Copy
The EuropeanAI Newsletter
Covering Artificial Intelligence in Europe

Welcome to the EuropeanAI Newsletter covering the European AI and technology ecosystem. If you want to catch up on the archives, they're available here.

You can share the subscription link here and read the last newsletter here

 

In case you missed it, the deadline to provide input to the European Commission’s White Paper has been extended to the 14th of June. Click here for a summary of the White Paper on AI published some weeks ago. A recent discussion by two members of the High Level Expert Group on AI on the role of ethical AI in European legislative proposals can be found here.

The European Commission is holding an ongoing Technology Transfer Experiment call for SARS-CoV-2 related proposals with a maximum of €200,000 per application (deadline: 15/06/2020). The European AI Alliance is currently
hosting two databases for AI and robotics projects related to SARS-CoV-2 which can be updated by the community. There is also a vivid discussion on the AI Alliance platform, ranging from a data science analysis of the pandemic to the potential for x-ray detection. The OECD also set up a repository of international initiatives on AI.

There have been a bunch of proposals honoured by the Federal Ministry for Economic Affairs and Energy following the #WirVsVirus (we vs the virus) hackathon in Germany such as
ML for diagnosis through anonymised images, a chatbot using data analysis to help answer questions surrounding social distancing and an app to help students study remotely during the crisis. On the topic of EdTech, the European Commission is currently holding an open call offering around €6M to support innovative education projects in the EU. 


Right in the middle of the cool or creepy spectrum, you can now “remote control” a local equipped with a live video camera in the Faroe Islands and learn about the ecosystem there while island gazing from your couch. Cool if you care about learning more about the Faroe Islands without travelling, creepy if your main interest is to boss another human around on an app under the guise of remote tourism. Speaking of the couch, if you’re in the UK, sci-fi mini-series Devs is available on BBC iPlayer. For an effect that makes you feel right between following someone via live stream on a remote island and hunching over on your couch is this project, creating a 3D effect in photos. Hogwarts students want their money back.

In more academic news, Philosopher
Vincent C. Müller recently published his forthcoming chapter for the Stanford Encyclopedia of Philosophy on the Ethics of artificial intelligence and robotics. The chapter introduces the field, followed by sections on ethical tensions, machine ethics and artificial moral agency and the problem a possible future artificial superintelligence might pose.

Share Share
Tweet Tweet
Forward Forward
Read Later Read Later
Share Share
Policy, Strategy and Regulation

COVID-19 contact tracing and data privacy: a short story in three parts

PEPP-PT and DP-3T 

In light of the pandemic, PEPP-PT (Pan-European Privacy-Preserving Proximity Tracing) was established to build contact tracing platforms which hope to inform people if they have been potentially exposed to the virus through the publication of standards, building technology and providing services to developers and countries. 

At present, many businesses are channelling resources to fight SARS-CoV-2, with a surprising amount of them based on geolocation and geospatial data to augment contact tracing capabilities (for instance, using the Bluetooth APIs shared by Apple and Google in recent weeks). Though the intentions may be positive, care should be taken to ensure decisions are not being made which will negatively impact data privacy and other rights or ethical values -- now or in the future. We recommend following the work of Michael Veale who has spoken at length over the past weeks about the difficulties PEPP-PT faces and the significant lack of clarity around the project specifically.

A multi-stakeholder group of academics has published an alternative named Decentralised Privacy-Preserving Proximity Tracing which focuses on the provision of various privacy and security provisions such as graceful dismantling. The DP-3T (Decentralised Privacy-Preserving Proximity Tracing) team have responded to recent updates to PEPP-PTs site and documentation. Specifically, PEPP-PT has removed reference to DP-3T’s decentralised approach, signifying a shift to a centralised protocol which lacks transparency, erodes trust and is susceptible to function creep. A detailed response to the PEPP-PT updates can be found here. At the same time, many institutions have abandoned PEPP-PT due to the questionable nature of the project.

Notes in the margin: The Ada Lovelace Foundation published a report on digital contact tracing, named ‘Exit through the App Store?’. You may also be interested in signing this Joint Statement on contact tracing.

Ethical implications of the use of AI and other technology in the pandemic: IEEE edition

The Institute of Electrical and Electronics Engineers (IEEE) has released a Statement Regarding the Ethical Implementation of Artificial Intelligence Systems (AIS) for Addressing the COVID-19 Pandemic. It echoes those of multiple other institutions and organisations in that there should be a careful evaluation between the good that AI can do and the tensions it can create for fundamental rights, values and ethics in light of thoughtless development and deployment in a crisis situation. The statement briefly mentions the work of the IEEE on Ethically Aligned Design prior to delving into a number of responses to the global situation. They make 10 recommendations, touching on new measures for success, risk assessments, regulation, availability (of the internet etc.), mental health concerns and more.

