24-25: ARTIFICIËLE
INTELLIGENTIE:
MAATSCHAPPELIJKE
UITDAGINGEN
Inhoudsopgave
LES 1: INTRODUCTIE.......................................................................................3
GESCHIEDENIS VAN AI........................................................................................................... 4
MACHINE LEARNING HOE?...................................................................................................... 4
WAT IS ER VERANDERD?........................................................................................................ 5
LES 2: UNDERSTANDING AND INTERPRETING DEEP NEURAL NETWORKS............5
BACKGROUND...................................................................................................................... 5
DESIGNING A DEEP NEURAL NETWORK......................................................................................6
DEEP NEURAL NETWORK VOOR VISUELE DATA............................................................................7
GENERATIVE MODELS............................................................................................................ 8
autoencoders............................................................................................................... 8
Variational Autoencoders (VAE)....................................................................................9
Generative adversial networks.....................................................................................9
Diffusion models........................................................................................................ 11
CHALLENGES VOOR AI......................................................................................................... 11
LES 3: RESPONSIBLE AI.................................................................................12
FLAT FLOW FOR DATA SCIENCE ETHICS....................................................................................13
AI RISKS........................................................................................................................... 14
Onmiddelijke risico’s:................................................................................................. 14
Systemic risks............................................................................................................ 16
Long term impact....................................................................................................... 17
DE WEG VOORUIT............................................................................................................... 19
LES 4: IMPORTANCE OF SAFETY IN AI DESGIN................................................19
WAAR ZIEN WE AI IN GEBRUIK?............................................................................................. 20
IF AI INTERACTS PHYSICALLY WITH HUMANS, HOW DO WE MAKE SURE THAT IT DOES IT SAFELY?.........20
PROBLEMEN MET HET GEBRUIKEN VAN RL...............................................................................23
DESIGNING SAFETY IN RL..................................................................................................... 24
SAFE DESIGN OF AI ENABLED ROBOTICS..................................................................................24
LES 5: FAIRNESS AND GENERATIVE AI............................................................25
RESPONSIBLE AI................................................................................................................. 25
VERSCHILLENDE SOORTEN BIAS:............................................................................................ 25
HOE MEET JE FAIRNESS?...................................................................................................... 27
GENERATIVE AI RISKS......................................................................................................... 28
LES 6 SUSTAINABLE AI..................................................................................30
AI FOR SUSTAINABILITY (DE AI GEBRUIKEN VOOR BETERE DUURZAAMHEID).....................................30
, DATA COLLECTIE VOOR MEER DUURZAAMHEID..........................................................................31
OVERVIEW OF OUR AI FOR SUSTAINABILITY PROJECTS................................................................32
Bio-accelerated mineral weathering (BAM!)...............................................................32
Future arctic............................................................................................................... 33
Global fertilizer dataset..............................................................................................34
Curieuze Neuze: enviromental monitoring.................................................................34
ICOS Brasschaat: Forest monitoring...........................................................................34
ROL VAN AI IN SLIMME INFRASTRUCTUUR................................................................................35
DE NEGATIEVE IMPACT VAN AI OP DUURZAAMHEID....................................................................36
LES 7 DESIGNING FUTURES: HUMAN-CENTERED DESIGN IN TECHNOLOGIES.....37
WORLDBUILDING................................................................................................................ 39
DESIGN FICTION................................................................................................................. 39
CRITICAL DESIGN................................................................................................................ 39
INVENTION FACTORIES......................................................................................................... 39
HET GROTE GEVAAR VAN AUTOMATION...................................................................................40
AUGMENTATION................................................................................................................. 40
LES 8: THE MORAL DANGERS AND OPPORTUNITIES OF GENERATIVE,
MULTIMODAL LLMS AND OTHER AI................................................................41
3 GEBIEDEN VAN FILOSOFIE.................................................................................................. 41
ACCELERATED INNOVATION IN AI........................................................................................... 42
ETHIEK............................................................................................................................. 44
EPISTEMOLOGIE.................................................................................................................. 45
METAFYSICA...................................................................................................................... 46
THE ROLE OF AI................................................................................................................. 48
Spectrum of mind....................................................................................................... 48
Ethische implicaties.................................................................................................... 49
Epistemologische implicaties.....................................................................................49
Metafysische implicaties............................................................................................50
LES 9 AI EN RECHT........................................................................................50
ROL+RECHTSHANDHAVING................................................................................................... 50
