Showing 1–7 of 7 results for author: Konya, A

Deliberative Technology for Alignment

Authors: Andrew Konya, Deger Turan, Aviv Ovadya, Lina Qui, Daanish Masood, Flynn Devine, Lisa Schirch, Isabella Roberts, Deliberative Alignment Forum

Abstract: For humanity to maintain and expand its agency into the future, the most powerful systems we create must be those which act to align the future with the will of humanity. The most powerful systems today are massive institutions like governments, firms, and NGOs. Deliberative technology is already being used across these institutions to help align governance and diplomacy with human will, and moder… ▽ More For humanity to maintain and expand its agency into the future, the most powerful systems we create must be those which act to align the future with the will of humanity. The most powerful systems today are massive institutions like governments, firms, and NGOs. Deliberative technology is already being used across these institutions to help align governance and diplomacy with human will, and modern AI is poised to make this technology significantly better. At the same time, the race to superhuman AGI is already underway, and the AI systems it gives rise to may become the most powerful systems of the future. Failure to align the impact of such powerful AI with the will of humanity may lead to catastrophic consequences, while success may unleash abundance. Right now, there is a window of opportunity to use deliberative technology to align the impact of powerful AI with the will of humanity. Moreover, it may be possible to engineer a symbiotic coupling between powerful AI and deliberative alignment systems such that the quality of alignment improves as AI capabilities increase. △ Less

Submitted 6 December, 2023; originally announced December 2023.

Democratic Policy Development using Collective Dialogues and AI

Authors: Andrew Konya, Lisa Schirch, Colin Irwin, Aviv Ovadya

Abstract: We design and test an efficient democratic process for develo** policies that reflect informed public will. The process combines AI-enabled collective dialogues that make deliberation democratically viable at scale with bridging-based ranking for automated consensus discovery. A GPT4-powered pipeline translates points of consensus into representative policy clauses from which an initial policy i… ▽ More We design and test an efficient democratic process for develo** policies that reflect informed public will. The process combines AI-enabled collective dialogues that make deliberation democratically viable at scale with bridging-based ranking for automated consensus discovery. A GPT4-powered pipeline translates points of consensus into representative policy clauses from which an initial policy is assembled. The initial policy is iteratively refined with the input of experts and the public before a final vote and evaluation. We test the process three times with the US public, develo** policy guidelines for AI assistants related to medical advice, vaccine information, and wars & conflicts. We show the process can be run in two weeks with 1500+ participants for around $10,000, and that it generates policy guidelines with strong public support across demographic divides. We measure 75-81% support for the policy guidelines overall, and no less than 70-75% support across demographic splits spanning age, gender, religion, race, education, and political party. Overall, this work demonstrates an end-to-end proof of concept for a process we believe can help AI labs develop common-ground policies, governing bodies break political gridlock, and diplomats accelerate peace deals. △ Less

Submitted 3 November, 2023; originally announced November 2023.

Comments: Report produced as part of OpenAI Democratic inputs to AI grant program (2023)

Faster Peace via Inclusivity: An Efficient Paradigm to Understand Populations in Conflict Zones

Authors: Jordan Bilich, Michael Varga, Daanish Masood, Andrew Konya

Abstract: United Nations practice shows that inclusivity is vital for mediation to be successful in hel** end violent conflict and establish lasting peace. However, current methods for understanding the views and needs of populations during dynamic situations create tension between inclusivity and efficiency. This work introduces a novel paradigm to mitigate such tension. In partnership with collaborators… ▽ More United Nations practice shows that inclusivity is vital for mediation to be successful in hel** end violent conflict and establish lasting peace. However, current methods for understanding the views and needs of populations during dynamic situations create tension between inclusivity and efficiency. This work introduces a novel paradigm to mitigate such tension. In partnership with collaborators at the United Nations we develop a realtime large-scale synchronous dialogue process (RLSDP) to understand stakeholder populations on an hour timescale. We demonstrate a machine learning model which enables each dialogue cycle to take place on a minute-timescale. We manage a key risk related to machine learning result trustworthiness by computing result confidence from a fast and reliable estimation of posterior variance. Lastly, we highlight a constellation of risks stemming from this new paradigm and suggest policies to mitigate them. △ Less

Submitted 1 November, 2023; originally announced November 2023.

Comments: AI for Social Good workshop at NeurIPS (2019), Vancouver, Canada

Gradient-driven diffusion and pattern formation in crowded mixtures

Authors: Prithviraj Nandigrami, Brandy Grove, Andrew Konya, Robin L. B. Selinger

Abstract: Gradient-driven diffusion in crowded, multicomponent mixtures is a topic of high interest because of its role in biological processes such as transport in cell membranes. In partially phase-separated solutions, gradient-driven diffusion affects microstructure, which in turn affects diffusivity; a key question is how this complex coupling controls both transport and pattern formation. To examine th… ▽ More Gradient-driven diffusion in crowded, multicomponent mixtures is a topic of high interest because of its role in biological processes such as transport in cell membranes. In partially phase-separated solutions, gradient-driven diffusion affects microstructure, which in turn affects diffusivity; a key question is how this complex coupling controls both transport and pattern formation. To examine these mechanisms, we study a two-dimensional multi-component lattice gas model, where "tracer" molecules diffuse between a source and a sink separated by a solution of sticky "crowder" molecules that cluster to form dynamically evolving obstacles. In the high temperature limit, crowders and tracers are miscible and transport may be predicted analytically. At intermediate temperatures, crowders phase separate into clusters that drift toward the tracer sink. As a result, steady-state tracer diffusivity depends non-monotonically on both temperature and crowder density and we observe a variety of complex microstructures. In the low temperature limit, crowders rapidly aggregate to form obstacles that are kinetically arrested; if crowder density is near the percolation threshold, resulting tracer diffusivity shows scaling behavior with the same scaling exponent as the random resistor network model. Though highly idealized, this simple model reveals fundamental mechanisms governing coupled gradient-driven diffusion, phase separation, and microstructural evolution in crowded solutions. △ Less

Submitted 20 October, 2016; v1 submitted 12 July, 2016; originally announced July 2016.

