Filters








139 Hits in 6.2 sec

Chess as a Testing Grounds for the Oracle Approach to AI Safety [article]

James D. Miller, Roman Yampolskiy, Olle Haggstrom, Stuart Armstrong
2020 arXiv   pre-print
As the oracles would be vastly more intelligent than the player in the domain of chess, experience with these oracles might help us prepare for future artificial general intelligence oracles.  ...  This paper proposes a possibly practical means of using machine learning to create two classes of narrow AI oracles that would provide chess advice: those aligned with the player's interest, and those  ...  We propose using chess as a testing ground and simplified model for a future circumstance under which we have a powerful AGI oracle.  ... 
arXiv:2010.02911v1 fatcat:kya3crxpmvfzxgxalaakszsmoe

Pattern knowledge and search:The SUPREM architecture

Hans Berliner, Carl Ebeling
1989 Artificial Intelligence  
In the past, this has led to a schism between animate and machine searchers: the former opting for the few nodes, deep understanding approach, and the latter for many nodes but shallow understanding.  ...  merely to be able to apply what would pass as a first-cut approximation by a true human expert in the field.  ...  Also, it is possible for material on the board to increase due to pawn promotion, and thus king safety could again become a consideration, after having been dismissed as inconsequential in the Oracle analysis  ... 
doi:10.1016/0004-3702(89)90056-8 fatcat:xeijdn26g5hh5ksgnwgkbavsh4

Automation and the Future of Command and Control: The End of Auftragstaktik?

Rosario M. Simonetti, Paolo Tripodi
2020 MCU Journal  
Political and/or strategic leaders might find it appealing to pursue preemptive or preventive wars as a strategy to acquire asymmetric advantage over the enemy.  ...  The impact of new technologies and the increased speed in the future battlespace may overcentralize command and control functions at the political or strategic level and, as a result, bypass the advisory  ...  Department of Defense has a more conservative approach to AI.  ... 
doi:10.21140/mcuj.2020110106 fatcat:zovers5k6rfnzjcnlazs4ljoru

Responses to catastrophic AGI risk: a survey

Kaj Sotala, Roman V Yampolskiy
2014 Physica Scripta  
It has been suggested that AGI may pose a catastrophic risk to humanity. After summarizing the arguments for why AGI may pose such a risk, we survey the field's proposed responses to AGI risk.  ...  We consider societal proposals, proposals for external constraints on AGI behaviors, and proposals for creating AGIs that are safe due to their internal design.  ...  Oracle AIs are unlikely to remain as pure oracles for long.  ... 
doi:10.1088/0031-8949/90/1/018001 fatcat:2ekclllvfjazrc6skoyobx5jye

Machine Learning Testing: Survey, Landscapes and Horizons [article]

Jie M. Zhang University College London, Nanyang Technological University)
2019 arXiv   pre-print
This paper provides a comprehensive survey of Machine Learning Testing (ML testing) research.  ...  It covers 144 papers on testing properties (e.g., correctness, robustness, and fairness), testing components (e.g., the data, learning program, and framework), testing workflow (e.g., test generation and  ...  ACKNOWLEDGEMENT Before submitting, we sent the paper to those whom we cited, to check our comments for accuracy and omission.  ... 
arXiv:1906.10742v2 fatcat:p5c54cy4pjc5flzm7shybk3qxe

The Singularity Controversy, Part I: Lessons Learned and Open Questions: Conclusions from the Battle on the Legitimacy of the Debate [article]

Amnon H. Eden
2016 arXiv   pre-print
This report seeks to inform policy makers on the nature and the merit of the arguments for and against the concerns associated with a potential technological singularity.  ...  Part I describes the lessons learned from our investigation of the subject, separating the argu-ments of merit from the fallacies and misconceptions that confuse the debate and undermine its rational resolution  ...  Concerns for AI safety are not limited to the emergence of a race of artificially superintelligent agents for AIs.  ... 
arXiv:1601.05977v1 fatcat:fl77khpywbfqhcpntdbw52sbbm

Introduction to computability logic

Giorgi Japaridze
2003 Annals of Pure and Applied Logic  
The formalism used as a specification language for computational problems, called the universal language, is a non-disjoint union of the formalisms of classical, intuitionistic and linear logics, with  ...  The same appears to be the case for intuitionistic and linear logics (understood in a broad sense and not necessarily identified with the particular known axiomatic systems).  ...  The goal of this paper is to introduce a basic logic of computability and provide a solid formal ground for future, more advanced, studies of this exciting subject.  ... 
doi:10.1016/s0168-0072(03)00023-x fatcat:izj4ic4fhvg3vna6gwbfo4fktq

The Incentives that Shape Behaviour [article]

