AIBullisharXiv – CS AI · May 127/10
🧠Researchers identify a fundamental geometric flaw in decoder-based Vision-Language Models where visual embeddings become over-aligned with linguistic patterns, causing systematic hallucinations. The study introduces quantitative methods to characterize this bias and proposes training-free and fine-tuning solutions that reduce hallucinations across multiple benchmarks without computational overhead.
AIBullisharXiv – CS AI · May 127/10
🧠Researchers propose a self-captioning workflow with a Multimodal Interaction Gate to improve vision language models by amplifying redundant information between vision and text modalities. The approach addresses hallucination and robustness issues by converting unique modal interactions into shared redundancies, reducing visual-induced errors by 38.3% and improving consistency by 16.8%.
AIBullisharXiv – CS AI · May 127/10
🧠Researchers propose Theorem-SFT, a novel supervised fine-tuning approach that teaches language models to apply mathematical rules explicitly rather than memorize surface-level correlations between problems and solutions. The method demonstrates significant performance improvements across benchmarks while revealing that feed-forward layers, not memorization itself, are the primary locus of reasoning capability.
AINeutralarXiv – CS AI · May 117/10
🧠Researchers developed a method to extract and analyze search trees from LLM reasoning traces, revealing that large language models use shallower, more myopic planning strategies compared to humans. While LLMs generate extended chain-of-thought reasoning, their actual decision-making is driven primarily by shallow search rather than deep lookahead, contrasting sharply with human expert planning.
AIBullisharXiv – CS AI · May 117/10
🧠Researchers introduce a mechanistic-interpretability toolkit using Sparse Autoencoders and linear probes to diagnose AI agent failures before they occur, addressing a critical gap in enterprise AI deployment where tool-use errors in long-horizon workflows create cascading safety and financial risks.
🏢 Nvidia
AIBearisharXiv – CS AI · May 97/10
🧠A peer-reviewed study evaluates explainability methods in AI systems used for automatic target recognition in safety-critical applications, revealing that popular post-hoc explanation techniques have significant limitations including spurious explanations and vulnerability to manipulation. The research argues that current XAI approaches are insufficient for deployment in high-stakes environments and calls for more robust, causally-grounded methods that prioritize system assurance over visual plausibility.
AIBullisharXiv – CS AI · May 97/10
🧠Researchers introduce PCNET, a probabilistic circuit-based method that detects hallucinations in large language models as geometric anomalies in the factual manifold, achieving 99% detection accuracy. The approach uses PC-LDCD decoding to correct hallucinations selectively without corrupting originally correct outputs, demonstrating significant improvements across multiple benchmarks.
AINeutralarXiv – CS AI · Apr 207/10
🧠Researchers demonstrate through causal experiments that hallucinations in language models arise from early trajectory commitments governed by asymmetric attractor dynamics. Using controlled prompt bifurcation on Qwen2.5-1.5B, they show that 44% of test prompts diverge into factual or hallucinated outputs at the first token, with activation patterns revealing that corrupting correct trajectories is far easier than recovering hallucinated ones—suggesting hallucination represents a stable but difficult-to-escape attractor state.
AIBullisharXiv – CS AI · Apr 207/10
🧠Researchers introduce Sequential Internal Variance Representation (SIVR), a novel supervised framework for detecting hallucinations in large language models by analyzing token-wise and layer-wise variance patterns in hidden states. The method demonstrates superior generalization compared to existing approaches while requiring smaller training datasets, potentially enabling practical deployment of hallucination detection systems.
AINeutralarXiv – CS AI · Apr 157/10
🧠Researchers have identified a critical vulnerability in large language models where safety guardrails fail across low-resource languages despite strong performance in high-resource ones. The team proposes LASA (Language-Agnostic Semantic Alignment), a new method that anchors safety protocols at the semantic bottleneck layer, dramatically reducing attack success rates from 24.7% to 2.8% on tested models.
AIBullisharXiv – CS AI · Apr 157/10
🧠Researchers propose a two-stage LLM framework that uses one model to translate XAI technical outputs into natural language and a second model to verify accuracy, faithfulness, and completeness before delivering explanations to users. The framework includes iterative refinement mechanisms and demonstrates improved reliability across multiple XAI techniques and LLM families.
AINeutralarXiv – CS AI · Apr 147/10
🧠Researchers introduce Pando, a benchmark that evaluates mechanistic interpretability methods by controlling for the 'elicitation confounder'—where black-box prompting alone might explain model behavior without requiring white-box tools. Testing 720 models, they find gradient-based attribution and relevance patching improve accuracy by 3-5% when explanations are absent or misleading, but perform poorly when models provide faithful explanations, suggesting interpretability tools may provide limited value for alignment auditing.
