-
How critically can an AI think? A framework for evaluating the quality of thinking of generative artificial intelligence
Authors:
Luke Zaphir,
Jason M. Lodge,
Jacinta Lisec,
Dom McGrath,
Hassan Khosravi
Abstract:
Generative AI such as those with large language models have created opportunities for innovative assessment design practices. Due to recent technological developments, there is a need to know the limits and capabilities of generative AI in terms of simulating cognitive skills. Assessing student critical thinking skills has been a feature of assessment for time immemorial, but the demands of digita…
▽ More
Generative AI such as those with large language models have created opportunities for innovative assessment design practices. Due to recent technological developments, there is a need to know the limits and capabilities of generative AI in terms of simulating cognitive skills. Assessing student critical thinking skills has been a feature of assessment for time immemorial, but the demands of digital assessment create unique challenges for equity, academic integrity and assessment authorship. Educators need a framework for determining their assessments vulnerability to generative AI to inform assessment design practices. This paper presents a framework that explores the capabilities of the LLM ChatGPT4 application, which is the current industry benchmark. This paper presents the Mapping of questions, AI vulnerability testing, Grading, Evaluation (MAGE) framework to methodically critique their assessments within their own disciplinary contexts. This critique will provide specific and targeted indications of their questions vulnerabilities in terms of the critical thinking skills. This can go on to form the basis of assessment design for their tasks.
△ Less
Submitted 20 June, 2024;
originally announced June 2024.
-
Beyond Simple Structure-Function Relationships: Interplay Between Cis/Trans Isomerization and Geometrically Constrained Metal/Molecule Coupling Efficiency in Single-Molecule Junctions
Authors:
Nathan D. Bamberger,
Dylan Dyer,
Keshaba N. Parida,
Tarek H. El-Asssad,
Dawson Pursell,
Dominic V. McGrath,
Manuel Smeu,
Oliver L. A. Monti
Abstract:
Structure-function relationships constitute an important tool to investigate the fundamental principles of molecular electronics. Most commonly, this involves identifying a potentially important molecular structural element, followed by designing and synthesizing a set of related organic molecules, and finally interpretation of their experimental and/or computational quantum transport properties i…
▽ More
Structure-function relationships constitute an important tool to investigate the fundamental principles of molecular electronics. Most commonly, this involves identifying a potentially important molecular structural element, followed by designing and synthesizing a set of related organic molecules, and finally interpretation of their experimental and/or computational quantum transport properties in the light of this structural element. Though this has been extremely powerful in many instances, we demonstrate here the common need for more nuanced relationships even for relatively simple structures, using both experimental and computational results for a series of stilbene derivatives as a case study. In particular, we show that the presence of multiple competing and subtle structural factors can combine in unexpected ways to control quantum transport in these molecules. Our results clarify the reasons for previous widely varying and often contradictory reports on charge-transport in stilbene derivatives, and highlight the need for refined multidimensional structure-property relationships in single molecule electronics.
△ Less
Submitted 4 February, 2022; v1 submitted 31 January, 2022;
originally announced February 2022.
-
Phenalenyls as tunable excellent molecular conductors and switchable spin filters
Authors:
Manuel Smeu,
Oliver L. A. Monti,
Dominic McGrath
Abstract:
We demonstrate a new class of molecules for exceptional performance in molecular electronics and spintronics. Phenalenyl-based radicals are stable radicals whose electronic properties can be tuned readily by heteroatom substitution. We employ density functional theory-based non-equilibrium Green's function (NEGF-DFT) calculations to show that this class of molecules exhibits tunable spin- and char…
▽ More
We demonstrate a new class of molecules for exceptional performance in molecular electronics and spintronics. Phenalenyl-based radicals are stable radicals whose electronic properties can be tuned readily by heteroatom substitution. We employ density functional theory-based non-equilibrium Green's function (NEGF-DFT) calculations to show that this class of molecules exhibits tunable spin- and charge-transport properties in single molecule junctions. Our simulations identify the design principles and interplay between unusually high conductivity and strong spin-filtering: Paired with moderate conductance ($10^{-3} G_0$), two of the four radicals investigated exhibit above 80% spin filter efficiency that is moreover tunable via bias control. Conversely, two radicals that make modest spin filters are excellent conductors with a low bias conductance reaching $0.48 G_0$. This is made possible by the unusually good alignment of the singly occupied or unoccupied molecular orbital with the Fermi level of the electrodes, overcoming the limitations of Fermi level pinning that typically plague molecular electronics. We show that this interplay between excellent conductance and high spin-filter efficiency is determined by the energy alignment between the singly (un)occupied molecular orbital and the Fermi level of the electrodes, and that for phenalenyls this can be readily controlled with judicious heteroatom substitution.
