AI AS A FORCE OF GOOD Mariarosaria Taddeo Digital Ethics Lab - Oxford Internet Institute, University of Oxford Alan Turing Institute, London @RosariaTaddeo
AI Definition Outline AI Challenges Ethics for AI: how and what
AI is a growing resource of interactive, autonomous, and self-learning agency, which can be used to perform tasks that Challenges as Risks would otherwise require human intelligence to be executed successfully Floridi & Taddeo, Science, 2018
5 Challenges Challenges as Risks Guang-Zhong Yang et al. Science Robotics, 2018
Wrongdoing profiling becomes discrimination
Trust & little control understanding predictability
AI failures remain human responsibilities Taddeo & Floridi, Nature, 2018a
Pilots need to keep landing airplanes, so they can if AI won t
AI is becoming an invisible facilitator that mediates our interactions in a convenient, barely noticeable way
More AI must mean more humanity
AI Definition Outline AI Challenges Ethics for AI: how and what
It is the hardware the data the software
It is the hardware + the data + the software
It is the hardware + the data + the software The environment
Ethics for AI is developed along three axes X: Ethics of Data Y: Ethics of Algorithms Z: Ethics of Practices Privacy re-identification group privacy Trust in whom? Transparency of what? Responsibility/ac countability Ethical design of requirements Ethical auditing of algorithms Deontological code Consent Privacy of data subjects Secondary use
Ethics for AI is developed along three axes X: Ethics of Data Y: Ethics of Algorithms Z: Ethics of Practices Which principles?
5 documents & 44 principles Montreál Declaration for Responsible AI, Universite de Montre al General Principles of Ethically Aligned Design, IEEE Statement on Artificial Intelligence, Robotics and Autonomous Systems by the EU Commission European Group on Ethics in Science and New Technologies Five overarching principles for an AI code, UK House of Lords Asilomar Principles, FHI Cowls & Floridi, 2018
Avoid discrimination Sustainability Overuse Cowls & Floridi, 2018
Common good Avoid discrimination Sustainability Share benefits Overuse Autonomy Cowls & Floridi, 2018
Common good Avoid discrimination Sustainability Misuse Share benefits Overuse Autonomy Avoid Well-being breaches of privacy Cowls & Floridi, 2018
Common good Caution Avoid discrimination Sustainability Misuse Control Human dignity Overuse Autonomy Share benefits Well-being Avoid breaches of privacy Equity & Solidarity Cowls & Floridi, 2018
Common good Caution Avoid discrimination Sustainability Misuse Control Human dignity A convergence starts to emerge Autonomy Overuse Share benefits Well-being Avoid breaches of privacy Equity & Solidarity Cowls & Floridi, 2018
Common good Caution Avoid discrimination Sustainability Misuse Control Human dignity Overuse Autonomy Share benefits Well-being Avoid breaches of privacy Equity & Solidarity
Translational ethics Ethical principles Viable Challenges as Risks guidelines for AI innovation
We shape our buildings and afterwards our buildings shape us
Every Bit as Good
Thank you Mariarosaria Taddeo @RosariaTaddeo mariarosaria.taddeo@oii.ox.ac.uk http://digitalethicslab.oii.ox.ac.uk/mariarosaria-taddeo/ http://rosariataddeo.net COPYRIGHT DISCLAIMER Texts, marks, logos, names, graphics, images, photographs, illustrations, artwork, audio clips, video clips, and software copyrighted by their respective owners are used on these slides for non-commercial, educational and personal purposes only. Use of any copyrighted material is not authorised without the written consent of the copyright holder. Every effort has been made to respect the copyrights of other parties. If you believe that your copyright has been misused, please direct your correspondence to: mariarosaria.taddeo@oii.ox.ac.uk stating your position and I shall endeavour to correct any misuse as early as possible.