Ryan Carey, Eric Langlois, Tom Everitt, Shane Legg
2021 arXiv   pre-print
To this end, we introduce structural causal influence models, a hybrid of the influence diagram and structural causal model frameworks.  ...  Finally, we illustrate how these incentives predict agent incentives in both fairness and AI safety applications.  ...  For example: • A chess player (or chess-playing AI) often has an incentive to capture its opponent's queen. • A pilot (or autopilot) has an incentive to move to a safe altitude. • A walking human (or humanoid  ... 
arXiv:2001.07118v2 fatcat:7axuljgh3bhupmvcxz6mwjno4a

OSTP-AWhite House Office of Science and Technology Policy Request for Information on the Future of Artificial Intelligence

Kris Kitchen
2017 Figshare  
White House Office of Science and Technology Policy Request for Information on the Future of Artificial Intelligence  ...  believes that to use AI's Additional research and expansion of these techniques for safe development, testing, and operation should be conducted as well as linking these methods to the emerging legal  ...  Like other economic transitions, AI has the potential for a dramatic increase in prosperity.  ... 
doi:10.6084/m9.figshare.4640305 fatcat:26aelpe4ejbupao5yh2mys2pzq

Supervising strong learners by amplifying weak experts [article]

Paul Christiano, Buck Shlegeris, Dario Amodei
2018 arXiv   pre-print
One solution is to have humans provide a training signal by demonstrating or judging performance, but this approach fails if the task is too complicated for a human to directly evaluate.  ...  We propose Iterated Amplification, an alternative training strategy which progressively builds up a training signal for difficult problems by combining solutions to easier subproblems.  ...  the decomposition oracle H for each task.  ... 
arXiv:1810.08575v1 fatcat:yf3ittk6fraotftmxmkqop2yl4

Artificial Intelligence, Values and Alignment [article]

Iason Gabriel
2020 arXiv   pre-print
A principle-based approach to AI alignment, which combines these elements in a systematic way, has considerable advantages in this context.  ...  Third, the central challenge for theorists is not to identify 'true' moral principles for AI; rather, it is to identify fair principles for alignment, that receive reflective endorsement despite widespread  ...  Acknowledgements This paper has benefited greatly from the input and advice of a large number of people. Within DeepMind, particular thanks are owed to Jan Leike  ... 
arXiv:2001.09768v1 fatcat:g7fhejzw7vef3lwe5ih3lclrny

Digital Twin: Values, Challenges and Enablers [article]

Adil Rasheed, Omer San, Trond Kvamsdal
2019 arXiv   pre-print
Our aim is to provide a detailed coverage of the current challenges and enabling technologies along with recommendations and reflections for various stakeholders.  ...  A digital twin can be defined as an adaptive model of a complex physical system.  ...  ACKNOWLEDGMENT The authors would like to thank Professor Harald Martens for his helpful comments and suggestions on an earlier draft of this manuscript.  ... 
arXiv:1910.01719v1 fatcat:3r2bbivffbeizelake7oo3xgbm

Video fire detection – Review

A. Enis Çetin, Kosmas Dimitropoulos, Benedict Gouverneur, Nikos Grammalidis, Osman Günay, Y. Hakan Habiboǧlu, B. Uǧur Töreyin, Steven Verstockt
2013 Digital signal processing (Print)  
It is possible to cover an area of 100 km 2 using a single pan-tiltzoom camera placed on a hilltop for wildfire detection.  ...  It is also possible to use security cameras and special purpose infrared surveillance cameras for fire detection.  ...  If a false alarm is issued during the training phase, the oracle gives feedback to the algorithm by declaring a no-smoke decision value ( y = −1) for the false alarm region.  ... 
doi:10.1016/j.dsp.2013.07.003 fatcat:fry5gkdqqbd77egujfx7uznii4

Digital Twin: Values, Challenges and Enablers from a Modeling Perspective

Adil Rasheed, Omer San, Trond Kvamsdal
2020 IEEE Access  
Our aim is to provide a detailed coverage of the current challenges and enabling technologies along with recommendations and reflections for various stakeholders.  ...  Also referred to as a computational megamodel, device shadow, mirrored system, avatar or a synchronized virtual prototype, there can be no doubt that a digital twin plays a transformative role not only  ...  The algorithms [139] without any prior knowledge and starting to play random gains, in 24 hours, achieved super-human level performance in chess, shogi as well as Go and convincingly defeated a work-champion  ... 
doi:10.1109/access.2020.2970143 fatcat:nmdakwa2urdmbbcjblsjrkypum

Ethical Artificial Intelligence [article]

Bill Hibbard
2015 arXiv   pre-print
This article also discusses how future AI will differ from current AI, the politics of AI, and the ultimate use of AI to help understand the nature of the universe and our place in it.  ...  This article makes the case for utility-maximizing agents and for avoiding infinite sets in agent definitions.  ...  This idea is important both for the theory of AI and for development of practical AI systems. Hutter grounds his approach in the philosophies of Occam and Epicurus.  ... 
arXiv:1411.1373v9 fatcat:3rl7nqozr5gapncjsufatv7b2m
« Previous Showing results 1 — 15 out of 139 results