Slide 1
Slide 1 text
About the good and the bad things of [AI] Algorithms and the European AI Act
Marc Salomon, Professor of Decision Science, University of Amsterdam
Slide 2
Slide 2 text
1
WHITE BOX MODELS
(= explain what happens reasonably well)
BLACK BOX MODELS
(= difficult to explain)
Decision tree
Linear regression Neural network
This talk is about “the good” and “the bad” of algorithms
Slide 3
Slide 3 text
2
What is the fear ?
That we use (AI) algorithms that make decisions
- intentionally or not - with far-reaching social,
health, security and / or financial consequences
Slide 4
Slide 4 text
3
Slide 5
Slide 5 text
4
OMG, this does not happen here, ….
Dutch childcare tax benefit affair.
Algorithms determine who gets on the blacklist
of the tax authorities and may run into big
financial problems
Slide 6
Slide 6 text
5
Harm to our democratic system – Cambridge Analytica
• Floating voters were
identified using Facebook
data
• They were presented
information (via Facebook
and other channels) in
favour of voting Trump
Slide 7
Slide 7 text
Harm to our wallets – digital cartels
6
Slide 8
Slide 8 text
7
But, …., there is also another fear to take very seriously
Harm due to not using AI,
for instance to save lives,
because they can not be
developed due to some laws
Slide 9
Slide 9 text
8
A difficult balance
Forbidding “the good”
(= too strict rules)
Continuing with “the bad”
(= too loose rules)
Slide 10
Slide 10 text
9
Extra complication in the discussion on algorithms:
different people/cultures/ethical schools have different
perspectives on “good” and “bad”
(“good” and “bad” are subjective)
Slide 11
Slide 11 text
MIT Moral Machine to test people's perspectives
Slide 12
Slide 12 text
MIT Moral Machine to test people's perspectives
Slide 13
Slide 13 text
12
Different “schools” in ethics think differently
Slide 14
Slide 14 text
13
Different “schools” in ethics think differently
Utilitarian approach: The greatest benefit to the greatest group.
The Fairness of Justice Approach: All people should be treated the same
Many other ethical schools….
Source: https://www.scu.edu/ethics/ethics-resources/ethical-decision-making/thinking-ethically/
Slide 15
Slide 15 text
14
How to deal with this ?
Different ethical schools Different applicable laws
Technical limitations
(math, CS, audit)
Slide 16
Slide 16 text
15
Mathematical limitations
Explainability
(= how good could we explain
the outcome of the algorithm to people) Quality of outcome of the algorithm
Slide 17
Slide 17 text
Europe’s answer
European AI Act
16
Slide 18
Slide 18 text
Source: EU Guidelines for Trustworthy AI, Independent High Level Expert Group 17
Slide 19
Slide 19 text
Source: EU Guidelines for Trustworthy AI, Independent High Level Expert Group 18
Slide 20
Slide 20 text
Source: EU Guidelines for Trustworthy AI, Independent High Level Expert Group 19
Slide 21
Slide 21 text
Act classifies risk of applications and domains
Source: Eve Gaumond, Lawfare Blog, June 2021
20
Slide 22
Slide 22 text
21
Hurdles with the law
Slide 23
Slide 23 text
22
Hurdles with the European Act
1. Many concepts are very subjective. Not all rational thinkers would describe
“harmful discrimination” in the same way
Slide 24
Slide 24 text
23
Hurdles with the European Act
1. Many concepts are very subjective. Not all rational thinkers would describe
“harmful discrimination” in the same way
2. Translation of the legal text into mathematics is difficult. Different mathematical
concepts to describe harmful discrimination.
Slide 25
Slide 25 text
24
Hurdles with the European Act
1. Many concepts are very subjective. Not all rational thinkers would describe
“harmful discrimination” in the same way
2. Translation of the legal text into mathematics is difficult. Different mathematical
concepts to describe harmful discrimination.
3. Monitoring / auditing concepts, including XAI, still not in mature stage.
Slide 26
Slide 26 text
25
Hurdles with the European Act
1. Many concepts are very subjective. Not all rational thinkers would describe
“harmful discrimination” in the same way
2. Translation of the legal text into mathematics is difficult. Different mathematical
concepts to describe harmful discrimination.
3. Monitoring / auditing concepts, including XAI, still not in mature stage.
4. Algorithms have very positive contributions: how to keep them ?
Slide 27
Slide 27 text
26
Slide 28
Slide 28 text
27
Slide 29
Slide 29 text
28
Slide 30
Slide 30 text
29
Slide 31
Slide 31 text
30
Time to share some “best practices”
Slide 32
Slide 32 text
31
Best practices: do a communication campaign
Slide 33
Slide 33 text
32
Best practices: provide reliable information
Slide 34
Slide 34 text
33
Slide 35
Slide 35 text
34
Slide 36
Slide 36 text
35
Best practices: be careful at buying algorithms
Slide 37
Slide 37 text
36
What are your thoughts on trustworthy AI ?
Marc Salomon,
Professor of Decision Science
Program Director MBA Big Data & Business Analytics
University of Amsterdam
[email protected]
www.linkedin.com/in/marcsalomon