This document discusses the ethics of increasing automation and the evolving knowledge infrastructure. It notes challenges around trusting automated analysis with no human in the loop, and understanding complex and evolving data sources. Breakout groups are proposed to discuss interventions for improving research quality, issues around reproducibility with different data types, and the impact of large online data on reproducibility in social sciences. The document references challenges around safety vs security, and tradeoffs between hardening systems and adaptive response.
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Ethics of Automation
1. David De Roure
@dder
The Ethics of Automation
A dystopian view of our
evolving knowledge infrastructure
DIRECTOR, UNIVERSITY OF OXFORD E-RESEARCH CENTRE
4. Energy Efficient
Computing
Infrastructure
(STFC)
De-identified admin
(inc. health) data
Business
data
Open data
(public sector)
Social media
data
Research
data
Longitudinal
survey data
Open data
Securely held data
Environment
data
Business and LG Data
Research Centres
(ESRC)
Admin Data Research
Centres (ESRC)
High Performance
Data Environment
(NERC)
Clinical
data
Medical Bioinformatics (MRC)
Understanding Populations
(ESRC)
Clinical Practice Datalink
(MHRA, NIHR)
100,000 Genome Project NHS)
Research Data Facility (EPSRC)
European Bioinformatics
Institute (EMBL)
Bioscience E-Infrastructure
(BBSRC)
Square Kilometre Array (STFC)
Digital Transformations
(AHRC)
Archive
data
Open Data
Institute
Commercial
Research
Understanding Populations
(ESRC)
7. Social Media Triangle
social media
data and
analytics
social media for
engagement with
research
social media
as a subject
of research
Sam McGregor
8. New Forms of Data
▶ Internet data, derived from social
media and other online interactions
(including data gathered by
connected people and devices, eg
mobile devices, wearable
technology, Internet of Things)
▶ Tracking data, monitoring the
movement of people and objects
(including GPS/geolocation data,
traffic and other transport sensor
data, CCTV images etc)
▶ Satellite and aerial imagery (eg
Google Earth, Landsat, infrared,
radar mapping etc) http://www.oecd.org/sti/sci-tech/new-data-for-
understanding-the-human-condition.htm
10. Blackett Review of IoT
▶ The Internet of Things
describes a world in which
everyday objects are
connected to a network so that
data can be shared
▶ But it is really as much about
people as the inanimate object
▶ It is impossible to anticipate
all the social changes that
could be created by connecting
billions of devices
https://www.gov.uk/government/publications/internet-of-things-blackett-review
11. PETRAS Privacy, Ethics, Trust, Reliability, Acceptability, and Security
for the Internet of Things
• The fusion of the cyber, physical and human elements
• Scale: from 1mm3 devices to large infrastructure systems
• Managing devices throughout their (decades long) lifetimes
• New and evolving threat landscape
• Continue to operate when partially compromised
The Challenges are numerous
• Safety vs Security
• Security vs Efficiency
• Hardening vs Adaptive Response
Tradeoffs
Emil Lupu
19. Edwards, P. N., et al. (2013) Knowledge Infrastructures: Intellectual Frameworks and Research
Challenges. Ann Arbor: Deep Blue. http://hdl.handle.net/2027.42/97552
21. Research in the Wild (West)
Imagine you are a conference chair… or responsible for urban
planning, or security. Confidence in results is getting harder:
What interventions should we make to improve confidence
and quality? What (socio-)technology can we adopt?
Trusting the analysis that
is occurring
Automation of workflows,
crowd-sourced data reduction,
software vulnerabilities, increasing
adoption of machine learning,
and no critical human in the loop
Knowing what the data is,
where it has come from,
and what we can do with it
Multiple and partial data sources,
at speed and scale, in an evolving
ecosystem of data processing
intermediaries, with complex
permissions for data use
22. Breakout topics
Group A – Lecture Theatre
Chair: Prof David de Roure
What interventions can we make to
improve the quality of research in
our increasingly automated
research communication
ecosystem?
Group C – Ho Tim Seminar Room
Chair: Prof Eric Meyer
Repurposed data (e.g. Twitter,
Wikipedia), data exhaust (e.g.
mobile phones, Loyalty cards),
proprietary data, data that comes
with non-sharing or non-disclosure
agreements (e.g. Twitter), other sorts
of commercial data, data limited in
distribution because of differing
national legal frameworks, etc.
What does reproducibility mean in
these contexts?
Group B – Louey Seminar Room
Chair: Dr Suzy Moat
Reproducibility in the social
sciences: does the arrival of large
online data sources make the
outlook better or worse?”