AIBullisharXiv – CS AI · Apr 147/10
🧠FACT-E is a new evaluation framework that uses controlled perturbations to assess the faithfulness of Chain-of-Thought reasoning in large language models, addressing the problem of models generating seemingly coherent explanations with invalid intermediate steps. By measuring both internal chain consistency and answer alignment, FACT-E enables more reliable detection of flawed reasoning and selection of trustworthy reasoning trajectories for in-context learning.
AINeutralarXiv – CS AI · Apr 77/10
🧠A new research study reveals that truth directions in large language models are less universal than previously believed, with significant variations across different model layers, task types, and prompt instructions. The findings show truth directions emerge earlier for factual tasks but later for reasoning tasks, and are heavily influenced by model instructions and task complexity.
AINeutralarXiv – CS AI · Apr 77/10
🧠Researchers identified a sparse routing mechanism in alignment-trained language models where gate attention heads detect content and trigger amplifier heads that boost refusal signals. The study analyzed 9 models from 6 labs and found this routing mechanism distributes at scale while remaining controllable through signal modulation.
AINeutralarXiv – CS AI · Apr 67/10
🧠Researchers analyzed the geometric structure of layer updates in deep language models, finding they decompose into a dominant tokenwise component and a geometrically distinct residual. The study shows that while most updates behave like structured reparameterizations, functionally significant computation occurs in the residual component.
AINeutralarXiv – CS AI · Mar 267/10
🧠Researchers analyzed how large language models (4B-72B parameters) internally represent different ethical frameworks, finding that models create distinct ethical subspaces but with asymmetric transfer patterns between frameworks. The study reveals structural insights into AI ethics processing while highlighting methodological limitations in probing techniques.
AINeutralarXiv – CS AI · Mar 177/10
🧠Researchers introduce Distributional Semantics Tracing (DST), a new framework for explaining hallucinations in large language models by tracking how semantic representations drift across neural network layers. The method reveals that hallucinations occur when models are pulled toward contextually inconsistent concepts based on training correlations rather than actual prompt context.
AINeutralarXiv – CS AI · Mar 47/102
🧠Researchers developed linear probes that can predict whether large language models will answer questions correctly by analyzing neural activations before any answer is generated. The method works across different model sizes and generalizes to out-of-distribution datasets, though it struggles with mathematical reasoning tasks.
AIBullisharXiv – CS AI · Mar 46/102
🧠Researchers developed SAE-based Transferability Score (STS), a new metric using sparse autoencoders to predict how well fine-tuned large language models will perform across different domains without requiring actual training. The method achieves correlation coefficients above 0.7 with actual performance changes and provides interpretable insights into model adaptation.
AINeutralarXiv – CS AI · Mar 46/103
🧠Researchers found that narrow finetuning of Large Language Models leaves detectable traces in model activations that can reveal information about the training domain. The study demonstrates that these biases can be used to understand what data was used for finetuning and suggests mixing pretraining data into finetuning to reduce these traces.
AINeutralarXiv – CS AI · Feb 277/105
🧠Researchers propose Geodesic Integrated Gradients (GIG), a new method for explaining AI model decisions that uses curved paths instead of straight lines to compute feature importance. The method addresses flawed attributions in existing approaches by integrating gradients along geodesic paths under a model-induced Riemannian metric.
AIBearisharXiv – CS AI · 3d ago6/10
🧠Researchers demonstrate that Vision-Language Models (VLMs) used for optical character recognition produce fluent but visually unsupported text, relying heavily on language priors rather than actual image content. Testing on Ancient Greek critical editions reveals VLMs generate plausible errors while traditional OCR produces local noise, with token-level grounding analysis showing model-specific vulnerabilities to hallucination.
AINeutralarXiv – CS AI · 3d ago6/10
🧠Researchers present CODE, a novel approach to knowledge editing in large language models that replaces fact overwriting with causal reasoning. By embedding causal narratives and on-policy distillation into model parameters, CODE reduces self-refutation rates from 95.6% to 1.8%, enabling LLMs to evolve knowledge coherently rather than storing isolated facts.
AIBearisharXiv – CS AI · 3d ago6/10
🧠Researchers demonstrate a successful attack on Introspection Adapters, a technique proposed by Shenoy et al., by exploiting symmetry properties in the system. The findings highlight potential vulnerabilities in adapter-based AI architectures that could have implications for model security and trustworthiness.
