The rapid emergence of foundation models, particularly Large Language Models (LLMs) and Vision-Language Models (VLMs), has introduced a transformative paradigm in robotics. These models offer powerful capabilities in semantic understanding, high-level reasoning, and cross-modal generalization, enabling significant advances in perception, planning, control, and human-robot interaction. This critical review provides a structured synthesis of recent developments, categorizing applications across simulation-driven design, open-world execution, sim-to-real transfer, and adaptable robotics. Unlike existing surveys that emphasize isolated capabilities, this work highlights integrated, system-level strategies and evaluates their practical feasibility in real-world environments. Key enabling trends such as procedural scene generation, policy generalization, and multimodal reasoning are discussed alongside core bottlenecks, including limited embodiment, lack of multimodal data, safety risks, and computational constraints. Through this lens, this paper identifies both the architectural strengths and critical limitations of foundation model-based robotics, highlighting open challenges in real-time operation, grounding, resilience, and trust. The review concludes with a roadmap for future research aimed at bridging semantic reasoning and physical intelligence through more robust, interpretable, and embodied models.

Recent large language models have shown promising capabilities in long-form reasoning, following structured chains of thought before arriving at a final answer. However, we observe that these reasoning paths tend to include substantial redundancy; analyzing attention patterns reveals that attention scores are widely scattered, particularly incorrect answers exhibit greater attention sparsity. In this paper, we demonstrate that deliberately removing this redundancy in the reasoning process significantly improves performance through clear thinking, i.e., removing distraction. Specifically, we systematically identify reasoning redundancy by measuring token-level attention scores to a special end-of-thinking token, which is appended to an explicit instruction inserted to conclude each intermediate reasoning step. Furthermore, we propose structure-aware pruning that prioritizes removing tokens in low-contributing reasoning chunks over individual tokens. After evicting redundant tokens, we remove the injected end-of-thinking instruction, then resume the reasoning generation. We demonstrate that our method significantly improves overall accuracy across reasoning-intensive benchmarks without any training involved. In particular, our method shows strong performance on challenging mathematical competition benchmarks such as AIME and AMC, where reasoning redundancy is more prevalent.

Enhancing reasoning capabilities remains a central focus in the LLM reasearch community. A promising direction involves requiring models to simulate code execution step-by-step to derive outputs for given inputs. However, as code is often designed for large-scale systems, direct application leads to over-reliance on complex data structures and algorithms, even for simple cases, resulting in overfitting to algorithmic patterns rather than core reasoning structures. To address this, we propose TeaR, which aims at teaching LLMs to reason better. TeaR leverages careful data curation and reinforcement learning to guide models in discovering optimal reasoning paths through code-related tasks, thereby improving general reasoning abilities. We conduct extensive experiments using two base models and three long-CoT distillation models, with model sizes ranging from 1.5 billion to 32 billion parameters, and across 17 benchmarks spanning Math, Knowledge, Code, and Logical Reasoning. The results consistently show significant performance improvements. Notably, TeaR achieves a 35.9% improvement on Qwen2.5-7B and 5.9% on R1-Distilled-7B.

Language models trained with reinforcement learning (RL) can engage in reward hacking--the exploitation of unintended strategies for high reward--without revealing this behavior in their chain-of-thought reasoning. This makes the detection of reward hacking difficult, posing risks for high-stakes applications. We propose verbalization fine-tuning (VFT), a pre-RL fine-tuning intervention that trains models to explicitly acknowledge when they are influenced by prompt cues--hints which point to incorrect answers (e.g., "a Stanford professor thinks the answer is A"). To evaluate VFT, we subsequently train models with RL on environments where held-out prompt cues signal which incorrect answers will receive high reward, incentivizing models to exploit these cues instead of reasoning correctly. We measure how often models exploit these cues without verbalizing it. After RL, only 6% of the VFT-trained model's responses consist of undetected reward hacks. In comparison, when we perform RL without VFT, the rate of undetected reward hacks goes up to 88%; with a debiasing baseline intervention, this increases further to 99%. VFT achieves this by substantially increasing how often models verbalize the influence of cues, from 8% to 43% after VFT, and up to 94% after RL. Baselines remain low even after RL (11% and 1%). Our results show that teaching models to explicitly verbalize reward hacking behavior before RL significantly improves their detection, offering a practical path toward more transparent and safe AI systems.

Steering vectors are a lightweight method for controlling text properties by adding a learned bias to language model activations at inference time. So far, steering vectors have predominantly been evaluated in multiple-choice settings, while their effectiveness in free-form generation tasks remains understudied. Moving "Beyond Multiple Choice," we thoroughly evaluate the effectiveness of steering vectors in adaptively controlling topical focus, sentiment, toxicity, and readability in abstractive summaries of the NEWTS dataset. We find that steering effectively controls the targeted summary properties, but high steering strengths consistently degrade both intrinsic and extrinsic text quality. Compared to steering, prompting offers weaker control, while preserving text quality. Combining steering and prompting yields the strongest control over text properties and offers the most favorable efficacy-quality trade-off at moderate steering strengths. Our results underscore the practical trade-off between control strength and text quality preservation when applying steering vectors to free-form generation tasks.

