Samenvatting Korfvak
Artificiële Intelligentie: maatschappelijke
uitdagingen
Universiteit Antwerpen
Dit document bevat theorie, notities, figuren, uitleg en voorbeelden. Achteraan is er een
oefentoets. Alles wat je nodig hebt om te slagen voor het examen.
De symbolen maken het makkelijker om de leerstof te onthouden.
1
,Inhoudsopgave
Les 1: inleiding ............................................................................................................................. 8
Geschiedenis .................................................................................................................................. 8
📜 Korte geschiedenis van AI ......................................................................................................... 8
🤖 Hoe werkt Machine Learning?................................................................................................... 9
💡 Extra uitleg: ............................................................................................................................. 9
🧠 Belangrijke mijlpalen in AI....................................................................................................... 10
🔧 Wat maakte deze sprongen mogelijk? ..................................................................................... 10
🧠 Recurrent Neural Networks (RNNs) ......................................................................................... 11
🧠 Deep Neural Networks (DNNs) ............................................................................................... 11
🧠 Very Deep Neural Networks .................................................................................................... 11
🔑 Belangrijk idee: Pre-training .................................................................................................... 11
🧪 Stap 1: Pre-training (auto-encoder) ......................................................................................... 11
🎯 Stap 2: Fine-tuning op echte taak ............................................................................................ 12
💡 Waarom dit nuttig is: .............................................................................................................. 12
🤖 Deep Neural Networks – kernpunten ....................................................................................... 13
Les 2: Understanding and Interpreting Deep Neural Networks ....................................................... 14
📌 1. Introductie en achtergrond ................................................................................................. 14
🧱 2. Architectuur van Deep Neural Networks (DNNs) .................................................................. 14
🖼 3. DNNs voor visuele data (zoals afbeeldingen) ....................................................................... 15
🔍 4. Feature-extractie in CNNs .................................................................................................. 15
⚙ 5. Generatieve modellen ........................................................................................................ 16
🔄 Autoencoder – Samenvatting .................................................................................................. 16
❗ 6. Belangrijke uitdagingen....................................................................................................... 17
🧠 7. Interpretability en Explainable AI (XAI) ................................................................................. 17
✅ 8. Samenvatting ..................................................................................................................... 18
Les 3: responsible AI .................................................................................................................... 18
🧠
📊 Wat is Responsible AI? ...............................................................................................................20
⚖ Basisbegrippen ..........................................................................................................................20
🔁 Waarom is Responsible AI belangrijk? .........................................................................................20
⚠ FAT-Flow: Ethische principes in het data science proces ..............................................................20
Belangrijkste risico’s ...................................................................................................................21
📌 Immediate Risks .................................................................................................................... 21
🏢 Systemic Risks....................................................................................................................... 21
🔎🌍 Existentiële Risico’s................................................................................................................ 21
👥Uitlegbaarheid (Explainable AI)....................................................................................................21
💡Vooroordelen en discriminatie ....................................................................................................21
🧭Oplossingen en strategieën .........................................................................................................21
De weg vooruit ...........................................................................................................................22
🧠
Les4: Importance of Safety in the design of AI systems .................................................................. 22
🔍Supervised Learning ...................................................................................................................22
🤖
Unsupervised Learning ...............................................................................................................22
Wat is Reinforcement Learning (RL)? ...........................................................................................23
2
, 🌍 RL in de praktijk ..........................................................................................................................23
⚠ Risico’s van RL in de echte wereld ...............................................................................................23
🧱 De basis van RL: Gridworld .........................................................................................................24
📐 Wiskundige onderbouw: Markov Decision Processes (MDP) .........................................................24
📊 Value Iteration & Q-learning ........................................................................................................25
🔐 Waarom veiligheid cruciaal is in RL ..............................................................................................25
🧠 Leren van mensen ......................................................................................................................25
🛡 Strategieën voor veilige AI ...........................................................................................................25
🚗 Voorbeelden van veilige RL-toepassingen ....................................................................................26
🧭 Conclusie: de weg vooruit ...........................................................................................................26
Les 5: Fairness and genAI ............................................................................................................. 26
🤖 Wat is Responsible AI? ...............................................................................................................26
⚖ Wat betekent Fairness in AI? .......................................................................................................26
📉
Bronnen van Bias in Data ............................................................................................................27
📊
