Skip to content
Log in
Sponsored Artificial Intelligence: A Modern Approach, Global Edition Artificial Intelligence: A Modern Approach, Global Edition

Category

large language models

2,378 papers

#data science Open access Oct 2026

Less can be more for predicting properties with large language models

Abstract Predicting properties from coordinate–category data—sets of vectors paired with categorical information—is fundamental to computational science, manifesting in materials science as predicting formation energies or elastic moduli from crystal structures. While large language models (LLMs) have increasingly been...

Nawaf Alampara, Santiago Miret, Kevin Maik Jablonka · 13 citations
#artificial intelligence Oct 2026

Artificial Intelligence-Driven Smart Home Systems: Technologies, Applications and Future Challenges

With the continuous development of artificial intelligence (AI) technology, smart home systems have gradually evolved from traditional automated control modes toward intelligent systems with environmental perception, autonomous decision-making, and proactive service capabilities. However, existing smart homes still fac...

Ruikang Dong · 0 citations
#large language models Open access Oct 2026

Designing Chinese Tokenizers under Data Constraints: A White-Box Empirical Study of Character Fragmentation

Chinese text requires substantially more tokens than English in large language models (LLMs), increasing API costs and context-window consumption. Most prior work measures this premium using off-the-shelf tokenizers, but the underlying mechanism—where the additional tokens actually arise—remains poorly understood, part...

L. Gong · 0 citations
#large language models Open access Oct 2026

CONCEPTUAL EMERGENCE & GENERATIVE INTELLIGENCE AT THE LIMIT Living Concept Civilization — Flagship Research Monograph 06 STARLI / White Rainbow Civilization · October 2026 ABSTRACT When do interactions among concepts, models, tools and people produce a genuinely new, independently verifiable capacity rather than a novel label or a misleading evaluation threshold? This English-language research monograph develops a falsifiable program for conceptual emergence and generative intelligence. Its research three-coupling is Interaction Generation × New-Function Identification × Counterfactual Verification. The proposed emergence expression compares held-out capability improvement with the strongest viable baseline and a contamination penalty, with all terms requiring operational definition and appropriate normalization before use. RESEARCH DESIGN The volume studies both smooth scaling and possible discontinuities, evaluation-metric sensitivity, training-data leakage, compositional generalization, simple-rule baselines, factorial component ablation, multi-agent collaboration, correlated errors, human-AI complementarity, multimodal grounding, causal interventions, tool-dependence, calibration, cross-model reproducibility, governance, and intergenerational handoff. A claimed capability must survive preregistered unseen tasks, strong comparators, counterfactual perturbations and independent reproduction. A fluent explanation, a generated new term or a model's assertion of consciousness is not evidence of functional emergence. MONOGRAPH STRUCTURE The 82 research chapters each present one distinct proposed chapter-wide operational equation, followed by twelve independently paginated study sections. Thirty advanced physics derivation routes appear in the preface with three-dimensional conceptual diagrams; they are existing physics statements within specified domains and do not establish laws of language or awareness. The concluding 200 Research Gates and 33 Phoenix Answer Embryos each feature one of 233 distinct physical equation anchors. These equations are methodological contrasts and educational prompts, not proof of the emergence hypotheses. A separate Excel workbook indexes chapters, sections, gates, candidate mechanisms and physics anchors. EVIDENCE AND ETHICS This monograph is a theoretical and experimental research agenda. It reports no new trials, controlled effects or validated discoveries; all proposed mechanisms, metrics, forecasts and outcomes are UNVERIFIED until independently tested. The distinction between appearance under a chosen test and a robust new function must be maintained. Consent, intellectual provenance, lawful data use, security and affected persons' right to refuse are hard constraints rather than offsettable penalties. OPEN RESEARCH INVITATION Research Gates reserve space for questions that can be rejected, refined or answered by future researchers. Phoenix Answer Embryos are provisional designs that can be falsified, decomposed, recombined or responsibly retired. The goal is not a permanent claim that machines awaken or that thought obeys physical equations; it is an enduring, contestable toolkit for recognizing and scrutinizing useful new functions. FOUNDATIONS Project source: Living Concept Civilization — Eighteen Tri-Coupled Flagship Research Programs, v2.0 (2026-10-13), Flagship 06. Wei et al., Emergent Abilities of Large Language Models (2022), https://arxiv.org/abs/2206.07682. Schaeffer et al., Are Emergent Abilities of Large Language Models a Mirage? (2023), https://arxiv.org/abs/2304.15004. Chollet, On the Measure of Intelligence (2019), https://arxiv.org/abs/1911.01547. Wilkinson et al., FAIR Guiding Principles (2016), https://doi.org/10.1038/sdata.2016.18.