The advancement of large language models (LLMs) has outpaced traditional evaluation methodologies. This progress presents novel challenges, such as measuring human-like psychological constructs, moving beyond static and task-specific benchmarks, and establishing human-centered evaluation. These challenges intersect with psychometrics, the science of quantifying the intangible aspects of human psychology, such as personality, values, and intelligence. This review paper introduces and synthesizes the emerging interdisciplinary field of LLM Psychometrics, which leverages psychometric instruments, theories, and principles to evaluate, understand, and enhance LLMs. The reviewed literature systematically shapes benchmarking principles, broadens evaluation scopes, refines methodologies, validates results, and advances LLM capabilities. Diverse perspectives are integrated to provide a structured framework for researchers across disciplines, enabling a more comprehensive understanding of this nascent field. Ultimately, the review provides actionable insights for developing future evaluation paradigms that align with human-level AI and promote the advancement of human-centered AI systems for societal benefit. A curated repository of LLM psychometric resources is available at https://github.com/valuebyte-ai/Awesome-LLM-Psychometrics.

Natural language-based assessment (NLA) is an approach to second language assessment that uses instructions - expressed in the form of can-do descriptors - originally intended for human examiners, aiming to determine whether large language models (LLMs) can interpret and apply them in ways comparable to human assessment. In this work, we explore the use of such descriptors with an open-source LLM, Qwen 2.5 72B, to assess responses from the publicly available S&I Corpus in a zero-shot setting. Our results show that this approach - relying solely on textual information - achieves competitive performance: while it does not outperform state-of-the-art speech LLMs fine-tuned for the task, it surpasses a BERT - based model trained specifically for this purpose. NLA proves particularly effective in mismatched task settings, is generalis-able to other data types and languages, and offers greater in-terpretability, as it is grounded in clearly explainable, widely applicable language descriptors.

AI has driven significant progress in the nutrition field, especially through multimedia-based automatic dietary assessment. However, existing automatic dietary assessment systems often overlook critical non-visual factors, such as recipe-specific ingredient substitutions that can significantly alter nutritional content, and rarely account for individual dietary needs, including allergies, restrictions, cultural practices, and personal preferences. In Switzerland, while food-related information is available, it remains fragmented, and no centralized repository currently integrates all relevant nutrition-related aspects within a Swiss context. To bridge this divide, we introduce the Swiss Food Knowledge Graph (SwissFKG), the first resource, to our best knowledge, to unite recipes, ingredients, and their substitutions with nutrient data, dietary restrictions, allergen information, and national nutrition guidelines under one graph. We establish a LLM-powered enrichment pipeline for populating the graph, whereby we further present the first benchmark of four off-the-shelf (<70 B parameter) LLMs for food knowledge augmentation. Our results demonstrate that LLMs can effectively enrich the graph with relevant nutritional information. Our SwissFKG goes beyond recipe recommendations by offering ingredient-level information such as allergen and dietary restriction information, and guidance aligned with nutritional guidelines. Moreover, we implement a Graph-RAG application to showcase how the SwissFKG's rich natural-language data structure can help LLM answer user-specific nutrition queries, and we evaluate LLM-embedding pairings by comparing user-query responses against predefined expected answers. As such, our work lays the foundation for the next generation of dietary assessment tools that blend visual, contextual, and cultural dimensions of eating.

Large Language Models (LLM) have achieved remarkable performance across a large number of tasks, but face critical deployment and usage barriers due to substantial computational requirements. Model compression methods, which aim to reduce model size while preserving its capacity, are an important means to mitigate these issues. Promising approaches along these lines, such as structured pruning, typically require costly empirical search for optimal variants and may run the risk of ignoring better solutions. In this work we introduce GeLaCo, an evolutionary approach to LLM compression via layer collapse. Our approach supports an efficient exploration of the compression solution space via population-based search and a module-wise similarity fitness function capturing attention, feed-forward, and hidden state representations. GeLaCo also supports both single and multi-objective evolutionary compression search, establishing the first Pareto frontier along compression and quality axes. We evaluate GeLaCo solutions via both perplexity-based and generative evaluations over foundational and instruction-tuned models, outperforming state-of-the-art alternatives.

Neurosymbolic artificial intelligence (AI) systems combine neural network and classical symbolic AI mechanisms to exploit the complementary strengths of large-scale, generaliz-able learning and robust, verifiable reasoning. Numerous classifications of neurosymbolic AI illustrate how these two components can be integrated in distinctly different ways. In this work, we propose reinterpreting instruction-tuned large language models as model-grounded symbolic AI systems --where natural language serves as the symbolic layer, and grounding is achieved through the model's internal representation space. Within this framework, we investigate and develop novel learning and reasoning approaches that preserve structural similarities to traditional learning and reasoning paradigms. Preliminary evaluations across axiomatic deductive reasoning procedure of varying complexity provides insights into the effectiveness of our approach towards learning efficiency and reasoning reliability.