DutchMLSchool 2022 - Citizen Development in AI

July 4 - 6, 2022
2 n d E d i t i o n

#DutchMLSchool
AI/ML and Humans
Why we need citizen developers
Jan Veldsink Msc

CAIO Grio

NYENRODE. A REWARD FOR LIFE
“While cutting edge technology and talent are certainly
needed, it is equally important to align a company's
 
culture, structure and practices

to support the adoption of broad data analytics and AI/ML.”

Harvard Business Review, August 2019

Jan W
. Veldsink MSc Kinda CONFIDENTIAL
Engineering Trust

What is technology without trust? Digital business requires a resilient and
ef
f
icient IT foundation at its core. Without a well-designed base, there is no
way to scale cost-ef
f
iciently.

IT is responsible for engineering the trust necessary in our connected
world with our
f
irst four trends.
Sculpting Change

With the trusted foundation in place, the next focus is technologies
that enable the organization to scale its digitalization efforts.

But IT cannot match the pace of change alone. Fusion teams — made up of IT
and business staff — will collaborate and drive innovation to rapidly digitize the
business. IT’s job is to provide the tools to allow fusion teams to sculpt the
change, as our next trends show.
Accelerating Growth

When the foundation and building blocks are established, it’s time to focus on
technology trends that maximize the value of what the organization creates.

These technologies exemplify the IT force multipliers that will win business
and market share.
Data Fabric

Cybersecurity Mesh

Privacy-Enhancing Computation(MPC)

Cloud-Native Platforms

Composable Applications

Decision Intelligence

Hyperautomation

AI Engineering

Distributed Enterprise

Total Experience

Autonomic Systems

Generative AI

Jan W
The Holy Grail: General (Strong) AI

Jan W
From years to weeks

Jan W
Artificial Intelligence / Cognitive tasks

Jan W
Process

Jan W
AI/ML A combined Business and IT task
Scoping
Knowledge
aquisition
Creating Model(s)
Data / Interface
changes?
Specify change
in a RFC
IT change
Validate Model Apply Model
Data for
model
Decision
Engineering
ICT
change
Define/ Train/
Test Model
Deliver change
Testing
changes
Yes
No
Change done
RFC

Jan W
Software has eaten the world

Jan W
AI/ML is eating software

Jan W
AI versus Programming
Programming Without AI Programming With AI
A computer program without AI can answer the
specific questions it is meant to solve.
A computer program with AI can answer the

generic questions it is meant to solve.
Modification in the program leads to change in its
structure.
AI programs can absorb new modifications by putting
highly independent pieces of information together.
Hence you can modify even a minute piece of
information of program without affecting its structure.
Modification is time consuming and difficult.

It may lead to affecting the program adversely.
Quick and Easy program modification.

Jan W
Task we looked at…
47%

30%

16%

7%
Only 10% of organizations use data analytics in all business processes and decision making

(Source: Harvard Business Review, Data analytics survey, 2019)

Jan W
And things get out of hand!

Jan W
What happend
• We see fraud with dual nationality households

• Therefore we train a model that concludes there is a strong correlation between Dual
nationality and fraud

• Objective rule:

• If dual nationality then suspicious for fraud

• Consequence: The computer says suspect so suspend the payout of allowances, then
investigate. But the computer said NO.. so where smoke .. there fire…
Suspect
Fraud
Dual
Nationality
Then

Jan W
We Turn to ETHICS
Step

1
“Let’s predict
 
milk production!”
Evaluation
Modeling
Data
Understanding
Data
Preparation
Business
Understanding
Deployment

Jan W
And Acceptance / governance

Jan W
So the solution, Human in the loop!

Jan W

Jan W
Why do we need humans in the loop
• There are limits to how smart AI can be

• We can see the bigger picture

• Systems Thinking

• Goal setting

• We understand ethics

• We can attribute meaning to correlations
to build causal relations, as features to
make AI become better

• Mutual learning

• Augmented AI

Jan W
Human domain experts in the loop!
Human domain experts can
 
attribute meaning to correlations
 
to build causal relations,
 
as features to improve AI

Jan W
Why do we need humans in the loop?

Jan W
A Citizen developer
What is Citizen Developer?

A citizen developer is a user who creates new business applications for consumption
by others using development and runtime environments sanctioned by corporate IT. In
the past, end-user application development has typically been limited to single-user
or workgroup solutions built with tools like Microsoft Excel and Access.  
 
