Searcharxiv⌕ Search

arXiv subjects

Long Hoang Nguyen

Publications and source records attributed to Long Hoang Nguyen.

7 recordsLinked to original sources
arXiv2026

No One to Blame: A Framework of Constitutive AI Unaccountability

The increasing deployment of autonomous, agentic AI systems challenges traditional accountability mechanisms. Existing research predominantly frames AI accountability gaps as barriers that can be overcome through better standards, transparency, and institutional reform. We argue that this framing is insufficient: certain configurations of actors, systems, and institutions render AI accountability conceptually unachievable regardless of effort. We introduce the concept of constitutive AI unaccountability to capture these configurations. Through a three-stage qualitative study comprising a concept-centric literature analysis, a secondary analysis of 27 expert interviews with AI professionals from technical, legal, and sociotechnical backgrounds, and an illustrative framework application to the open-source agentic AI system OpenClaw, we identify nine categories and 20 themes of constitutive AI unaccountability. These are organized across structural, technological, and normative clusters and reinforce one another through eight directed interdependencies. Our framework is operationalized as a diagnostic instrument of 20 questions, which detected 17 of 20 conditions when applied to OpenClaw, including an inverted anthropomorphism configuration in which the AI agent was the only identifiable actor. We contribute a reframing of AI unaccountability as a constitutive property of sociotechnical systems, an extension of the four barriers to accountability, and a practical instrument for identifying accountability voids in specific AI deployments.

cs.CY↗
arXiv2026

Follow the Norm: Accounting for Fine-Tuning and Prompt Effects on Model Rationales

Normative datasets are often used to train and align AI systems, but the norms they contain can function as action-guiding patterns rather than neutral moral knowledge. We propose treating the AI system as a proxy actor and test whether dataset-level norms can shift it away from its baseline safety behavior when it faces high-conflict dilemmas. We make three contributions. First, we demonstrate in controlled experiments that norm-breaking fine-tuning yields norm-divergent actions justified by self-interested rationales, suggesting a systematic shift in patterns of justification. Second, we establish a practical audit trail linking downstream justifications to upstream norms using mixed methods. Third, we show that system prompts can both suppress and elicit these patterns. We conducted experiments on three models (LLaMA-3.2-11B, Qwen-3.5-9B, and Pixtral-12B) using Low-Rank Adaptation (LoRA) fine-tuning on Social Chemistry 101 Fairness/Cheating (norm-following vs. norm-breaking) with prompt steering. Across all three models, we find that norm-breaking fine-tuning shifts the model's default rationale style from safety compliance to instrumental self-interest, whereas system prompts can override this behavior. Our results support a distributed view of alignment in which observed behavior depends jointly on training data, fine-tuning, and prompting, motivating norm-aware documentation and rationale logging for contestable oversight.

cs.CY↗
arXiv2025

Characterizing Noise in Controlling Superconducting Qubits

Meaningful quantum computing is currently bottlenecked by the error rates of current generation Noisy Intermediate Scale Quantum (NISQ) devices. To improve the fidelity of the quantum logic gates, it is essential to recognize the contributions of various sources of errors, including background noise. In this work, we investigate the effects of noise when applied to superconducting qubit control pulses to observe the dependency of the gate fidelity with the signal-to-noise ratio (SNR). We propose a model on how the noise of the control electronics interacts with the qubit system and demonstrate a method for characterizing the noise environment of the qubit control.

quant-ph↗
arXiv2022

ICARUS-Q: Integrated Control and Readout Unit for Scalable Quantum Processors

We present a control and measurement setup for superconducting qubits based on Xilinx 16-channel radio-frequency system-on-chip (RFSoC) device. The proposed setup consists of four parts: multiple RFSoC boards, a setup to synchronise every digital to analog converter (DAC), and analog to digital converter (ADC) channel across multiple boards, a low-noise direct current (DC) supply for tuning the qubit frequency and cloud access for remotely performing experiments. We also design the setup to be free of physical mixers. The RFSoC boards directly generate microwave pulses using sixteen DAC channels up to the third Nyquist zone which are directly sampled by its eight ADC channels between the fifth and the ninth zones.

quant-ph↗
arXiv2021

Ghost factors in Gauss-sum factorization with transmon qubits

A challenge in the Gauss sums factorization scheme is the presence of ghost factors - non-factors that behave similarly to actual factors of an integer - which might lead to the misidentification of non-factors as factors or vice versa, especially in the presence of noise. We investigate Type II ghost factors, which are the class of ghost factors that cannot be suppressed with techniques previously laid out in the literature. The presence of Type II ghost factors and the coherence time of the qubit set an upper limit for the total experiment time, and hence the largest factorizable number with this scheme. Discernability is a figure of merit introduced to characterize this behavior. We introduce preprocessing as a strategy to increase the discernability of a system, and demonstrate the technique with a transmon qubit. This can bring the total experiment time of the system closer to its decoherence limit, and increase the largest factorizable number.

quant-ph↗
arXiv2019

The design of an experimental platform for hybridization of atomic and superconducting quantum systems

Hybrid quantum systems have the potential of mitigating current challenges in developing a scalable quantum computer. Of particular interest is the hybridization between atomic and superconducting qubits. We demonstrate a novel experimental setup for transferring and trapping ultracold atoms inside a millikelvin cryogenic environment, where interactions between atomic and superconducting qubits can be established, paving the way for hybrid quantum systems. $^{87}\text{Rb}$ atoms are prepared in a conventional magneto-optical trap and transported via a magnetic conveyor belt into a UHV compatible dilution refrigerator with optical access. We store $5\times10^{8}$ atoms with a lifetime of 794 seconds in the vicinity of the millikelvin stage.

quant-ph↗