Building Trustworthy AI by Addressing its 16+2 Desiderata with Goal-Directed Commonsense Reasoning

TL;DR

The paper proposes Building Trustworthy AI by addressing 16+2 Desiderata with Goal-Directed Commonsense Reasoning, positioning s(CASP) as a lightweight, explainable symbolic reasoning engine that can integrate with machine learning. It argues that combining a robust commonsense knowledge base with a goal-directed reasoner can achieve trustworthiness aspects such as explainability, consistency, and reasoning across multiple possible worlds, while overcoming LLM brittleness. The authors map each desideratum to concrete s(CASP) features (e.g., constructive negation, default reasoning, abductive rules, deontic constraints, and multi-world semantics) and illustrate applications including VECSR and reliable chatbots, with a commercially deployed case (IRMA). They advocate open-source accessibility and future work on efficiency and KB maintenance to enable practical, trustworthy AI systems across domains.

Abstract

Current advances in AI and its applicability have highlighted the need to ensure its trustworthiness for legal, ethical, and even commercial reasons. Sub-symbolic machine learning algorithms, such as the LLMs, simulate reasoning but hallucinate and their decisions cannot be explained or audited (crucial aspects for trustworthiness). On the other hand, rule-based reasoners, such as Cyc, are able to provide the chain of reasoning steps but are complex and use a large number of reasoners. We propose a middle ground using s(CASP), a goal-directed constraint-based answer set programming reasoner that employs a small number of mechanisms to emulate reliable and explainable human-style commonsense reasoning. In this paper, we explain how s(CASP) supports the 16 desiderata for trustworthy AI introduced by Doug Lenat and Gary Marcus (2023), and two additional ones: inconsistency detection and the assumption of alternative worlds. To illustrate the feasibility and synergies of s(CASP), we present a range of diverse applications, including a conversational chatbot and a virtually embodied reasoner.