UITDAGINGEN VAN AI VOOR ANTIDISCRIMINATIERECHT...............................................................50
UITDAGINGEN VAN AI VOOR AUTEURSRECHT............................................................................51
UITDAGINGEN VAN AI VOOR CONSUMENTENRECHT.....................................................................53
LES 10 DE REGULERING VAN AI IN EUROPA....................................................54
EU STRATEGIE VOOR AI:...................................................................................................... 54
AI-VERORDENING............................................................................................................... 54
Doelstellingen............................................................................................................ 54
Definities.................................................................................................................... 54
risk-based approach................................................................................................... 55
AI EN AANSPRAKELIJKHEID.................................................................................................... 64
AI EN GEGEVENSCBESCHERMING............................................................................................ 65
AVG............................................................................................................................ 65
AVG: AI en rechten van data subject..........................................................................68
AI en compatibiliteit met AVG....................................................................................70
LES 11 PRIVACY BY DESIGN IN AI...................................................................72
INTRODUCTIE BEGINSELEN VAN GEGEVENSVERWERFIING:............................................................72
PRIVACY BY DESIGN............................................................................................................ 74
Gegevensbeschermingseffectbeoordeling (DPIA).......................................................74
, Anonieme gegevens................................................................................................... 75
PRIVACY BY DESIGN IN AI..................................................................................................... 75
Voorbeelden............................................................................................................... 76
LES 12: HARMONIZING HUMAN-AL SYNERGY..................................................77
HUMAN-MACHINE INTERACTION (HMI)....................................................................................77
THE CHALLENGES VAN AUTONOMIE (OF THE RISICO VAN SLECHTE HMI).........................................79
Automatisation bias -> 2 soorten...............................................................................80
Out-of-the-loop unfamiliarity (OOLU)..........................................................................80
UAS & AGV’s............................................................................................................... 80
Moral crumple zone.................................................................................................... 81
The cognitieve prijs van automatisatie.......................................................................81
FAST AND SLOW THINKING................................................................................................... 82
Vormen van cognitive biases.....................................................................................83
Compliance by desgin................................................................................................ 84
Automation complacency........................................................................................... 84
AI-(il)literacy en vatbaarheid voor beïnvloeding.........................................................85
Cognitive forcing functions......................................................................................... 86
Gebruiksgemak vs diepgang van de verwerking........................................................86
Designing for human-AI synergy................................................................................86
AI TRANSPARENCY, VERIFICATION & HUMAN OVERSIGHT.............................................................87
White box paradox..................................................................................................... 88
Algoritme aversie....................................................................................................... 88
Human-centered AI.................................................................................................... 88
De paradox van transparantie....................................................................................89
Wat maakt een AI-uitleg zinvol?.................................................................................89
HOW WILL AI CHANGE US?................................................................................................... 92
Bias amplificatie......................................................................................................... 93
A Third Knowledge Revolution?..................................................................................93
The Paradox of Personalization.................................................................................94
Digitale amnesie........................................................................................................ 95
AI als een sociale partner...........................................................................................96
Mechanismen van beïnvloeding: hoe AI engagement versterkt:...............................97
ENKELE TIPS VOOR HET EXAMEN...................................................................98
LES 1: INTRODUCTIE
, GESCHIEDENIS VAN AI
1950 turing test om Engima code te kraken (Duitsers) -> kunnen we nu menselijke
intelligentie benadere?
1960 → ELIZA chatbot, werkt met regels, gebruiker herkent alleen woorden in een
gestructureerd patroon
1970 → AI-winter, de funding ging weg
1980 → expertsystemen, via databanken antwoorden krijgen, werd gebruikt voor
diagnoses, experts exact kennis aan het systeem die het kon verwerken
1990 → Machine learning → boom
o Supervised learning: veel info, dingen kunnen voorstellen
(wat zijn chihuahua’s vs muffins)
Laat computer zelf programma’s schrijven
o Reinforcement learning
MACHINE LEARNING HOE?
1. Meetbare karakteristieken
2. Model opstellen
3. Dan met model voorstellingen doen
→ eenvoudig voorbeeld is belastingfraude controle
4. Model locatie en inkomen vergelijking
5. We bekomen een formule die zo goed mogelijk een uitkomst kan voorspellen
→ algoritme beginnen met random beginsituatie, aanpassen tot een goed model
Reinforcement learning
Principe: Een vorm van AI-training die niet gebaseerd is op een vooraf gelabelde
dataset, maar op interactie en feedback.
Methodiek: Werkt via een lus van Agent (AI), Omgeving (context) en Interpreter
(beloningsmechanisme).
o Actie → Feedback → Optimalisatie.
Toepassingen:
o spelletjes: Strategiebepaling in complexe omgevingen (bv. schaken of de
klassieke AI-demo Pong).
o Chatbots: Optimalisatie van antwoorden door menselijke feedback (RLHF),
wanneer statische data alleen onvoldoende nuance biedt.
Voordeel: Kan zelf strategieën ontdekken vanuit een willekeurige beginsituatie
(random start), mits er een duidelijk evaluatiekader is.