Hoe meet je Bias in Data? ...........................................................................................................27
🧪
Voorbeeldanalyse: hoe eerlijk zijn modellen? ...............................................................................27
📏
Fairness-metric 1: Demographic Parity ........................................................................................27
⚖
Fairness-metric 2: Equalized Opportunity & Equalized Odds .........................................................28
🧠
Ethische keuzes & juridische uitdagingen.....................................................................................28
📍
Voorbeeld: COMPAS – bias in strafrecht .......................................................................................28
👁
Oplossing: Human in the Loop & Transparantie ............................................................................28
🧬
⚠ Wat is Generative AI? ..................................................................................................................28
📈 Risico’s van Generative AI ...........................................................................................................29
ChatGPT verhoogt productiviteit (maar niet zonder risico) .............................................................29
Les🌱
6: AI for sustainability. ........................................................................................................... 29
🧩 Wat is Sustainable AI? ................................................................................................................29
🌍Vier kernprincipes van Sustainable AI ..........................................................................................30
🛰Wat is AI for Sustainability? .........................................................................................................30
🔬Hoe verzamelen we data? ...........................................................................................................30
Voorbeelden van AI-onderzoek voor duurzaamheid ......................................................................31
🪨 Bio-accelerated Mineral Weathering (BAM!) ............................................................................ 31
❄ Future Arctic – Klimaatonderzoek ........................................................................................... 31
🌍 Global Fertilizer Dataset ......................................................................................................... 31
🌳 CurieuzeNeuzen in de Tuin ..................................................................................................... 31
🧠🌲 ICOS Brasschaat – Bosmonitoring .......................................................................................... 31
⚠Brede toepassingen van AI voor duurzaamheid ............................................................................32
Negatieve milieu-impact van AI ...................................................................................................32
3
, 📜 Beleidskader: EU AI Act (2024) ....................................................................................................32
💡 Oplossingen voor duurzamere AI .................................................................................................32
🧭 Samenvatting voor het examen (volgens jouw notities) .................................................................33
Les 7: Designing Futures: human-centered design in technologies ................................................ 33
🎯 Thema: AI Design Futures ...........................................................................................................33
🤖 Humane AI-producten — Hype vs. realiteit...................................................................................33
🧠 AI & Ontwerpen vandaag ............................................................................................................34
🧱 Wat maakt een AI-product écht innovatief?..................................................................................34
🔮 Futures Thinking – Ontwerpen voor de toekomst ..........................................................................34
Wat is futures thinking? .............................................................................................................. 34
🌍 Worldbuilding – Werelden creëren ...............................................................................................34
📦 Design Fiction – Fictieve prototypes.............................................................................................34
🧨 Critical Design – Technologie bevragen ........................................................................................35
🧪 Voorbeeldinstituut: MIT Media Lab ..............................................................................................35
🧬
Convergentie van disciplines ......................................................................................................35
🦾
Automatisering vs. Augmentatie ..................................................................................................35
💡
Vier vormen van Human Augmentation ........................................................................................35
🔚
Conclusie: wat leer je hieruit? .....................................................................................................36
Les 8: DE MORELE GEVAREN EN KANSEN VAN GENERATIEVE, MULTIMODALE LLMS EN ANDERE AI .. 36
⚖
Thema: Ethiek, Epistemologie & Metafysica in AI ..........................................................................36
🧩
1. Ethische vragen – “Ought implies can” .....................................................................................36
📚
🧠 2. Epistemologische vragen – Wat is kennis? ................................................................................36
🧠 3. Metafysische vragen – Wat is echt? Wat betekent ‘bestaan’? .....................................................37
🌐 AI zet taal om in geometrie ..........................................................................................................37
🧬 AI als multimodaal vertaalstation ................................................................................................37
🐝 Collectieve intelligentie: zijn wij deel van iets groters? ..................................................................38
🧠CASE STUDY 1: Dierlijke intelligentie en communicatie .................................................................38
🧠CASE STUDY 2: Bewustzijn, morfogenese & AI..............................................................................38
🧭Metafysische implicaties van AI...................................................................................................38
📌Samenvattend – Drie domeinen in interactie ................................................................................39
Mogelijke examenvragen (uit de slides) ........................................................................................39
Les⚖
9: AI en recht ......................................................................................................................... 39
🏛Recht vs. Ethiek ..........................................................................................................................39
👩⚖Domeinen van het recht (relevant bij AI) .......................................................................................39
🚫
Handhaving van het recht ...........................................................................................................40
Antidiscriminatierecht en AI ........................................................................................................40
4
Artificiële Intelligentie: maatschappelijke
uitdagingen
Universiteit Antwerpen
Dit document bevat theorie, notities, figuren, uitleg en voorbeelden. Achteraan is er een
oefentoets. Alles wat je nodig hebt om te slagen voor het examen.
