AI Ethics

Short URL to this document: http://okf.fi/mydata2018-ethics

Slack channel: https://mydataglobal.slack.com/messages/CAR71JHPV

Our Data Rights

Practical AI ethics


AI Ethics

Embedding privacy and open data into AI ethics 


By putting together tiny ones and zeros, we are piling up huge quantities of personal data. Quite likely, we will never manage our personal data ourselves. Indeed, we do need personal control over personal data, but an average reasonable person will soon hand this over to her artificially intelligent helper - a trusted, ethically safe helper. The experience of corporate entities “taking care” of what we see on the web, what product we buy, who are our friends and what thoughts we think, is luckily just too recent to let it slip between the lines that we factor in some ethics in our new artificial helpers.

There is barely any ethical layer in the current digital domain. This track aims for raising awareness about the many nice and hard ways we need to embed ethics into the solutions we’re building for the future. Ethics needs to become more than just a nice word in the personal data talks. It needs to become a practical tool that goes beyond the transparent algorithms and decision making, to the fair remuneration of the individuals behind AI training data, and beyond.

Keywords: artificial intelligence, algorithmic transparency, automated decision making, AI training data, ethics


Our Data Rights

Wednesday: Alvar

13:15 - 14:30

TIME

Jeni Tennison (Open Data Institute), Frederike Kaltheuner (Privacy International), Paul-Olivier Dehaye (PersonalData.io), Karin Christiansen (OKI), Martin Tisné (moderator)

MODERATOR AND PRESENTERS

Bring together three communities that are not talking to each other enough to chart a joint way forward: AI ethics, open data, privacy/surveillance. The outcome is a coming together, a silo busting, of these 3 groups around a joint data rights framework for individuals, companies and governments, which will then be translated into (1) campaigns, (2) regulations, (3) policy change. We expect to discuss the following policy ideas: (1) Incl transparency of automated decision making systems under national and local RTI laws. (2) Implement principles of algorithmic transparency for all automated decision making tools that are developed for and by the public sector. (3) Include principles of algorithmic transparency in Open Contracting Data Standard. (4) Ensure that AI training data is available as open data.

Session is moderated by: Martin Tisné

We need to add short bio’s for each of the panelists.

  1. Add room setup requirements (flipcharts, layout, tech etc) here: https://mydata2018.org/session-practicalities/
  2. Update here below the example script...

Start        :                        13:15

Opening:                          (x minutes, moderator)                 

Intro of panel:                
Discussions:

Questions:

Closing:                        14:25 (5 minutes, moderator)

End:                                14:30


Practical AI Ethics

Wednesday: Alvar

15:00 - 16:15

TIME

Fabio Massimo Zanzotto, Erlin Gulbenkoglu, Natalia Rincón, Reggie Rusan (m)

MODERATOR AND PRESENTERS

AI will have the biggest impact on our lives for the foreseeable future.  We must harness its potential and power Ethically with personal data privacy at the heart. The Practical AI Ethics tract will tackle these challenges and opportunities!


<b>Every Artificial Intelligence has Human Knowledge in the Loop</b>

Fabio Massimo Zanzotto, University of Rome Tor Vergata


Recognizing that every Artificial Intelligence (AI) system has humans in the loop, we propose Human-in-the-loop Artificial Intelligence (HitAI) as a fairer paradigm for building AI systems. In fact,
HitAI will reward the legitimate owners of the knowledge used in these systems. Any decisions of AI systems generating revenues will repay the legitimate owners of the knowledge used for taking those decisions. This is a call for AI reserchers! As modern Merry Men, HitAIresearchers should fight for a fairer Robin Hood Artificial Intelligence that gives back what it steals.

<b>How to build AI while honouring data subject rights?</b>

Erlin Gulbenkoglu, Silo.AI

Privacy is reborn in the digital age with the demand of data subjects on their rights. In AI research and development more attention is directed to building interpretable, anonymous, secure and fair AI systems. We will discuss the challenges and best practices of model interpretability, building anonymous models, data minimisation and anonymisation based on Erlin’s experience at Silo.AI on privacy by design principle and fair usage of AI.

<b>Ethical AI for Smart Cities </b>

Natalia Rincón, CEO at CHAOS architects

Session is moderated by: Reggie Rusan

  1.  Add room setup requirements (flipcharts, layout, tech etc) here: https://mydata2018.org/session-practicalities/
  2. Update here below the example script…

Start:                                15:00 (allow 2 min for settling in)

Opening:                        .. (5 minutes, moderator)                 

Intro of presenter 1:                .. (1 minute, moderator)

Presenter 1:                        .. (15/20 minutes)


Q&A for presenter 1:                .. (5 minutes)

Intro of presenter 2:                .. (1 minute)

Presenter 2:                        .. (15/20 minutes)

Q&A for presenter 2:        .. (5 minutes)

Closing:                        16:10 (5 minutes, moderator)

End:                                16:15