INTELLIGENTIE:
MAATSCHAPPELIJKE
UITDAGINGEN
Inhoudsopgave
LES 1: INTRODUCTIE.......................................................................................3
GESCHIEDENIS VAN AI........................................................................................................... 4
MACHINE LEARNING HOE?...................................................................................................... 4
WAT IS ER VERANDERD?........................................................................................................ 5
LES 2: UNDERSTANDING AND INTERPRETING DEEP NEURAL NETWORKS............5
BACKGROUND...................................................................................................................... 5
DESIGNING A DEEP NEURAL NETWORK......................................................................................6
DEEP NEURAL NETWORK VOOR VISUELE DATA............................................................................7
GENERATIVE MODELS............................................................................................................ 8
autoencoders............................................................................................................... 8
Variational Autoencoders (VAE)....................................................................................9
Generative adversial networks.....................................................................................9
Diffusion models........................................................................................................ 11
CHALLENGES VOOR AI......................................................................................................... 11
LES 3: RESPONSIBLE AI.................................................................................12
FLAT FLOW FOR DATA SCIENCE ETHICS....................................................................................13
AI RISKS........................................................................................................................... 14
Onmiddelijke risico’s:................................................................................................. 14
Systemic risks............................................................................................................ 16
Long term impact....................................................................................................... 17
DE WEG VOORUIT............................................................................................................... 19
LES 4: IMPORTANCE OF SAFETY IN AI DESGIN................................................19
WAAR ZIEN WE AI IN GEBRUIK?............................................................................................. 20
IF AI INTERACTS PHYSICALLY WITH HUMANS, HOW DO WE MAKE SURE THAT IT DOES IT SAFELY?.........20
PROBLEMEN MET HET GEBRUIKEN VAN RL...............................................................................23
DESIGNING SAFETY IN RL..................................................................................................... 24
SAFE DESIGN OF AI ENABLED ROBOTICS..................................................................................24
LES 5: FAIRNESS AND GENERATIVE AI............................................................25
RESPONSIBLE AI................................................................................................................. 25
VERSCHILLENDE SOORTEN BIAS:............................................................................................ 25
HOE MEET JE FAIRNESS?...................................................................................................... 27
GENERATIVE AI RISKS......................................................................................................... 28
LES 6 SUSTAINABLE AI..................................................................................30
AI FOR SUSTAINABILITY (DE AI GEBRUIKEN VOOR BETERE DUURZAAMHEID).....................................30
, DATA COLLECTIE VOOR MEER DUURZAAMHEID..........................................................................31
OVERVIEW OF OUR AI FOR SUSTAINABILITY PROJECTS................................................................32
Bio-accelerated mineral weathering (BAM!)...............................................................32
Future arctic............................................................................................................... 33
Global fertilizer dataset..............................................................................................34
Curieuze Neuze: enviromental monitoring.................................................................34
ICOS Brasschaat: Forest monitoring...........................................................................34
ROL VAN AI IN SLIMME INFRASTRUCTUUR................................................................................35
DE NEGATIEVE IMPACT VAN AI OP DUURZAAMHEID....................................................................36
LES 7 DESIGNING FUTURES: HUMAN-CENTERED DESIGN IN TECHNOLOGIES.....37
WORLDBUILDING................................................................................................................ 39
DESIGN FICTION................................................................................................................. 39
CRITICAL DESIGN................................................................................................................ 39
INVENTION FACTORIES......................................................................................................... 39
HET GROTE GEVAAR VAN AUTOMATION...................................................................................40
AUGMENTATION................................................................................................................. 40
LES 8: THE MORAL DANGERS AND OPPORTUNITIES OF GENERATIVE,
MULTIMODAL LLMS AND OTHER AI................................................................41
3 GEBIEDEN VAN FILOSOFIE.................................................................................................. 41
ACCELERATED INNOVATION IN AI........................................................................................... 42
ETHIEK............................................................................................................................. 44
EPISTEMOLOGIE.................................................................................................................. 45
METAFYSICA...................................................................................................................... 46
THE ROLE OF AI................................................................................................................. 48
Spectrum of mind....................................................................................................... 48
Ethische implicaties.................................................................................................... 49
Epistemologische implicaties.....................................................................................49
Metafysische implicaties............................................................................................50
LES 9 AI EN RECHT........................................................................................50
ROL+RECHTSHANDHAVING................................................................................................... 50
