Artificial Intelligence From a LEA perspective...Artificial Intelligence From a LEA perspective Oscar Wijsman Business Expert Intelligence Netherlands Police @oscarwijsman Netherlands

Artificial Intelligence

From a LEA perspective

Oscar Wijsman

Business Expert Intelligence

Netherlands Police

@oscarwijsman

Netherlands Police

one single police force since 2012

with 65.000 colleagues

10 Regional Units

1 National Unit

1 Police Service Centre

75 Data Scientists

10 PhD candidates

22

What does AI mean to us?

Many definitions & focal points worldwide….

• Human design and/or human in control

• Machines that can reason

• Taking action ≠ taking decision

• What is “autonomous”

• Experimenting ≠ will use it

33

How are we involved

1. Working group on AI – strategic level

2. Political & legislative discussions

e.g. NL Strategic Action Plan AI, EU EC

3. NL Police AI lab – linked with academia

4. AI is part of our Data Science activities

both experiments and building applications

44

Collaboration among various universities

Part of ICAI https://icai.ai/police-lab-ai/

10 PhD candidates, 2 post-docs and 2 programming teams

Goal: developing AI knowledge for law enforcement to steer towards a more

effective, fair and transparent police

National Police AI LabNational Police AI Lab

55

For good and for bad…

AI in the lab – concept development1. Understand, explain decisions or decision support, reasoning

2. Ensure legal and ethical correctness, guidelines, piloting, learning by doing

3. Get insight in phenomena, behaviour, pitfalls

and in the operation – daily use1. Image, text & pattern recognition & interpretation, entity extraction (ML and DL)

2. Unravel complex (criminal) networks, relations

3. Task automation (narrow AI)

But our criminal adversaries will use it too…

66

AI applications for LEA’s can be

1. Biometrics (face, sound, movement, fingers, feet..)

2. NLP (multi-lingual, automatic translation)

3. Fraud detection

4. Recommendation engines

5. Deep fakes (detecting)

6. Malware & (prevent) cyber attacks

7. Complex networks and more…

77

Ethics are important

and must be at the core of how we use AI

But many approaches and frameworks…

88

https://cyber.harvard.edu/publication/2020/principled-aihttps://cyber.harvard.edu/publication/2020/principled-ai

99

https://cyber.harvard.edu/publication/2020/principled-aihttps://cyber.harvard.edu/publication/2020/principled-ai

1010

https://cyber.harvard.edu/publication/2020/principled-ai. https://cyber.harvard.edu/publication/2020/principled-ai.

1111

How many frameworks do we need?

Which ones should a police Data Scientists or Engineer be conscious of?

1212

What role do Interpol, Europol and national LEA’s have?

What would our (correct) order of challenges be?12

1313

1414

In the future we will have1. Strategy and commitment, both LEA and EU LEA’s

2. Better legal framework & regulation, human centric

3. Cross-border (cloud) infrastructure, open source

4. International cooperation and sharing of (our own)

knowledge, best practise and solutions AI platform

5. Unbiased data sets / common ontology, taxonomy

6. Research hub

7. New and different role for Europol (and Interpol)

1515

Opportunities

1616

Risks

1717

1818

Questions in the panelQuestions in the panel

1919