△ Less
Submitted 15 July, 2021; v1 submitted 15 July, 2021;
originally announced July 2021.
-
Grid-Based Correlation Analysis to Identify Rare Quantum Transport Behaviors
Authors:
Nathan D. Bamberger,
Dylan Dyer,
Keshaba N. Parida,
Dominic V. McGrath,
Oliver L. A. Monti
Abstract:
Most single-molecule transport experiments produce large and stochastic datasets containing a wide range of behaviors, presenting both a challenge to their analysis, but also an opportunity for discovering new physical insights. Recently, several unsupervised clustering algorithms have been developed to help extract and separate distinct features from single-molecule transport data. However, these…
▽ More
Most single-molecule transport experiments produce large and stochastic datasets containing a wide range of behaviors, presenting both a challenge to their analysis, but also an opportunity for discovering new physical insights. Recently, several unsupervised clustering algorithms have been developed to help extract and separate distinct features from single-molecule transport data. However, these clustering approaches have been primarily designed and used to extract major dataset components, and are consequently likely to struggle with identifying very rare features and behaviors which may nonetheless contain physically meaningful information. In this work, we thus introduce a completely new analysis framework specifically designed for rare event detection in single-molecule break junction data to help unlock such information and provide a new perspective with different implicit assumptions than clustering. Our approach leverages the concept of correlations of breaking traces with their own history to robustly identify paths through distance-conductance space that correspond to reproducible rare behaviors. As both a demonstrative and important example, we focus on rare conductance plateaus for short molecules, which can be essentially invisible when examining raw data. We show that our grid-based correlation tools successfully and reproducibly locate these rare plateaus in real experimental datasets, including in situations that traditional clustering approaches find challenging. This result enables a broader variety of molecules to be considered in the future, and suggests that our new approach is a useful tool for detecting rare yet meaningful behaviors in single molecule transport data more generally.
△ Less
Submitted 17 August, 2021; v1 submitted 27 May, 2021;
originally announced May 2021.
-
Correlated Energy-Level Alignment Effects Determine Substituent-Tuned Single-Molecule Conductance
Authors:
Jeffrey A. Ivie,
Nathan D. Bamberger,
Keshaba N. Parida,
Stuart Shepard,
Dylan Dyer,
Aldilene Saraiva-Souza,
Roland Himmelhuber,
Dominic V. McGrath,
Manuel Smeu,
Oliver L. A. Monti
Abstract:
The rational design of single molecule electrical components requires a deep and predictive understanding of structure-function relationships. Here we explore the relationship between chemical substituents and the conductance of metal-single molecule-metal junctions, using functionalized oligophenylenevinylenes as a model system. Using a combination of mechanically controlled break-junction experi…
▽ More
The rational design of single molecule electrical components requires a deep and predictive understanding of structure-function relationships. Here we explore the relationship between chemical substituents and the conductance of metal-single molecule-metal junctions, using functionalized oligophenylenevinylenes as a model system. Using a combination of mechanically controlled break-junction experiments and various levels of theory including non-equilibrium Green's functions, we demonstrate that the connection between gas-phase molecular electronic structure and in-junction molecular conductance is complicated by the involvement of multiple mutually correlated and opposing effects that contribute to energy level alignment in the junction. We propose that these opposing correlations represent powerful new "design principles," because their physical origins make them broadly applicable, and they are capable of predicting the direction and relative magnitude of observed conductance trends. In particular, we show that they are consistent with the observed conductance variability not just within our own experimental results, but also within disparate molecular series reported in literature and, crucially, with the trend in variability across these molecular series, which previous simple models fail to explain. The design principles introduced here can therefore aid in both screening and suggesting novel design strategies for maximizing conductance tunability in single-molecule systems.