UITDAGINGEN VAN AI VOOR ANTIDISCRIMINATIERECHT...............................................................50
UITDAGINGEN VAN AI VOOR AUTEURSRECHT............................................................................51
UITDAGINGEN VAN AI VOOR CONSUMENTENRECHT.....................................................................53
LES 10 DE REGULERING VAN AI IN EUROPA....................................................54
EU STRATEGIE VOOR AI:...................................................................................................... 54
AI-VERORDENING............................................................................................................... 54
Doelstellingen............................................................................................................ 54
Definities.................................................................................................................... 54
risk-based approach................................................................................................... 55
AI EN AANSPRAKELIJKHEID.................................................................................................... 64
AI EN GEGEVENSCBESCHERMING............................................................................................ 65
AVG............................................................................................................................ 65
AVG: AI en rechten van data subject..........................................................................68
AI en compatibiliteit met AVG....................................................................................70
LES 11 PRIVACY BY DESIGN IN AI...................................................................72
INTRODUCTIE BEGINSELEN VAN GEGEVENSVERWERFIING:............................................................72
PRIVACY BY DESIGN............................................................................................................ 74
Gegevensbeschermingseffectbeoordeling (DPIA).......................................................74
, Anonieme gegevens................................................................................................... 75
PRIVACY BY DESIGN IN AI..................................................................................................... 75
Voorbeelden............................................................................................................... 76
LES 12: HARMONIZING HUMAN-AL SYNERGY..................................................77
HUMAN-MACHINE INTERACTION (HMI)....................................................................................77
THE CHALLENGES VAN AUTONOMIE (OF THE RISICO VAN SLECHTE HMI).........................................79
Automatisation bias -> 2 soorten...............................................................................80
Out-of-the-loop unfamiliarity (OOLU)..........................................................................80
UAS & AGV’s............................................................................................................... 80
Moral crumple zone.................................................................................................... 81
The cognitieve prijs van automatisatie.......................................................................81
FAST AND SLOW THINKING................................................................................................... 82
Vormen van cognitive biases.....................................................................................83
Compliance by desgin................................................................................................ 84
Automation complacency........................................................................................... 84
AI-(il)literacy en vatbaarheid voor beïnvloeding.........................................................85
Cognitive forcing functions......................................................................................... 86
Gebruiksgemak vs diepgang van de verwerking........................................................86
Designing for human-AI synergy................................................................................86
AI TRANSPARENCY, VERIFICATION & HUMAN OVERSIGHT.............................................................87
White box paradox..................................................................................................... 88
Algoritme aversie....................................................................................................... 88
Human-centered AI.................................................................................................... 88
De paradox van transparantie....................................................................................89
Wat maakt een AI-uitleg zinvol?.................................................................................89
HOW WILL AI CHANGE US?................................................................................................... 92
Bias amplificatie......................................................................................................... 93
A Third Knowledge Revolution?..................................................................................93
The Paradox of Personalization.................................................................................94
Digitale amnesie........................................................................................................ 95
AI als een sociale partner...........................................................................................96
Mechanismen van beïnvloeding: hoe AI engagement versterkt:...............................97
ENKELE TIPS VOOR HET EXAMEN...................................................................98
LES 1: INTRODUCTIE
, GESCHIEDENIS VAN AI
1950 turing test om Engima code te kraken (Duitsers) -> kunnen we nu menselijke
intelligentie benadere?
1960 → ELIZA chatbot, werkt met regels, gebruiker herkent alleen woorden in een
gestructureerd patroon
1970 → AI-winter, de funding ging weg
1980 → expertsystemen, via databanken antwoorden krijgen, werd gebruikt voor
diagnoses, experts exact kennis aan het systeem die het kon verwerken
1990 → Machine learning → boom
o Supervised learning: veel info, dingen kunnen voorstellen
(wat zijn chihuahua’s vs muffins)
Laat computer zelf programma’s schrijven
o Reinforcement learning
MACHINE LEARNING HOE?
1. Meetbare karakteristieken
2. Model opstellen
3. Dan met model voorstellingen doen
→ eenvoudig voorbeeld is belastingfraude controle
4. Model locatie en inkomen vergelijking
5. We bekomen een formule die zo goed mogelijk een uitkomst kan voorspellen
→ algoritme beginnen met random beginsituatie, aanpassen tot een goed model
Reinforcement learning
Principe: Een vorm van AI-training die niet gebaseerd is op een vooraf gelabelde
dataset, maar op interactie en feedback.
Methodiek: Werkt via een lus van Agent (AI), Omgeving (context) en Interpreter
(beloningsmechanisme).
o Actie → Feedback → Optimalisatie.
Toepassingen:
o spelletjes: Strategiebepaling in complexe omgevingen (bv. schaken of de
klassieke AI-demo Pong).
o Chatbots: Optimalisatie van antwoorden door menselijke feedback (RLHF),
wanneer statische data alleen onvoldoende nuance biedt.
Voordeel: Kan zelf strategieën ontdekken vanuit een willekeurige beginsituatie
(random start), mits er een duidelijk evaluatiekader is.
