arXiv subjects

John H. Miller

Publications and source records attributed to John H. Miller.

5 recordsLinked to original sources

arXiv2025

Detecting AI Assistance in Abstract Complex Tasks

Detecting assistance from artificial intelligence is increasingly important as they become ubiquitous across complex tasks such as text generation, medical diagnosis, and autonomous driving. Aid detection is challenging for humans, especially when looking at abstract task data. Artificial neural networks excel at classification thanks to their ability to quickly learn from and process large amounts of data -- assuming appropriate preprocessing. We posit detecting help from AI as a classification task for such models. Much of the research in this space examines the classification of complex but concrete data classes, such as images. Many AI assistance detection scenarios, however, result in data that is not machine learning-friendly. We demonstrate that common models can effectively classify such data when it is appropriately preprocessed. To do so, we construct four distinct neural network-friendly image formulations along with an additional time-series formulation that explicitly encodes the exploration/exploitation of users, which allows for generalizability to other abstract tasks. We benchmark the quality of each image formulation across three classical deep learning architectures, along with a parallel CNN-RNN architecture that leverages the additional time series to maximize testing performance, showcasing the importance of encoding temporal and spatial quantities for detecting AI aid in abstract tasks.

cs.AI↗

arXiv2023

The Role of Heuristics and Biases During Complex Choices with an AI Teammate

Behavioral scientists have classically documented aversion to algorithmic decision aids, from simple linear models to AI. Sentiment, however, is changing and possibly accelerating AI helper usage. AI assistance is, arguably, most valuable when humans must make complex choices. We argue that classic experimental methods used to study heuristics and biases are insufficient for studying complex choices made with AI helpers. We adapted an experimental paradigm designed for studying complex choices in such contexts. We show that framing and anchoring effects impact how people work with an AI helper and are predictive of choice outcomes. The evidence suggests that some participants, particularly those in a loss frame, put too much faith in the AI helper and experienced worse choice outcomes by doing so. The paradigm also generates computational modeling-friendly data allowing future studies of human-AI decision making.

cs.AI↗

arXiv2018

Transport evidence for three dimensional topological superconductivity in doped $β$-PdBi$_2$

Interest in topological states of matter exploded over a decade ago with the theoretical prediction and experimental detection of three-dimensional topological insulators, especially in bulk materials that can be tuned out of it by doping. However, their superconducting counterpart, the time-reversal invariant three-dimensional topological superconductor, has evaded discovery thus far. In this work, we provide transport evidence that K-doped $β$-PdBi$_2$ is a 3D time-reversal-invariant topological superconductor. In particular, we find signatures of Majorana surface states protected by time-reversal symmetry--the hallmark of this phase--in soft point-contact spectroscopy, while the bulk system shows signatures of odd-parity pairing via upper-critical field and magnetization measurements. Odd-parity pairing can be argued, using existing knowledge of the band structure of $β$-PdBi$_2$, to result in 3D topological superconductivity. Moreover, we find that the undoped system is a trivial superconductor. Thus, we discover $β$-PdBi$_2$ as a unique material that, on doping, can potentially undergo an unprecedented topological quantum phase transition in the superconducting state.

cond-mat.supr-con↗

arXiv2016

Paradoxes in Leaky Microbial Trade

Microbes produce metabolic resources that are important for cell growth yet leak across membranes into the extracellular environment. Other microbes in the same environment can use these resources and adjust their own metabolic production accordingly---causing other resources to leak into the environment. The combined effect of these processes is an economy in which organismal growth and metabolic production are coupled to others in the community. We propose a model for the co-evolving dynamics of metabolite concentrations, production regulation, and population frequencies for the case of two cell types, each requiring and capable of producing two metabolites. In this model, beneficial trade relations emerge without any coordination, via individual-level production decisions that maximize each cell's growth rate given its perceived environment. As we vary production parameters of the model, we encounter three paradoxical behaviors, where a change that should intuitively benefit some cell type, actually harms it. (1) If a cell type is more efficient than its counterpart at producing a metabolite and becomes even more efficient, its frequency in the population can decrease. (2) If a cell type is less efficient than its counterpart at producing a metabolite but becomes less inefficient, the growth rate of the population can decrease. (3) Finally, if a cell type controls its counterpart's production decisions so as to maximize its own growth rate, the ultimate growth rate it achieves can be lower than if the two cell types each maximized their own growth. These three paradoxes highlight the complex and counter-intuitive dynamics that emerge in simple microbial economies.

q-bio.PE↗

arXiv2005

Towards Experimental Tests of Quantum Effects in Cytoskeletal Proteins

It has become increasingly evident that fabrication of novel biomaterials through molecular self-assembly is going to play a significant role in material science and possibly the information technology of the future. Tubulin, microtubules (MTs) and the cytoskeleton are dynamic, self-assembling systems and we asked whether their structure and function contain the clues on how to fabricate biomolecular information processing devices. Here we review our neurobiological studies of transgenic Drosophila that strongly suggest the microtubular cytoskeleton is near the 'front lines' of intracellular information manipulation and storage. We also establish that spectroscopic techniques such as refractometry, surface plasmon resonance sensing and dielectric spectroscopy, coupled with molecular dynamic simulations and (quantum) electrodynamic analytical theory are useful tools in the study of the electrodynamic and possible quantum effects in cytoskeletal proteins. Implicit in our driving question is the possibility that if tubulin and MTs can indeed be cast as the basis of a classical or quantum computer, then perhaps nature has already done so and tubulin and MTs already play such a role in living neural and other cells. If quantum mechanics is found to be important in cellular function (through its involvement in proteins) it is natural to ask whether there are indeed quantum effects pertinent to consciousness. We conclude this work by suggesting a number of ways towards experimentally testing this "Quantum Consciousness Idea" (QCI), directly at the protein level.

physics.bio-ph↗