in which we use those causal beliefs to make further inferences, decisions, predictions, and so on.
Causal reasoning
Causal structure
Causal model
Causal decision theory
Causal analysis
Causality
Cite
Citations (64)
Abstract We used a new method to assess how people can infer unobserved causal structure from patterns of observed events. Participants were taught to draw causal graphs, and then shown a pattern of associations and interventions on a novel causal system. Given minimal training and no feedback, participants in Experiment 1 used causal graph notation to spontaneously draw structures containing one observed cause, one unobserved common cause, and two unobserved independent causes, depending on the pattern of associations and interventions they saw. We replicated these findings with less‐informative training (Experiments 2 and 3) and a new apparatus (Experiment 3) to show that the pattern of data leads to hidden causal inferences across a range of prior constraints on causal knowledge.
Causal structure
Causal model
Causality
Causal reasoning
Causal analysis
Cite
Citations (47)
This chapter provides an introduction to how humans learn and reason about multiple causal relations connected together in a causal structure. The first half of the chapter focuses on how people learn causal structures. The main topics involve learning from observations versus interventions, learning temporal versus atemporal causal structures, and learning the parameters of a causal structure including individual cause-effect strengths and how multiple causes combine to produce an effect. The second half of the chapter focuses on how individuals reason about the causal structure, such as making predictions about one variable given knowledge about other variables, once the structure has been learned. Some of the most important topics involve reasoning about observations versus interventions, how well people reason compared to normative models, and whether causal structure beliefs bias reasoning. In both sections the author highlights open empirical and theoretical questions.
Causal structure
Causal reasoning
Causal model
Causal decision theory
Causality
Cite
Citations (17)
Large Language Models (LLMs) exhibit exceptional abilities for causal analysis between concepts in numerous societally impactful domains, including medicine, science, and law. Recent research on LLM performance in various causal discovery and inference tasks has given rise to a new ladder in the classical three-stage framework of causality. In this paper, we advance the current research of LLM-driven causal discovery by proposing a novel framework that combines knowledge-based LLM causal analysis with data-driven causal structure learning. To make LLM more than a query tool and to leverage its power in discovering natural and new laws of causality, we integrate the valuable LLM expertise on existing causal mechanisms into statistical analysis of objective data to build a novel and practical baseline for causal structure learning. We introduce a universal set of prompts designed to extract causal graphs from given variables and assess the influence of LLM prior causality on recovering causal structures from data. We demonstrate the significant enhancement of LLM expertise on the quality of recovered causal structures from data, while also identifying critical challenges and issues, along with potential approaches to address them. As a pioneering study, this paper aims to emphasize the new frontier that LLMs are opening for classical causal discovery and inference, and to encourage the widespread adoption of LLM capabilities in data-driven causal analysis.
Causality
Causal structure
Causal reasoning
Causal model
Leverage (statistics)
Causal analysis
Cite
Citations (12)
Inferring the effect of interventions within complex systems is a fundamental problem of statistics. A widely studied approach employs structural causal models that postulate noisy functional relations among a set of interacting variables. The underlying causal structure is then naturally represented by a directed graph whose edges indicate direct causal dependencies. In a recent line of work, additional assumptions on the causal models have been shown to render this causal graph identifiable from observational data alone. One example is the assumption of linear causal relations with equal error variances that we will take up in this work. When the graph structure is known, classical methods may be used for calculating estimates and confidence intervals for causal effects. However, in many applications, expert knowledge that provides an a priori valid causal structure is not available. Lacking alternatives, a commonly used two-step approach first learns a graph and then treats the graph as known in inference. This, however, yields confidence intervals that are overly optimistic and fail to account for the data-driven model choice. We argue that to draw reliable conclusions, it is necessary to incorporate the remaining uncertainty about the underlying causal structure in confidence statements about causal effects. To address this issue, we present a framework based on test inversion that allows us to give confidence regions for total causal effects that capture both sources of uncertainty: causal structure and numerical size of nonzero effects.
Causal structure
Causal model
Cite
Citations (0)