Notes in the margin: Although the IEEE mentions that technology platforms can increase human wellbeing by e.g. enabling humans to stay in contact, it should be noted that popular applications such as zoom have faced a variety of security issues (zoombombing, security exploits).

What about the European Commission?

In light of all this, the European Commission recently published guidance on app development and data protection in light of fighting SARS-CoV-2. The guidance follows the ‘Recommendation to support exit strategies through mobile data and apps’ adopted on the 8th of April.  The Guidance takes contributions from the European Data Protection Board (EDPB) into account and outlines several aspects to ensure the trustworthiness of an app in question. This includes: strict limits on data storage, clarity of the purpose of processing, involvement of the Data Protection authorities,   and opening the source code of the app open for public review and encrypted transmissions to health authorities. The package is accompanied by an EU toolbox on contact tracing apps.

Notes in the margin: You may also be interested in reading the European Parliamentary Research Service's analysis on Ten Technologies to fight coronavirus or the adopted guidance from the European Data Protection Board.
Ecosystem

Trustworthy AI by any other name 
While AI principles seemingly continue to mushroom, it remains difficult to verify whether or not (and to what degree) the recommendations outlined therein have been acted upon and implemented by organizations. A multi-disciplinary group of 58 researchers from institutions across the globe recently attempted to address this problem. They published a report entitled “Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims”. The report proposes 10 mechanisms to improve the verifiability of claims made about a given AI-based system. The mechanisms provided can serve the purpose of enabling verifiability of claims from the perspective of developers and industry, but also to ensure that civil society, government and other stakeholders can evaluate claims.

The report is divided into three sections covering institutional mechanisms, software mechanisms and hardware mechanisms. Each section outlines relevant recommendations ranging from privacy-preserving ML, compute support for academia to bias and safety bounties.
Notes in the margin: Pretty thrilled to have been able to contribute to this effort!
 
Crystal clear?

To our eyes, a household mirror may appear a near-perfect surface, but under a microscope, it reveals its chaotic nature. Glass is a somewhat enigmatic material, like a liquid which has been frozen in time with some inherent level of noise baked into it.

Researchers at LIGO (Laser Interferometer Gravitational-Wave Observatory) require extreme levels of accuracy, with LIGO’s current instrument relying on a mirror with a 99.9999% reflective surface, built using layered transparent ‘silica’ glass-like that we use in our homes, and reflective ‘tantala’ glass. The noise, thought to be caused by atoms switching between two different configurations, can result in the surface of the glass shifting in shape by the width of an atom, which may not sound like much, but becomes a problem for LIGO who measure distances one-thousandth that size. As LIGO looks towards their next-generation system (Advanced LIGO Plus), their team is investigating novel solutions to reduce that noise by more than half.

What does AI have to say about glass? 

On a related note, researchers at DeepMind have been looking into glass transition, a process affecting a wide range of ‘locally constrained’ systems, like ironing clothes, freezing ice cream, or even traffic jams. The team built a graph neural network to simulate ‘glassy dynamics’, predicting the movements of particles in glasses with high levels of accuracy. Their model achieves 96% accuracy against the ground truth of a simulation over very short timescales, and 64% over the relaxation period of glass, a 40% improvement over the previous state of the art system. The movements of glassy particles during their relaxation period are like that of normal glass over thousands of years.

Notes in the margin: Though DeepMind’s research doesn’t directly address LIGO’s challenge, perhaps simulating when and how the surface of their mirror will change could go some way to improving the accuracy of their measurements. 
Enjoy learning about Europe? Share the subscription link with friends, colleagues, enemies..

Contact Charlotte at:

www.charlottestix.com
@charlotte_stix


Ben Gilburt (Sopra Steria) co-wrote this edition. 

Interesting events, numbers or policy developments that should be included? Send an email!

Disclaimer: this newsletter is personal opinion only and does not represent the opinion of any organisation.

Copyright © Charlotte Stix, All rights reserved.
Twitter Twitter
Website Website
Email Email






This email was sent to <<Email Address>>
why did I get this?    unsubscribe from this list    update subscription preferences
EuropeanAI · 16 Mill Lane · Cambridge, UK CB2 1SB · United Kingdom

Email Marketing Powered by Mailchimp