De symbolen maken het makkelijker om de leerstof te onthouden.
1
,Inhoudsopgave
Les 1: inleiding ............................................................................................................................. 8
Geschiedenis .................................................................................................................................. 8
📜 Korte geschiedenis van AI ......................................................................................................... 8
🤖 Hoe werkt Machine Learning?................................................................................................... 9
💡 Extra uitleg: ............................................................................................................................. 9
🧠 Belangrijke mijlpalen in AI....................................................................................................... 10
🔧 Wat maakte deze sprongen mogelijk? ..................................................................................... 10
🧠 Recurrent Neural Networks (RNNs) ......................................................................................... 11
🧠 Deep Neural Networks (DNNs) ............................................................................................... 11
🧠 Very Deep Neural Networks .................................................................................................... 11
🔑 Belangrijk idee: Pre-training .................................................................................................... 11
🧪 Stap 1: Pre-training (auto-encoder) ......................................................................................... 11
🎯 Stap 2: Fine-tuning op echte taak ............................................................................................ 12
💡 Waarom dit nuttig is: .............................................................................................................. 12
🤖 Deep Neural Networks – kernpunten ....................................................................................... 13
Les 2: Understanding and Interpreting Deep Neural Networks ....................................................... 14
📌 1. Introductie en achtergrond ................................................................................................. 14
🧱 2. Architectuur van Deep Neural Networks (DNNs) .................................................................. 14
🖼 3. DNNs voor visuele data (zoals afbeeldingen) ....................................................................... 15
🔍 4. Feature-extractie in CNNs .................................................................................................. 15
⚙ 5. Generatieve modellen ........................................................................................................ 16
🔄 Autoencoder – Samenvatting .................................................................................................. 16
❗ 6. Belangrijke uitdagingen....................................................................................................... 17
🧠 7. Interpretability en Explainable AI (XAI) ................................................................................. 17
✅ 8. Samenvatting ..................................................................................................................... 18
Les 3: responsible AI .................................................................................................................... 18
🧠
📊 Wat is Responsible AI? ...............................................................................................................20
⚖ Basisbegrippen ..........................................................................................................................20
🔁 Waarom is Responsible AI belangrijk? .........................................................................................20
⚠ FAT-Flow: Ethische principes in het data science proces ..............................................................20
Belangrijkste risico’s ...................................................................................................................21
📌 Immediate Risks .................................................................................................................... 21
🏢 Systemic Risks....................................................................................................................... 21
🔎🌍 Existentiële Risico’s................................................................................................................ 21
👥Uitlegbaarheid (Explainable AI)....................................................................................................21
💡Vooroordelen en discriminatie ....................................................................................................21
🧭Oplossingen en strategieën .........................................................................................................21
De weg vooruit ...........................................................................................................................22
🧠
Les4: Importance of Safety in the design of AI systems .................................................................. 22
🔍Supervised Learning ...................................................................................................................22
🤖
Unsupervised Learning ...............................................................................................................22
Wat is Reinforcement Learning (RL)? ...........................................................................................23
2
, 🌍 RL in de praktijk ..........................................................................................................................23
⚠ Risico’s van RL in de echte wereld ...............................................................................................23
🧱 De basis van RL: Gridworld .........................................................................................................24
📐 Wiskundige onderbouw: Markov Decision Processes (MDP) .........................................................24
📊 Value Iteration & Q-learning ........................................................................................................25
🔐 Waarom veiligheid cruciaal is in RL ..............................................................................................25
🧠 Leren van mensen ......................................................................................................................25
🛡 Strategieën voor veilige AI ...........................................................................................................25
🚗 Voorbeelden van veilige RL-toepassingen ....................................................................................26
🧭 Conclusie: de weg vooruit ...........................................................................................................26
Les 5: Fairness and genAI ............................................................................................................. 26
🤖 Wat is Responsible AI? ...............................................................................................................26
⚖ Wat betekent Fairness in AI? .......................................................................................................26
📉
Bronnen van Bias in Data ............................................................................................................27
📊