△ Less
Submitted 16 October, 2020;
originally announced October 2020.
-
Unsupervised Segmentation-Based Machine Learning as an Advanced Analysis Tool for Single Molecule Break Junction Data
Authors:
Nathan D. Bamberger,
Jeffrey A. Ivie,
Keshaba N. Parida,
Dominic V. McGrath,
Oliver L. A. Monti
Abstract:
Improved understanding of charge-transport in single molecules is essential for harnessing the potential of molecules e.g. as circuit components at the ultimate size limit. However, interpretation and analysis of the large, stochastic datasets produced by most quantum transport experiments remains an ongoing challenge to discovering much-needed structure-property relationships. Here, we introduce…
▽ More
Improved understanding of charge-transport in single molecules is essential for harnessing the potential of molecules e.g. as circuit components at the ultimate size limit. However, interpretation and analysis of the large, stochastic datasets produced by most quantum transport experiments remains an ongoing challenge to discovering much-needed structure-property relationships. Here, we introduce Segment Clustering, a novel unsupervised hypothesis generation tool for investigating single molecule break junction distance-conductance traces. In contrast to previous machine learning approaches for single molecule data, Segment Clustering identifies groupings of similar pieces of traces instead of entire traces. This offers a new and advantageous perspective into dataset structure because it facilitates the identification of meaningful local trace behaviors that may otherwise be obscured by random fluctuations over longer distance scales. We illustrate the power and broad applicability of this approach with two case studies that address common challenges encountered in single molecule studies: First, Segment Clustering is used to extract primary molecular features from a varying background to increase the precision and robustness of conductance measurements, enabling small changes in conductance in response to molecular design to be identified with confidence. Second, Segment Clustering is applied to a known data mixture to qualitatively separate distinct molecular features in a rigorous and unbiased manner. These examples demonstrate two powerful ways in which Segment Clustering can aid in the development of structure-property relationships in molecular quantum transport, an outstanding challenge in the field of molecular electronics.
△ Less
Submitted 3 July, 2020; v1 submitted 22 April, 2020;
originally announced April 2020.
-
Student experiences of virtual reality - a case study in learning special relativity
Authors:
D. McGrath,
M. Wegener,
T. J. McIntyre,
C. M. Savage,
M. Williamson
Abstract:
We present a study of student learning through the use of virtual reality. A software package is used to introduce concepts of special relativity to students in a game-like environment where users experience the effects of travelling at near light speeds. From this new perspective, space and time are significantly different to that experienced in everyday life. The study explores how students ha…
▽ More
We present a study of student learning through the use of virtual reality. A software package is used to introduce concepts of special relativity to students in a game-like environment where users experience the effects of travelling at near light speeds. From this new perspective, space and time are significantly different to that experienced in everyday life. The study explores how students have worked with this environment and how these students have used this experience in their study of special relativity. A mixed method approach has been taken to evaluate the outcomes of separate implementations of the package at two universities. Students found the simulation to be a positive learning experience and described the subject area as being less abstract after its use. Also, students were more capable of correctly answering concept questions relating to special relativity, and a small but measurable improvement was observed in the final exam.
△ Less
Submitted 1 November, 2009;
originally announced November 2009.
-
Teaching Physics Using Virtual Reality
Authors:
C. M. Savage,
D. McGrath,
T. J. McIntyre,
M. Wegener,
M. Williamson
Abstract:
We present an investigation of game-like simulations for physics teaching. We report on the effectiveness of the interactive simulation "Real Time Relativity" for learning special relativity. We argue that the simulation not only enhances traditional learning, but also enables new types of learning that challenge the traditional curriculum. The lessons drawn from this work are being applied to t…
▽ More
We present an investigation of game-like simulations for physics teaching. We report on the effectiveness of the interactive simulation "Real Time Relativity" for learning special relativity. We argue that the simulation not only enhances traditional learning, but also enables new types of learning that challenge the traditional curriculum. The lessons drawn from this work are being applied to the development of a simulation for enhancing the learning of quantum mechanics.
△ Less
Submitted 29 October, 2009;
originally announced October 2009.