Latest preprints

This research evaluates the extent to which modern AI frameworks, specifically TensorFlow and scikit-learn, adhere to the SOLID design principles—Single Responsibility, Open/Closed, Liskov Substitution, Interface Segregation, and Dependency Inversion. Analyzing the frameworks' architectural documentation and design philosophies, this research investigates architectural trade-offs when balancing software engineering best practices with AI-specific needs. I examined each framework's documentation, source code, and architectural components to evaluate their adherence to these principles. The results show that both frameworks adopt certain aspects of SOLID design principles but make intentional trade-offs to address performance, scalability, and the experimental nature of AI development. TensorFlow focuses on performance and scalability, sometimes sacrificing strict adherence to principles like Single Responsibility and Interface Segregation. While scikit-learn’s design philosophy aligns more closely with SOLID principles through consistent interfaces and composition principles, it sticks closer to SOLID guidelines but with occasional deviations for performance optimizations and scalability. This research discovered that applying SOLID principles in AI frameworks depends on context, as performance, scalability, and flexibility often require deviations from traditional software engineering principles. This research contributes to understanding how domain-specific constraints influence architectural decisions in modern AI frameworks and how these frameworks strategically adapted design choices to effectively balance these contradicting requirements.

This paper proposes a novel approach to analyzing multi-hop reasoning in language models through Hamiltonian mechanics. We map reasoning chains in embedding spaces to Hamiltonian systems, defining a function that balances reasoning progression (kinetic energy) against question relevance (potential energy). Analyzing reasoning chains from a question-answering dataset reveals that valid reasoning shows lower Hamiltonian energy values, representing an optimal trade-off between information gathering and targeted answering. While our framework offers complex visualization and quantification methods, the claimed ability to ”steer” or ”improve” reasoning algorithms requires more rigorous empirical validation, as the connection between physical systems and reasoning remains largely metaphorical. Nevertheless, our analysis reveals consistent geometric patterns distinguishing valid reasoning, suggesting this physics-inspired approach offers promising diagnostic tools and new perspectives on reasoning processes in large language models.

Domain generalization (DG) aims to learn models that can generalize well to unseen domains by training only on a set of source domains. Sharpness-Aware Minimization (SAM) has been a popular approach for this, aiming to find flat minima in the total loss landscape. However, we show that minimizing the total loss sharpness does not guarantee sharpness across individual domains. In particular, SAM can converge to fake flat minima, where the total loss may exhibit flat minima, but sharp minima are present in individual domains. Moreover, the current perturbation update in gradient ascent steps is ineffective in directly updating the sharpness of individual domains. Motivated by these findings, we introduce a novel DG algorithm, Decreased-overhead Gradual Sharpness-Aware Minimization (DGSAM), that applies gradual domain-wise perturbation to reduce sharpness consistently across domains while maintaining computational efficiency. Our experiments demonstrate that DGSAM outperforms state-of-the-art DG methods, achieving improved robustness to domain shifts and better performance across various benchmarks, while reducing computational overhead compared to SAM.

We introduce Z-SASLM, a Zero-Shot Style-Aligned SLI (Spherical Linear Interpolation) Blending Latent Manipulation pipeline that overcomes the limitations of current multi-style blending methods. Conventional approaches rely on linear blending, assuming a flat latent space leading to suboptimal results when integrating multiple reference styles. In contrast, our framework leverages the non-linear geometry of the latent space by using SLI Blending to combine weighted style representations. By interpolating along the geodesic on the hypersphere, Z-SASLM preserves the intrinsic structure of the latent space, ensuring high-fidelity and coherent blending of diverse styles—all without the need for fine-tuning. We further propose a new metric, Weighted Multi-Style DINO VIT-B/8, designed to quantitatively evaluate the consistency of the blended styles. While our primary focus is on the theoretical and practical advantages of SLI Blending for style manipulation, we also demonstrate its effectiveness in a multi-modal content fusion setting through comprehensive experimental studies. Experimental results show that Z-SASLM achieves enhanced and robust style alignment. The implementation code can be found at: https://github.com/alessioborgi/Z-SASLM.

