The Anatomy of Explanations for Artificial Intelligence: How Explanations and Explainability Can Be Defined in the Context of Black-Box Algorithms and the GDPR

(1)

THE ANATOMY OF EXPLANATIONS FOR

ARTIFICIAL INTELLIGENCE

Saar Hoek, 6935362 MSc. Artificial Intelligence Utrecht University

HOW EXPLANATIONS AND EXPLAINABILITY CAN BE DEFINED IN THE CONTEXT OF

BLACK-BOX ALGORITHMS AND THE GDPR

OCTOBER 2022

SUPERVISED BY: PROF. DR. HENRY PRAKKEN

SECOND EXAMINER: DR. ANNEMARIE BORG

(2)

(3)

(4)

(5)

(6)

(7)

(8)

1 Art. 13-15 GDPR.

(9)

2 Art. 2(1) GDPR

3 Art. 4(1) GDPR

4 Art. 4(2) GDPR

(10)

(11)

(12)

5 These are some of the scholars that either develop post-hoc explainability methods or lay out their benefits.

This does not entail that they are opposed to inherently interpretable machines, merely that they express the need for additional post-hoc methods.

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

(32)

(33)

(34)

(35)

(36)

(37)

(38)

(39)

(40)

(41)

(42)

(43)

(44)

(45)

(46)

(47)

(48)

(49)

6 (UK Government (gov.uk) n.d.) https://www.gov.uk/dismissal, accessed 11 August 2022.

7 Rijksoverheid, https://www.rijksoverheid.nl/onderwerpen/ontslag/vraag-en-antwoord/wanneer-mag-ik- worden-ontslagen, accessed 11 August 2022.

(50)

(51)

(52)

8 art. R. 311-3-1-2 of the French Code on the relationships between the public and the administration.

(53)

(54)

(55)

(56)

(57)

(58)

(59)

(60)

(61)

(62)

(63)

(64)

(65)

(66)

(67)

(68)

(69)

9 Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995 on the protection of individuals with regard to the processing of personal data and on the free movement of such data.

(70)

10 GDPR, Recital 6.

(71)

11 Case 215/88 Casa Fleischhandels [1989] European Court of Justice ECR 2789 [31], https://eur- lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A61988CJ0215 [https://perma.cc/C8RK-45FP].

(72)

(73)

(74)

(75)

(76)

(77)

(78)

(79)

(80)

(81)

(82)

(83)

(84)

(85)

(86)

(87)

(88)

(89)

(90)

(91)

(92)

(93)

(94)

(95)

(96)

(97)

(98)

(99)

(100)

(101)

(102)

(103)

(104)

12 See Naruto v. Slater, et al., no. 16-15469 (9^th Cir. 23 April 2018), United States.

(105)

(106)

(107)

(108)

(109)

(110)

(111)

(112)

(113)

(114)

(115)

 

(116)

(117)

(118)

(119)

(120)

(121)

(122)

(123)

The Anatomy of Explanations for Artificial Intelligence: How Explanations and Explainability Can Be Defined in the Context of Black-Box Algorithms and the GDPR

Full text