Hoe meet je Bias in Data? ...........................................................................................................27
🧪
Voorbeeldanalyse: hoe eerlijk zijn modellen? ...............................................................................27
📏
Fairness-metric 1: Demographic Parity ........................................................................................27
⚖
Fairness-metric 2: Equalized Opportunity & Equalized Odds .........................................................28
🧠
Ethische keuzes & juridische uitdagingen.....................................................................................28
📍
Voorbeeld: COMPAS – bias in strafrecht .......................................................................................28
👁
Oplossing: Human in the Loop & Transparantie ............................................................................28
🧬
⚠ Wat is Generative AI? ..................................................................................................................28
📈 Risico’s van Generative AI ...........................................................................................................29
ChatGPT verhoogt productiviteit (maar niet zonder risico) .............................................................29
Les🌱
6: AI for sustainability. ........................................................................................................... 29
🧩 Wat is Sustainable AI? ................................................................................................................29
🌍Vier kernprincipes van Sustainable AI ..........................................................................................30
🛰Wat is AI for Sustainability? .........................................................................................................30
🔬Hoe verzamelen we data? ...........................................................................................................30
Voorbeelden van AI-onderzoek voor duurzaamheid ......................................................................31
🪨 Bio-accelerated Mineral Weathering (BAM!) ............................................................................ 31
❄ Future Arctic – Klimaatonderzoek ........................................................................................... 31
🌍 Global Fertilizer Dataset ......................................................................................................... 31
🌳 CurieuzeNeuzen in de Tuin ..................................................................................................... 31
🧠🌲 ICOS Brasschaat – Bosmonitoring .......................................................................................... 31
⚠Brede toepassingen van AI voor duurzaamheid ............................................................................32
Negatieve milieu-impact van AI ...................................................................................................32
3
, 📜 Beleidskader: EU AI Act (2024) ....................................................................................................32
💡 Oplossingen voor duurzamere AI .................................................................................................32
🧭 Samenvatting voor het examen (volgens jouw notities) .................................................................33
Les 7: Designing Futures: human-centered design in technologies ................................................ 33
🎯 Thema: AI Design Futures ...........................................................................................................33
🤖 Humane AI-producten — Hype vs. realiteit...................................................................................33
🧠 AI & Ontwerpen vandaag ............................................................................................................34
🧱 Wat maakt een AI-product écht innovatief?..................................................................................34
🔮 Futures Thinking – Ontwerpen voor de toekomst ..........................................................................34
Wat is futures thinking? .............................................................................................................. 34
🌍 Worldbuilding – Werelden creëren ...............................................................................................34
📦 Design Fiction – Fictieve prototypes.............................................................................................34
🧨 Critical Design – Technologie bevragen ........................................................................................35
🧪 Voorbeeldinstituut: MIT Media Lab ..............................................................................................35
🧬
Convergentie van disciplines ......................................................................................................35
🦾
Automatisering vs. Augmentatie ..................................................................................................35
💡
Vier vormen van Human Augmentation ........................................................................................35
🔚
Conclusie: wat leer je hieruit? .....................................................................................................36
Les 8: DE MORELE GEVAREN EN KANSEN VAN GENERATIEVE, MULTIMODALE LLMS EN ANDERE AI .. 36
⚖
Thema: Ethiek, Epistemologie & Metafysica in AI ..........................................................................36
🧩
1. Ethische vragen – “Ought implies can” .....................................................................................36
📚
🧠 2. Epistemologische vragen – Wat is kennis? ................................................................................36
🧠 3. Metafysische vragen – Wat is echt? Wat betekent ‘bestaan’? .....................................................37
🌐 AI zet taal om in geometrie ..........................................................................................................37
🧬 AI als multimodaal vertaalstation ................................................................................................37
🐝 Collectieve intelligentie: zijn wij deel van iets groters? ..................................................................38
🧠CASE STUDY 1: Dierlijke intelligentie en communicatie .................................................................38
🧠CASE STUDY 2: Bewustzijn, morfogenese & AI..............................................................................38
🧭Metafysische implicaties van AI...................................................................................................38
📌Samenvattend – Drie domeinen in interactie ................................................................................39
Mogelijke examenvragen (uit de slides) ........................................................................................39
Les⚖
9: AI en recht ......................................................................................................................... 39
🏛Recht vs. Ethiek ..........................................................................................................................39
👩⚖Domeinen van het recht (relevant bij AI) .......................................................................................39
🚫
Handhaving van het recht ...........................................................................................................40
Antidiscriminatierecht en AI ........................................................................................................40
4
