Summary Webinar 1 Year Knowledge Centre Data and Society

Omina Technologies > News > Conference > Webinar > Summary Webinar 1 Year Knowledge Centre Data and Society

Summary Webinar 1 Year Knowledge Centre Data and Society

  • Posted by: MartinCanter
1 jaar Kenniscentrum Data & Maatschappij

Key Take Aways 1 Year Knowledge Centre Data and Society

Knowledge Centre Data and Society launches Wikaipedia: links ALTAI (Assessment List for Trustworthy AI) tool with legislation and explains each of the 7 ethical principles

Imec: Update of Digimeter expected in April 2021

UGent (Tom Evens): 2021 plan includes work on acceptance of AI by business

CITIP (Jan De Bruyne and Thomas Gills) proposes digital ethicist as a new job role and introduces the Wikaipedia

Thomas More (Marijke Brants): need to reflect on whether you want to make future decisions in the same way as in the past

Minister Hilde Crevits

Hilde Crevits, Flemish minister of Economics, Innovation, Labour, Social Economics and Agriculture stresses the importance of including the citizen and values in the development of AI solutions. The Corona app is a great example of this.

Pieter Ballon (director Imec-SMIT)

Summary Webinar 1 Year Knowledge Centre Data and Society 1

Pieter Ballon states that we have talked enough about AI being the next big thing and that it is now time to talk about AI in layman terms, like in the book of Geertrui Mieke De Ketelare (director AI Imec): Mens versus Machine.

Three initiatives in 2020
1° Databus
AI Experience Centrum: aggregates 30 AI solutions from VUB
3° Kenniscentrum voor Data en Maatschappij: communication and responsible AI systems research:
Transparency of AI and accountability: why do I get this recommendation,
Practice: involving companies and users in AI. Development of AI blindspots cards to allow companies to detect        AI issues.

Peggy Valcke (Professor of Law KU Leuven)

  1. 3 reports created over the last year:

Artificiele Intelligentie en gegevensbescherming: AGV gids (GDPR) to help companies understand GDPR and its implications.
Ethical guideline for artificial intelligence: The High-level Expert Group on AI (AI HLEG)  published the Ethics Guidelines for Trustworthy AI which was the basis of the ALTAI tool. Now CITIP has made the link between ALTAI tool and the legislation and the most common 7 ethical principles: Wikaipedia
Report with an overview of the different responsibilities of different Belgian government institutions.

2. Networking: via AI4Belgium,
Jacob Turner (Regulating AI)

3. 3 books:  editor Jan De Bruyne
Autonome voertuigen en aansprakelijkheid
Artificiele intelligentie en maatschappij: target the generic public
Artificial Intelligence and the law: targets the legal  audience

It is crucial that society accepts AI.
This year we focussed on trust of the Flemish citizen in AI:
– Digimeter: April update
– AI in healthcare study

Next year:

  • AI in education: due to COVID-19 education is forced towards more and smarter digital transformation.
  • AI for business: acceptance of AI by business
  • Government and AI: part of the Belgian government relance plan

Ethical Guidelines for AI: Thomas Gils and Jan De Bruyne (CITIP KU Leuven)

Translation of ALTAI for company lawyers and AI specialists in Wikaipedia: Gids: Ethische principes en (niet-)bestaande juridische regels voor AI: link between ALTAI, 7 ethical principles and legislation.

For each of the 7 ethical principles Wikaipedia gives an easy to understand explanation, the link with legislation, makes policy recommendations and which ethical tools that could be useful.

Example of policy recomendation: product liability legislation is ok but it needs further clarification. Is software a product? What is a defective product?

Ethical AI

CITIP envisions a new job role: the digital ethicist.

The digital ethicist can develop and maintain an ethical culture within a company, and can represent the social perspective within innovation projects.

Digital Ethicist

AI for recruitment support: Marijke Brants (Thomas More Hogeschool)

It is clear that AI could be used to improve recruitment. AI could be used to automatically generate job descriptions, to personalize assessments and to pre-screen job applicants’ resumees.

AI HRM

However, it is important to reflect on whether you want to make future job recruitment decisions in the same way as in the past. Past job recruitment bias might perpetuate in the future when using AI:
-data bias: e.g. face recognition system trained mostly on Caucasian people underperforms on non-white face images.
– bias in algorithm
– bias in the system:  a team mostly consisting out of white people might more easily overlook that AI solution underperforms for non-white people. Team diversity is important to overcome bias.

algorithmic bias

Bias is a social construct:
– bias concept is dynamic: it is a social construct that changes over time. We need to continuously reflect on  the definition of bias.
– bias is context dependent
– there is no easy fix: excluding the sensitive variable is not the solution: removing gender from cv, but  particular words can still reveal the job applicant’s gender.

AI Bias Concept

The ethical framework must be addressed during the entire AI journey.
– Explainable AI (XAI) to help a job recruiter to explain why a job candidate is selected or rejected, explain to AI recruiter in layman terms how the AI tool makes the decision to improve the AI recruitment logic and remove AI bias from data or algorithm: the feedback loop.

Ethical AI process
Explainable AI