PAL improves few-shot reasoning accuracy by having LLMs generate executable programs rather than text-based chains of thought, outperforming much larger models on math and logic benchmarks.
hub
(2020), ‘The next decade in ai: Four steps towards robust artificial intelligence’
17 Pith papers cite this work, alongside 297 external citations. Polarity classification is still indexing.
17
Pith papers citing it
297
external citations · external index
hub tools
citation-role summary
background 3
citation-polarity summary
roles
background 3representative citing papers
PhyEditBench is a new benchmark for physics-aware image editing with real and synthetic instances plus a training-free PhyWorld baseline that uses test-time scaling to outperform SOTA models.
Invariant Gradient Alignment uses Logical Isomer Sets and a Continuous Gradient Conflict Mask to tighten OOD generalization bounds and boost empirical performance over ERM in reasoning distillation.
RAG models set new state-of-the-art results on open-domain QA by retrieving Wikipedia passages and conditioning a generative model on them, while also producing more factual text than parametric baselines.
AI occupational substitution is bounded more by commercial risk and liability than technical capability, producing high OAI for data scientists and near-zero for physical trades.
PHACT uses a propose-certify loop with deterministic derivation from fixed inputs to achieve zero false certifications in 80 adversarial trials across two models and temperatures.
AI benchmarks trap progress by operationalizing assumptions that redefine capabilities around the benchmarks themselves, and Epistematics provides an audit procedure to detect when evaluations cannot discriminate claimed capabilities from proxy behaviors.
Grounding LLMs via node-wise anchors in a traffic scenario taxonomy improves law-scenario matching by 29.1% and derived requirement accuracy by 36.9-38.2% on Chinese laws and 5,897 scenarios, enabling a compliance layer and real-time monitor for AVs.
ActivationReasoning grounds logical reasoning in LLM latent activations via SAEs to enable structured inference, concept composition, and behavior steering on multi-hop, abstraction, and safety tasks.
The paper defines possibility space, timing computation, and causal factum to make timing a computable variable, and illustrates the framework with automatic trajectory discovery and counterfactual timing on 3,276 breast-cancer patients.
Across 662 annotated Hugging Face threads, gated access (dominated by Llama), multimodal generation quality, and deployment/invocation complexity are the most prominent user concerns.
AI's compositional reasoning failures originate in psychological learning paradigms that shaped its architectures, and the ReSynth trimodular framework is proposed to embed systematicity structurally.
A causal attribution model is proposed that applies do-operators to quantify component contributions in LLMs' causal reasoning, motivating a fine-tuned model for pairwise causal discovery that combines knowledge and numerical data.
The paper proposes the Glassbox Framework in which Bayesian networks serve as transparent ante-hoc mediation layers for generative models to enable auditable reasoning traces and contestable outputs.
A position paper argues that post-hoc XAI explanations are unfaithful and paradoxical, proposing a shift to expert-based verification and certification of AI systems.
The paper defines Agent AI as interactive multimodal systems that perceive grounded data and generate embodied actions, arguing this approach can mitigate hallucinations in foundation models.
LLMs fail to detect hidden harmful intent, allowing systematic bypass of safety mechanisms through framing techniques, with reasoning modes often worsening the issue.
citing papers explorer
-
PAL: Program-aided Language Models
PAL improves few-shot reasoning accuracy by having LLMs generate executable programs rather than text-based chains of thought, outperforming much larger models on math and logic benchmarks.
-
PhyEditBench: A Real-World Multi-Stage Benchmark for Physics-Aware Image Editing
PhyEditBench is a new benchmark for physics-aware image editing with real and synthetic instances plus a training-free PhyWorld baseline that uses test-time scaling to outperform SOTA models.
-
Invariant Gradient Alignment for Robust Reasoning Distillation
Invariant Gradient Alignment uses Logical Isomer Sets and a Continuous Gradient Conflict Mask to tighten OOD generalization bounds and boost empirical performance over ERM in reasoning distillation.
-
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
RAG models set new state-of-the-art results on open-domain QA by retrieving Wikipedia passages and conditioning a generative model on them, while also producing more factual text than parametric baselines.
-
Bounded by Risk, Not Capability: Quantifying AI Occupational Substitution Rates via a Tech-Risk Dual-Factor Model
AI occupational substitution is bounded more by commercial risk and liability than technical capability, producing high OAI for data scientists and near-zero for physical trades.
-
Structural Certification for Reliable Physical Design with Language Models
PHACT uses a propose-certify loop with deterministic derivation from fixed inputs to achieve zero false certifications in 80 adversarial trials across two models and temperatures.
-
The Evaluation Trap: Benchmark Design as Theoretical Commitment
AI benchmarks trap progress by operationalizing assumptions that redefine capabilities around the benchmarks themselves, and Epistematics provides an audit procedure to detect when evaluations cannot discriminate claimed capabilities from proxy behaviors.
-
Towards Lawful Autonomous Driving: Deriving Scenario-Aware Driving Requirements from Traffic Laws and Regulations
Grounding LLMs via node-wise anchors in a traffic scenario taxonomy improves law-scenario matching by 29.1% and derived requirement accuracy by 36.9-38.2% on Chinese laws and 5,897 scenarios, enabling a compliance layer and real-time monitor for AVs.
-
ActivationReasoning: Logical Reasoning in Latent Activation Spaces
ActivationReasoning grounds logical reasoning in LLM latent activations via SAEs to enable structured inference, concept composition, and behavior steering on multi-hop, abstraction, and safety tasks.
-
To Use AI as Dice of Possibilities with Timing Computation
The paper defines possibility space, timing computation, and causal factum to make timing a computable variable, and illustrates the framework with automatic trajectory discovery and counterfactual timing on 3,276 breast-cancer patients.
-
When Models Meet Users: An Empirical Study of Perceptions of General LLMs and Multimodal LLMs on Hugging Face
Across 662 annotated Hugging Face threads, gated access (dominated by Llama), multimodal generation quality, and deployment/invocation complexity are the most prominent user concerns.
-
How Psychological Learning Paradigms Shaped and Constrained Artificial Intelligence
AI's compositional reasoning failures originate in psychological learning paradigms that shaped its architectures, and the ReSynth trimodular framework is proposed to embed systematicity structurally.
-
Enhancing Causal Reasoning in Large Language Models: A Causal Attribution Model for Precision Fine-Tuning
A causal attribution model is proposed that applies do-operators to quantify component contributions in LLMs' causal reasoning, motivating a fine-tuned model for pairwise causal discovery that combines knowledge and numerical data.
-
Beyond Post-hoc Explanation: Toward Glassbox AI via Probabilistic Mediation
The paper proposes the Glassbox Framework in which Bayesian networks serve as transparent ante-hoc mediation layers for generative models to enable auditable reasoning traces and contestable outputs.
-
Beyond Explainable AI (XAI): An Overdue Paradigm Shift and Post-XAI Research Directions
A position paper argues that post-hoc XAI explanations are unfaithful and paradoxical, proposing a shift to expert-based verification and certification of AI systems.
-
Agent AI: Surveying the Horizons of Multimodal Interaction
The paper defines Agent AI as interactive multimodal systems that perceive grounded data and generate embodied actions, arguing this approach can mitigate hallucinations in foundation models.
-
Beyond Context: Large Language Models' Failure to Grasp Users' Intent
LLMs fail to detect hidden harmful intent, allowing systematic bypass of safety mechanisms through framing techniques, with reasoning modes often worsening the issue.