Comments: 16 pages, 5 figures

Journal ref: Phys. Rev. E 95, 022107 (2017)

Modeling Defects, Shape Evolution, and Programmed Auto-origami in Liquid Crystal Elastomers

Authors: Andrew Konya, Vianney Gimenez-Pinto, Robin Selinger

Abstract: Liquid crystal elastomers represent a novel class of programmable shape-transforming materials whose shape change trajectory is encoded in the material's nematic director field. Using three-dimensional nonlinear finite element elastodynamics simulation, we model a variety of different actuation geometries and device designs: thin films containing topological defects, patterns that induce formation… ▽ More Liquid crystal elastomers represent a novel class of programmable shape-transforming materials whose shape change trajectory is encoded in the material's nematic director field. Using three-dimensional nonlinear finite element elastodynamics simulation, we model a variety of different actuation geometries and device designs: thin films containing topological defects, patterns that induce formation of folds and twists, and a bas-relief structure. The inclusion of finite bending energy in the simulation model reveals features of actuation trajectory that may be absent when bending energy is neglected. We examine geometries with a director pattern uniform through the film thickness encoding multiple regions of positive Gaussian curvature. Simulations indicate that heating such a system uniformly produces a disordered state with curved regions emerging randomly in both directions due to the film's up-down symmetry. By contrast, applying a thermal gradient by heating the material first on one side breaks up-down symmetry and results in a deterministic trajectory producing a more ordered final shape. We demonstrate that a folding zone design containing cut-out areas accommodates transverse displacements without war** or buckling; and demonstrate that bas-relief and more complex bent-twisted structures can be assembled by combining simple design motifs. △ Less

Submitted 14 February, 2016; originally announced February 2016.

Comments: 11 pages, 7 figures

Brownian Motion of Arbitrarily Shaped Particles in Two-Dimensions

Authors: Ayan Chakrabarty, Andrew Konya, Feng Wang, Jonathan V. Selinger, Kai Sun, Qi-Huo Wei

Abstract: Here we implement microfabricated boomerang particles with unequal arm lengths as a model for non-symmetry particles and study their Brownian motion in a quasi-two dimensional geometry by using high precision single particle motion tracking. We show that due to the coupling between translation and rotation, the mean squared displacements of a single asymmetric boomerang particle exhibit a non-line… ▽ More Here we implement microfabricated boomerang particles with unequal arm lengths as a model for non-symmetry particles and study their Brownian motion in a quasi-two dimensional geometry by using high precision single particle motion tracking. We show that due to the coupling between translation and rotation, the mean squared displacements of a single asymmetric boomerang particle exhibit a non-linear crossover from short time faster to long time slower diffusion, and the mean displacements for fixed initial orientation are non-zero and saturate out at long time. The measured anisotropic diffusion coefficients versus the tracking point position indicate that there exists one unique point, i.e. the center of hydrodynamic stress (CoH), at which all coupled diffusion coefficients vanish. This implies that in contrast to motion in 3D where the CoH only exists for high symmetry particles, the CoH always exists for Brownian motion in 2D. We develop an analytical model based on Langevin theory to explain the experimental results and show that among the 6 anisotropic diffusion coefficients only 5 are independent because the translation-translation coupling originates from the translation-rotation coupling. Finally we classify the behavior of 2D Brownian motion of arbitrarily shaped particles into four groups based on the particle shape symmetry group. △ Less

Submitted 22 October, 2014; originally announced October 2014.

Journal ref: Langmuir 30, 13844 (2014)

Brownian Motion of Boomerang Colloidal Particles

Authors: Ayan Chakrabarty, Andrew Konya, Feng Wang, Jonathan V. Selinger, Kai Sun, Qi-Huo Wei

Abstract: We investigate the Brownian motion of boomerang colloidal particles confined between two glass plates. Our experimental observations show that the mean displacements are biased towards the center of hydrodynamic stress (CoH), and that the mean-square displacements exhibit a crossover from short time faster to long time slower diffusion with the short-time diffusion coefficients dependent on the po… ▽ More We investigate the Brownian motion of boomerang colloidal particles confined between two glass plates. Our experimental observations show that the mean displacements are biased towards the center of hydrodynamic stress (CoH), and that the mean-square displacements exhibit a crossover from short time faster to long time slower diffusion with the short-time diffusion coefficients dependent on the points used for tracking. A model based on Langevin theory elucidates that these behaviors are ascribed to a superposition of two diffusive modes: the ellipsoidal motion of the CoH and the rotational motion of the tracking point with respect to the CoH. △ Less

Submitted 18 August, 2013; v1 submitted 16 August, 2013; originally announced August 2013.

Journal ref: Phys. Rev. Lett. 111, 160603 (2013)