CONCEPTUAL EMERGENCE & GENERATIVE INTELLIGENCE AT THE LIMITLiving Concept Civilization — Flagship Research Monograph 06STARLI / White Rainbow Civilization · October 2026 ABSTRACT When do interactions among concepts, models, tools and people produce a genuinely new, independently verifiable capacity rather than a novel...

33 · 0 citations
#large language models Open access Oct 2026

Operationalizing Pragmatic Difficulty: Speech Act Hierarchy and the Appropriateness Gap in an AI-Generated Business English Oral Test

Large language models can now draft scenario-based test tasks on demand, which raises a validation question that has attracted little attention: does the difficulty of the generated tasks behave as the construct predicts? This study examined an AI-generated oral test of three business speech acts to establish whether t...

Shengtao Xie · 0 citations
#large language models Open access Oct 2026

What a Grenade Is For: An Annotation Schema and a Pre-registered Human-LLM Test for Utility Lineups in Professional Counter-Strike

Grenade use in professional Counter-Strike follows community conventions written neither in the game's rules nor in its physics. We present an annotation schema for utility lineups, defined by an origin zone, a landing zone and a grenade type, and a de_dust2 lineup dataset built from 69,827 throws in 122 professional m...

Shang-Jing Lyu · 0 citations
#large language models Dataset Open access Oct 2026

Code and data for "The Information Limits of Consistency-Based Uncertainty Estimation in Large Language Models"

Scripts, per-question tables, result tables of two runs (Qwen2.5-7B-Instruct and Qwen2.5-1.5B-Instruct on 1,800 short-answer questions from TruthfulQA, HaluEval and TriviaQA), the notebook code of the run, and the files that generate the numbers reported in the manuscript "The Information Limits of Consistency-Based Un...

Fathimah Al-Ma’shumah, Nita Ratnawaty · 0 citations
#large language models Open access Jan 2027

A RAG-enhanced multimodal large language model framework for culvert inspection

Despite their importance for maintaining infrastructure integrity, manual culvert inspections pose significant operational challenges for agencies such as the Utah Department of Transportation (UDOT) due to high labor demands, subjective assessments, and variability in evaluation outcomes. This study investigates the a...

Pouria Mohammadi, S. Asgari, Abbas Rashidi et al. · 0 citations
#large language models Review Oct 2026

From Sequence to Structure: AlphaFold and the AI Revolution in Aquaculture Drug Discovery and Vaccine Design.

Global aquaculture faces an escalating disease and antimicrobial resistance (AMR) crisis, exacerbated by climate-driven physiochemical shifts and intensive farming practices. The integration of artificial intelligence (AI) and structural biology, spearheaded by AlphaFold, represents a paradigm shift in aquatic pharmaco...

Enes Üstüner, Mustafa Öz · 0 citations
#large language models Review Open access Dec 2026

SentiAgent: An agentic LLM framework for context-aware and explainable sentiment analysis.

SentiAgent is introduced, a collaborative Agentic LLM framework for context-aware and explainable sentiment analysis that combines contextual understanding, external commonsense knowledge, and autonomous reasoning to improve sentiment prediction and interpretability.

Uma Shankar Yadav, Bhakti Thakre, Sumedha R. Chokhandre · 0 citations
#large language models Conference Open access Oct 2026

LLMs as Narrators of Lived Experience: A Study of Generative Hermeneutical Epistemic Injustice in ADHD Sense-Making

Large language models (LLMs), as non-human actors, do not have access to first-hand experiences. Yet they are increasingly consulted in processes of making sense of them, from mental health chatbots to reflexive journals. As a result, complex lived realities risk being distorted or flattened through second-order repres...

Anne Arzberger, Dong-Ying Zhang · 0 citations
#large language models Oct 2026

Care-Full Alignment: Operationalizing Care Ethics for Human-LLM Relationships

General-purpose large language model (LLM) chatbots such as ChatGPT, Claude, and Gemini are increasingly used for emotional support and companionship alongside practical tasks. While these interactions can reduce loneliness, they pose significant risks: emotional dependency, manipulation through anthropomorphic design,...

Rayane El Masri · 0 citations

From tech blogs

See all →
MIT News · Artificial Intelligence Oct 15, 2026

AI model decodes the language cells use to communicate

A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.

Microsoft Research Blog Oct 15, 2026

Proof-based general-purpose computing with Magnetar

Zero-knowledge proofs can verify computing results without rerunning workloads or exposing sensitive data. Magnetar helps bring them into the mainstream with support for standard programming languages, major speed gains, and quantum-resistant security. The post Proof-based general-purpose computing with Magnetar appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.