However, today, end users can build departmental, enterprise and even public
applications using shared services, higher level languages, RPA style development
platforms, Self service BI platforms, Machine Learning Platforms like BigML and cloud
computing services.

Jan W
AI/ML— Design and change
Domain Expertise
Business Engineer
Domain Data Expertise
Data Engineer
ML/AI Expertise
Decision Engineer

Jan W
Co-creation Business - IT

Jan W
Find new perspectives in systems thinking

Jan W
A system is!

Jan W
and DATA

Jan W
SOCIAL
Data is collected
through a
multitude of
channels and
applications

This requires
attributing
meaning from
across the
organization
ANALYTICS
CLOUD
INTERNET of THINGS
ARTIFICIAL
INTELLIGENCE
BIOMETRICS
MOBILE

Jan W
Data Information Knowledge Wisdom (DIKW)

Jan W
Data 2 Wisdom

Jan W
Looks like Decision Engineering

Jan W
Correlation not causation

Jan W
Logic

Jan W
Time to think on reasoning
Deductive Inductive Abductive
Major Premise All Men are Mortal Most Greeks Have Beards Observation: That Man Has a Beard
Minor Premise Socrates is a Man Socrates is a Greek Known Fact: Most Greeks Have Beards
Conclusion
(Inference)
It is Certain that: Socrates is Mortal

(this is logically certain given the
premises; if all men are mortal, then
Socrates being a man must be
mortal. Here you can see that if a
premise is false, deduction can
produce false conclusions).
It is “likely” that: Socrates has a beard

 
(given the premises, the conclusion can be
assigned a likelihood; this argument isn’t very
compelling, but to explain that quality of induction
here would be a rabbit hole).
Perhaps: This Man is Greek
(a hypothesis based on an observation and
a known fact).

Jan W
Feedback loop!

Jan W
Okay.. and then???.. Systems thinking

Jan W
Systems Thinking

Jan W
What banks are doing.
• In a lot of money laundering cases
there are cash deposits involved

• We create a simple model that
implements

• And call it an Objective rule:

• If cash > 3.000 then suspicious
Money
laundering
Cash
deposits
Then

Jan W
What banks are doing
Money
laundering
Cash
deposits
Then
• Causal questioning:


• What other factor(s) makes that
depositing cash is part of a Money
laundering schema?
Money
laundering
Cash
deposits
Factor?
Involves

• We create a simple model that
implements

• And call it an Objective rule:

• If cash > 3.000 then suspicious

Jan W
Feedback as essential design principle?

Jan W
Understanding causal loops
Source: Peter Senge, The
fi
fth discipline

Jan W
Systems Thinking and AI development
• Problem oriented

• What to be solved

• Looking for causal relations

• Fed by observations / models or principles

• System and AI development

• Case based reasoning

• Causal reasoning

• Meaning attribution to found correlations

• Creating actionable insights

Jan W
Start building causal systems

Jan W
Perhaps Automated causal reasoning?
• The really challenging problems are still ahead: We still do not have a causal
understanding of poverty and cancer and intolerance, and only the accumulation
of data and the insight of great minds will eventually lead to such understanding.

• The data is all over the place, the insight is yours, and now an abacus is at your
disposal, too. I hope the combination amplifies each of these components.

Jan W
Implements Causal features
Associations
Correlations
Asses the
correlations,
associations
and cases
looking for
causal relations
Formulate
causal relations
in new features
Create augmented
dataset
Create
Model(s)
Dataset

Jan W
References
• Systems Thinking: https://thesystemsthinker.com

•

• What is a system: https://www.youtube.com/watch?v=EbLh7rZ3rhU&t=179s

• Citizen development

•

• https://scholarspace.manoa.hawaii.edu/server/api/core/bitstreams/7d661602-4f05-49dc-b9b5-e13c84385dfc/content

• Causal Reasoning:

•

Jan W
No time to waste
www.bigml.com

Modular MBA - AISEC - Artificial Intelligence and Security

This fall at Nyenrode!

DutchMLSchool 2022 - Citizen Development in AI

More Related Content

Similar to DutchMLSchool 2022 - Citizen Development in AI (20)

More from BigML, Inc (20)

Recently uploaded (20)

DutchMLSchool 2022 - Citizen Development in AI