AI for software engineering has made remarkable progress recently, becoming a notable success within generative AI. Despite this, there are still many challenges that need to be addressed before automated software engineering reaches its full potential. It should be possible to reach high levels of automation where humans can focus on the critical decisions of what to build and how to balance difficult tradeoffs while most routine development effort is automated away. Reaching this level of automation will require substantial research and engineering efforts across academia and industry. In this paper, we aim to discuss progress towards this in a threefold manner. First, we provide a structured taxonomy of concrete tasks in AI for software engineering, emphasizing the many other tasks in software engineering beyond code generation and completion. Second, we outline several key bottlenecks that limit current approaches. Finally, we provide an opinionated list of promising research directions toward making progress on these bottlenecks, hoping to inspire future research in this rapidly maturing field.

The trajectory of intelligence evolution is often framed around the emergence of artificial general intelligence (AGI) and its alignment with human values. This paper challenges that framing by introducing the concept of intelligence sequencing: the idea that the order in which AGI and decentralized collective intelligence (DCI) emerge determines the long-term attractor basin of intelligence. Using insights from dynamical systems, evolutionary game theory, and network models, it argues that intelligence follows a path-dependent, irreversible trajectory. Once development enters a centralized (AGI-first) or decentralized (DCI-first) regime, transitions become structurally infeasible due to feedback loops and resource lock-in. Intelligence attractors are modeled in functional state space as the co-navigation of conceptual and adaptive fitness spaces. Early-phase structuring constrains later dynamics, much like renormalization in physics. This has major implications for AI safety: traditional alignment assumes AGI will emerge and must be controlled after the fact, but this paper argues that intelligence sequencing is more foundational. If AGI-first architectures dominate before DCI reaches critical mass, hierarchical monopolization and existential risk become locked in. If DCI-first emerges, intelligence stabilizes around decentralized cooperative equilibrium. The paper further explores whether intelligence structurally biases itself toward an attractor based on its self-modeling method---externally imposed axioms (favoring AGI) vs. recursive internal visualization (favoring DCI). Finally, it proposes methods to test this theory via simulations, historical lock-in case studies, and intelligence network analysis. The findings suggest that intelligence sequencing is a civilizational tipping point: determining whether the future is shaped by unbounded competition or unbounded cooperation.

The relationship between metacognition and DeepSeek models represents a compelling and yet underexplored area of research. Metacognition refers to a system's capacity to monitor and comprehend its own cognitive processes, which includes the active regulation and adjustment of its procedures. In the DeepSeek-R1 and DeepSeek-R1-Zero, it is evident that the interactions between the system's monitoring and control processes are both present and crucial for achieving the coherent and often surprising levels of reasoning that define these models. Consequently, DeepSeek paves a new avenue for reasoning in Artificial Intelligence (AI) by prioritizing reinforcement learning (RL) over the more conventional approach of supervised fine-tuning (SFT). This study aims to analyse the implications of such innovations for machines' abilities to simulate behaviours based on self-reflection and to act accordingly. We will explore the extent to which these elements can be associated with metacognition, a trait traditionally considered to be uniquely human. Furthermore, from an educational standpoint, the significance of the relationship between advancements in DeepSeek and metacognition is highlighted, particularly in relation to the importance of prioritizing metacognitive approaches in education.

The research study explores the transformative role of AI technologies in enhancing technical aspects like editing, lighting, visual effects (VFX), computer-generated imagery (CGI), and sound design in filmmaking and cinema production. Through a comprehensive qualitative comparative analysis, the study examines 30 selected films from Hollywood, such as I, Robot (2004) and The Creator (2023), the Asian cinema industry, including After Yang (2021) and Jung-E (2023), and winners of the Artificial Intelligence Film Festival (AIFF) like PLSTC (2023), Expanded Childhood (2023), E^(IPi) + 1=0* (2024), and Dear Mom (2024). This analysis highlights significant transformations in pre-production, production, and post-production processes. The study traces the evolution from early CGI developments, manual editing, and complex simulations to AI-driven techniques and tools. These advancements include real-time rendering, automated content recognition, intelligent cut detection, real-time light adjustment, AI-driven storyboarding, predictive editing, AI-based motion capture, realistic particle simulation, deep-learning for texture synthesis, enhanced 3D modeling, CGI neural rendering, and AI-powered audio environments. Ethical considerations such as balancing creativity between human and AI technologies, intellectual property rights, and data privacy are also discussed. The study suggests developing frameworks and guidelines to ensure the best implementation of AI in cinema production. However, the research is limited by the number of films analyzed, the time frame, lack of quantitative methodologies, and challenges in conducting interviews with stakeholders. The findings aim to contribute to understanding AI's integration in cinema production by analyzing technical aspects, possibilities, limitations, guidelines, and ethical considerations.