Rationality is arguably a contested concept (like justice). ( Rationality Hypothesis) Suitably designed and trained NLMs may systematically display advanced rational skills.īy discussing the (Rationality Hypothesis), we put the more specific questions addressed in this study (see Q1–Q4 below) in a broader scientific context, sketching their potential relevance for a variety of disciplines and fields. The strong performance of NLMs in natural language understanding tasks triggers the more fundamental question whether NLMs are rational agents: They are, first and foremost, trained to fill in missing or next words in a text and they do predict a word by assigning probabilities to all words available in a given vocabulary. Technically, and leaving aside all the details, NLMs are essentially probabilistic word prediction machines. The performance of these systems has exploded with the advent of the so-called Transformer network architecture and has been increasing steadily over the last years (e.g., ) through further optimizations of machine learning algorithms and system design, increases in model size, or quantitatively and qualitatively improved training datasets. Neural language models (NLMs) are powerful natural language processing systems which have sparked a scientific revolution in the field of AI & NLP and excel at such diverse tasks as, e.g., machine translation, text summarization, question answering, or natural-language inference. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.Ĭompeting interests: The authors have declared that no competing interests exist. įunding: This work is supported by the Helmholtz Association Initiative and Networking Fund on the partition. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.ĭata Availability: The data is now available via wandb under the following links/repos. Received: JAccepted: JanuPublished: February 9, 2023Ĭopyright: © 2023 Betz, Richardson. Taylor’s University - Lakeside Campus: Taylor’s University, MALAYSIA We suggest that this hypothesis has empirical, yet also normative and conceptual ramifications far beyond the practical linguistic problems NLMs have originally been designed to solve.Ĭitation: Betz G, Richardson K (2023) Probabilistic coherence, logical consistency, and Bayesian learning: Neural language models as epistemic agents. All this, we conclude, confirms the Rationality Hypothesis, i.e., the claim that suitable trained NLMs may exhibit advanced rational skills. In addition, such self-training is found to have a pivotal role in rational evidential learning, too, for it seems to enable rankers to propagate a novel evidence item through their belief systems, successively re-adjusting individual degrees of belief. While pretrained rankers are found to suffer from global inconsistency (in agreement with, e.g., ), we observe that subsequent self-training on auto-generated texts allows rankers to gradually obtain a probabilistically coherent belief system that is aligned with logical constraints. Moreover, we introduce a procedure for eliciting a model’s degrees of belief, and define numerical metrics that measure the extent to which given degrees of belief violate (probabilistic, logical, and Bayesian) rationality constraints. To this purpose, we conduct computational experiments with rankers: T5 models that are pretrained on carefully designed synthetic corpora. It is argued that suitably trained neural language models exhibit key properties of epistemic agency: they hold probabilistically coherent and logically consistent degrees of belief, which they can rationally revise in the face of novel evidence.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |