Reinforcement learning-based post-training of large language models (LLMs) has recently gained attention, particularly following the release of DeepSeek R1, which applied GRPO for fine-tuning. Amid the growing hype around improved reasoning abilities attributed to RL post-training, we critically examine the formulation and assumptions underlying these methods. We start by highlighting the popular structural assumptions made in modeling LLM training as a Markov Decision Process (MDP), and show how they lead to a degenerate MDP that doesn't quite need the RL/GRPO apparatus. The two critical structural assumptions include (1) making the MDP states be just a concatenation of the actions-with states becoming the context window and the actions becoming the tokens in LLMs and (2) splitting the reward of a state-action trajectory uniformly across the trajectory. Through a comprehensive analysis, we demonstrate that these simplifying assumptions make the approach effectively equivalent to an outcome-driven supervised learning. Our experiments on benchmarks including GSM8K and Countdown using Qwen-2.5 base models show that iterative supervised fine-tuning, incorporating both positive and negative samples, achieves performance comparable to GRPO-based training. We will also argue that the structural assumptions indirectly incentivize the RL to generate longer sequences of intermediate tokens-which in turn feeds into the narrative of "RL generating longer thinking traces." While RL may well be a very useful technique for improving the reasoning abilities of LLMs, our analysis shows that the simplistic structural assumptions made in modeling the underlying MDP render the popular LLM RL frameworks and their interpretations questionable.

Bayesian networks and causal models provide frameworks for handling queries about external interventions and counterfactuals, enabling tasks that go beyond what probability distributions alone can address. While these formalisms are often informally described as capturing causal knowledge, there is a lack of a formal theory characterizing the type of knowledge required to predict the effects of external interventions. This work introduces the theoretical framework of causal systems to clarify Aristotle's distinction between knowledge that and knowledge why within artificial intelligence. By interpreting existing artificial intelligence technologies as causal systems, it investigates the corresponding types of knowledge. Furthermore, it argues that predicting the effects of external interventions is feasible only with knowledge why, providing a more precise understanding of the knowledge necessary for such tasks.

Bayesian optimal experimental design is a principled framework for conducting experiments that leverages Bayesian inference to quantify how much information one can expect to gain from selecting a certain design. However, accurate Bayesian inference relies on the assumption that one's statistical model of the data-generating process is correctly specified. If this assumption is violated, Bayesian methods can lead to poor inference and estimates of information gain. Generalised Bayesian (or Gibbs) inference is a more robust probabilistic inference framework that replaces the likelihood in the Bayesian update by a suitable loss function. In this work, we present Generalised Bayesian Optimal Experimental Design (GBOED), an extension of Gibbs inference to the experimental design setting which achieves robustness in both design and inference. Using an extended information-theoretic framework, we derive a new acquisition function, the Gibbs expected information gain (Gibbs EIG). Our empirical results demonstrate that GBOED enhances robustness to outliers and incorrect assumptions about the outcome noise distribution.

Hierarchical Bayesian models based on Gaussian processes a re considered useful for describing complex nonlinear statistical dependen cies among variables in real-world data. However, effective Monte Carlo algorithm s for inference with these models have not yet been established, except for sever al simple cases. In this study, we show that, compared with the slow inference ac hieved with existing program libraries, the performance of Riemannian-m anifold Hamiltonian Monte Carlo (RMHMC) can be drastically improved by optimisi ng the computation order according to the model structure and dynamical ly programming the eigendecomposition. This improvement cannot be achieved w hen using an existing library based on a naive automatic differentiator. W e nu merically demonstrate that RMHMC effectively samples from the posterior, allowin g the calculation of model evidence, in a Bayesian logistic regression on simula ted data and in the estimation of propensity functions for the American nation al medical expenditure data using several Bayesian multiple-kernel models. These results lay a foundation for implementing effective Monte Carlo algorithms for analysing real-world data with Gaussian processes, and highlight the need to deve lop a customisable library set that allows users to incorporate dynamically pr ogrammed objects and finely optimises the mode of automatic differentiation depe nding on the model structure.

Uncertainty quantification (UQ) for adaptively collected data, such as that coming from adaptive experiments, bandits, or reinforcement learning, is necessary for critical elements of data collection such as ensuring safety and conducting after-study inference. The data's adaptivity creates significant challenges for frequentist UQ, yet Bayesian UQ remains the same as if the data were independent and identically distributed (i.i.d.), making it an appealing and commonly used approach. Bayesian UQ requires the (correct) specification of a prior distribution while frequentist UQ does not, but for i.i.d. data the celebrated Bernstein-von Mises theorem shows that as the sample size grows, the prior 'washes out' and Bayesian UQ becomes frequentist-valid, implying that the choice of prior need not be a major impediment to Bayesian UQ as it makes no difference asymptotically. This paper for the first time extends the Bernstein-von Mises theorem to adaptively collected data, proving asymptotic equivalence between Bayesian UQ and Wald-type frequentist UQ in this challenging setting. Our result showing this asymptotic agreement does not require the standard stability condition required by works studying validity of Wald-type frequentist UQ; in cases where stability is satisfied, our results combined with these prior studies of frequentist UQ imply frequentist validity of Bayesian UQ. Counterintuitively however, they also provide a negative result that Bayesian UQ is not asymptotically frequentist valid when stability fails, despite the fact that the prior washes out and Bayesian UQ asymptotically matches standard Wald-type frequentist UQ. We empirically validate our theory (positive and negative) via a range of simulations.

Supply chain disruptions and volatile demand pose significant challenges to the UK automotive industry, which relies heavily on Just-In-Time (JIT) manufacturing. While qualitative studies highlight the potential of integrating Artificial Intelligence (AI) with traditional optimization, a formal, quantitative demonstration of this synergy is lacking. This paper introduces a novel stochastic learning-optimization framework that integrates Bayesian inference with inventory optimization for supply chain management (SCM). We model a two-echelon inventory system subject to stochastic demand and supply disruptions, comparing a traditional static optimization policy against an adaptive policy where Bayesian learning continuously updates parameter estimates to inform stochastic optimization. Our simulations over 365 periods across three operational scenarios demonstrate that the integrated approach achieves 7.4\% cost reduction in stable environments and 5.7\% improvement during supply disruptions, while revealing important limitations during sudden demand shocks due to the inherent conservatism of Bayesian updating. This work provides mathematical validation for practitioner observations and establishes a formal framework for understanding AI-driven supply chain resilience, while identifying critical boundary conditions for successful implementation.

This paper presents a comprehensive analysis of hyperparameter estimation within the empirical Bayes framework (EBF) for sparse learning. By studying the influence of hyperpriors on the solution of EBF, we establish a theoretical connection between the choice of the hyperprior and the sparsity as well as the local optimality of the resulting solutions. We show that some strictly increasing hyperpriors, such as half-Laplace and half-generalized Gaussian with the power in $(0,1)$, effectively promote sparsity and improve solution stability with respect to measurement noise. Based on this analysis, we adopt a proximal alternating linearized minimization (PALM) algorithm with convergence guaranties for both convex and concave hyperpriors. Extensive numerical tests on two-dimensional image deblurring problems demonstrate that introducing appropriate hyperpriors significantly promotes the sparsity of the solution and enhances restoration accuracy. Furthermore, we illustrate the influence of the noise level and the ill-posedness of inverse problems to EBF solutions.

Transformers have shown strong ability to model long-term dependencies and are increasingly adopted as world models in model-based reinforcement learning (RL) under partial observability. However, unlike natural language corpora, RL trajectories are sparse and reward-driven, making standard self-attention inefficient because it distributes weight uniformly across all past tokens rather than emphasizing the few transitions critical for control. To address this, we introduce structured inductive priors into the self-attention mechanism of the dynamics head: (i) per-head memory-length priors that constrain attention to task-specific windows, and (ii) distributional priors that learn smooth Gaussian weightings over past state-action pairs. We integrate these mechanisms into UniZero, a model-based RL agent with a Transformer-based world model that supports planning under partial observability. Experiments on the Atari 100k benchmark show that most efficiency gains arise from the Gaussian prior, which smoothly allocates attention to informative transitions, while memory-length priors often truncate useful signals with overly restrictive cut-offs. In particular, Gaussian Attention achieves a 77% relative improvement in mean human-normalized scores over UniZero. These findings suggest that in partially observable RL domains with non-stationary temporal dependencies, discrete memory windows are difficult to learn reliably, whereas smooth distributional priors flexibly adapt across horizons and yield more robust data efficiency. Overall, our results demonstrate that encoding structured temporal priors directly into self-attention improves the prioritization of informative histories for dynamics modeling under partial observability.

In this work, we study the problem of finding robust and safe policies in Robust Constrained Average-Cost Markov Decision Processes (RCMDPs). A key challenge in this setting is the lack of strong duality, which prevents the direct use of standard primal-dual methods for constrained RL. Additional difficulties arise from the average-cost setting, where the Robust Bellman operator is not a contraction under any norm. To address these challenges, we propose an actor-critic algorithm for Average-Cost RCMDPs. We show that our method achieves both \(ε\)-feasibility and \(ε\)-optimality, and we establish a sample complexities of \(\tilde{O}\left(ε^{-4}\right)\) and \(\tilde{O}\left(ε^{-6}\right)\) with and without slackness assumption, which is comparable to the discounted setting.

The complexity of human cognition has meant that psychology makes more use of theory and conceptual models than perhaps any other biomedical field. To enable precise quantitative study of the full breadth of phenomena in psychological and psychiatric medicine as well as cognitive aspects of AI safety, there is a need for a mathematical formulation which is both mathematically precise and equally accessible to experts from numerous fields. In this paper we formalize human psychodynamics via the diagrammatic framework of process theory, describe its key properties, and explain the links between a diagrammatic representation and central concepts in analysis of cognitive processes in contexts such as psychotherapy, neurotechnology, AI alignment, AI agent representation of individuals in autonomous negotiations, developing human-like AI systems, and other aspects of AI safety.