4. Why AI?
4
Photo by Ryoji Iwata on Unsplash
Human digital
Face to face
Digital first
5. 5
Case 1: Proactive Profiling
High chance of finding
a job within 6 months
Low chance of finding
a job within 6 months
Non active jobseeker Active jobseeker
6. 6
Case 1: Proactive Profiling
High chance on finding
a job within 6 months
Low chance on finding
a job within 6 months
Non active jobseeker Active jobseeker
High chance of finding
a job within 6 months
Low chance of finding
a job within 6 months
7. 7
Case 2: JobNet: Matching on Deep Learning Neural Network
profile data
job data
labels (clicks)
Match in 4 ways
(distance between vectors)
Job vector
Profile vector
[0.3 0.2 0.9 0.1 0.4 ... ]
[0.1 0.9 0.3 0.3 0.2 … ]
11. 11
AI 4 Good Needs Collaboration
Research & Development
AI Team
Data Protection Officer
Citizens
12. 12
AI 4 Good Needs Collaboration
Research & Development
AI Team
Data Protection Officer
Citizens
Business project leader AI
13. 13
AI 4 Good Needs Collaboration
Research & Development
AI Team
Data Protection Officer
Citizens
Start = goal project,
Privacy & ethics by design
Data & AI literacy
14. 14
AI 4 Good Needs Collaboration
Research & Development
AI Team
Data Protection Officer
Citizens
Oversight: get data
protection right
Risk assessmentData & AI literacy
Start = goal project,
Privacy & ethics by design
15. 15
AI 4 Good Needs Collaboration
Research & Development
AI Team
Data Protection Officer
Citizens
Goal AI project
Privacy & ethics by design
Asking the right questions
Risk assessment
Measuring data driven
ethics
Bias assessments
Exploit
Experiment
Explore
Experimental
bias
Design
issues
Execution
issues
Data
bias
Model bias
Communication
failures
Operational
bias
Conception
flaws
Use case is known
(from business) vs.
use case is not known
16. 16
AI 4 Good Needs Collaboration
Research & Development
AI Team
Measuring data driven
ethics
Bias assessments
Data Protection Officer
Citizens
Oversight: get data
protection right
Risk assessment: use cases
16
Co-creation via pilots,
POC’s, user testing, ...
AI Knowledge Center
Data & AI literacy
Start = goal project,
Privacy & ethics by design
17. 17
AI 4 Good Needs Collaboration
Research & Development
AI Team
Measuring data driven
ethics
Bias assessments
Data Protection Officer
Citizens
Oversight: get data
protection right
Risk assessment: use cases
17
Co-creation via pilots,
POC’s, user testing, ...
AI Knowledge Center
HR/Communication:
a culture to speak up
Data & AI literacy
Start = goal project,
Privacy & ethics by design
18. 18
Does your organisation develop AI products and
services according to ‘privacy and ethics by design’?
Photo by Element5 Digital on Unsplash
19. 19
Trustworthy AI is built on the strong foundations of ethics,
lawfulness and robust AI development and maintenance.
Photo by Shivendu Shukla on Unsplash
Notas del editor
3 ethical principles
Understand - measure - action & prevention
VDAB needs to invest in both technical and non-technical methods to: ensure proper visibility through the AI lifecycle, generate awareness of potential bias, embed the appropriate controls proactively, help teams make the right decisions to mitigate against unintended outputs and to ensure trustworthy AIs, meeting your responsible AI requirements.
AI 4 GOOD is een continu process: data & modellen evolueren en bias kan op elk moment opduiken.