Contemporary AI engenders hopes and fears – hopes of harnessing AI for productivity growth and innovation – fears of mass unemployment and conflict between humankind and an artificial super-intelligence. Before we let AI drive our hopes and fears, we need to understand what it is and what it is not. Then we need to understand how to implement AI in an ethical and responsible manner. Only then can we harness the power of AI to our benefit.
2. Agenda
• Introduction to AI and AI Applications
• Considerations for Developing AI Models
• The Good, Bad and Ugly of AI
• AI Governance
#ISSLearningFest
4. AI Applications
#ISSLearningFest
Retail
Banking & Finance
Manufacturing
Security
Automotive Logistics
F&B
Healthcare
Social and Lifestyle
Small Medium Enterprises
Multinational Corporations
Government Citizens
ConsumersOrganisations
Application
Domains
Example
Users
6. AI Current State
#ISSLearningFest
• Artificial Narrow Intelligence
Perform tasks within a specific domain and for certain use cases.
Often within certain boundaries and constraints.
• Artificial General Intelligence
Able to carry out tasks across several uses cases. Generalize it’s
ability across various domains. Almost human-like intelligence
with ability to perform acceptably in several general tasks.
Is AI able to:
- Perform tasks without supervision and monitoring?
- Think for cases not seen before and reacts accordingly?
- Generalize across various domains and tasks?
- Perform self-maintenance and upgrade?
- Provide the human touch?
https://www.jpost.com/jpost-tech/can-artificial-intelligence-understand-human-humor-635073
What is achievable and not achievable by AI?
7. Considerations for Developing AI Models
#ISSLearningFest
In the Context of:
• Transparency
• Fairness and Bias
• Audit
• Confidentiality
• Ethics
8. Transparency in AI Models
#ISSLearningFest
When would human have trust in AI models? When they
know:
• What is the decision taken by the model at each step ?
• Why did it make the decision?
• Will it make the same decision next time?
• What is the confidence of the decision made?
• How to tweak or correct the decision if it is not correct?
10. Bias in AI
#ISSLearningFest
What is Bias in AI?
Bias in AI happens when the AI model or algorithm discriminates against certain
group. It produces undesirable outcomes and does not provide a fair assessment
and judgement.
Bias in AI may be detected at the output stage. However, it may not always be
apparent unless enough cases were presented and analysis made.
Is bias caused at the output stage or much earlier stages?
It can be due to much earlier stages, e.g., at the data collection stage where
data collected may not be representative of the population and lacks diversity.
11. What could have gone wrong?
#ISSLearningFest
https://www.reuters.com/article/us-newzealand-passport-error/new-
zealand-passport-robot-tells-applicant-of-asian-descent-to-open-
eyes-idUSKBN13W0RL
12. AI Audit
#ISSLearningFest
AI Audit reinforce Trust, Transparency, Accountability and Responsibility
Audit provides an independent and unbiased view on the AI model
developed and provides certain level of assurance to users.
It also serves to address questions such as:
• Is there bias in the model?
• Are there any ethical issues in the modelling(e.g., one that may cause
harm)?
• Is there a structure and process in place for data access?
• Is there governance in place on who can modify the model?
• Does the model perform within a certain level of confidence?
13. AI Confidentiality
#ISSLearningFest
From AI
User
Perspective
From AI
Owner
Perspective
• Informed of data collected and has the option to opt in or opt out
• Usage of data is for specific purpose only
• AI model to benefit the user
• AI model will not harm the user
• Expect certain level of clarifications to be provided at request
• Gain trust from the users of AI
• AI is a valuable asset and hence need to be protected
• Performs AI Audit to the extent that no confidential info is being revealed
• Protects AI algorithm details, to prevent info being leaked to competitors
and to prevent AI algorithms from being exploited and attacked.
14. AI Ethics
#ISSLearningFest
3 Laws of Robotics
First Law: A robot may not injure a human being or, through inaction, allow a
human being to come to harm.
Second Law: A robot must obey the orders given it by human beings except where
such orders would conflict with the First Law.
Third Law: A robot must protect its own existence as long as such protection does
not conflict with the First or Second Laws
By Isaac Asimov
https://www.flickr.com/photos/itupictures/27254369347/
Example Considerations in AI Ethics
• How to prevent misuse of AI?
• Is AI able to take into consideration moral values?
• How much human intervention is required?
15. Agenda
• Introduction to AI and AI Applications
• Considerations for Developing AI Models
• The Good, Bad and Ugly of AI
• AI Governance
#ISSLearningFest
20. Operating Model
#ISSLearningFest
‘An operating model is a visualisation
(i.e. model or collection of models,
maps, tables and charts) that explains
how the organisation operates so as to
deliver value to its customers or
beneficiaries.’
– Andrew Campbell
Marketing Quote Policy
Issuance
Policy
Admin Claims Renewals
In its simplest form, it is a value delivery chain
Structure
Organisational design & reporting structure Committee(s) structure & charters
Oversight responsibilities
Board oversight &
responsibilities
Management accountability &
authority
Committee(s) authorities &
responsibilities
Talent & culture
Performance management &
incentives
Business & operating principles
Leadership development &
talent programmes
Infrastructure
Policies & procedures Reporting & communication Technology
In its complete form, it is a system of practices
21. Elements of the Operating Model
#ISSLearningFest
• Process flow to deliver the value proposition
(the value delivery chain)
• Organisation for the people who will do the
work, the structure of organisation units and
support functions, the decision rights and other
organisation elements
• Locations for the buildings and places where the
work will be done
• Information for the software applications and
databases needed to support the work
• Suppliers for those important suppliers
supporting the work who need special
relationships with the organisation
• Management system for the planning,
budgeting, performance management, and
people management processes needed to run
the organisation. These underpin the other five
elements
22. Where the Operating Model fits
#ISSLearningFest
Source: Julie Choo, The Strategy Journey Framework®
24. AI GRC
#ISSLearningFest
“New products and services, including those that incorporate or utilize artificial
intelligence and machine learning, can raise new or exacerbate existing ethical,
technological, legal, and other challenges, which may negatively affect our brands
and demand for our products and services and adversely affect our revenues and
operating results.”
– Alphabet, 2018
“AI algorithms may be flawed. Datasets may be insufficient or contain biased
information. Inappropriate or controversial data practices by Microsoft or others
could impair the acceptance of AI solutions. These deficiencies could undermine the
decisions, predictions, or analysis AI applications produce, subjecting us to
competitive harm, legal liability, and brand or reputational harm.”
– Microsoft, 2018
AI’s complexity and growing concerns
about its potentially harmful effects have
inspired calls for AI Oversight.
25. AI Oversight
#ISSLearningFest
“New products and services, including those that incorporate or utilize artificial intelligence and
machine learning, can raise new or exacerbate existing ethical, technological, legal, and other
challenges, which may negatively affect our brands and demand for our products and services and
adversely affect our revenues and operating results.” – Alphabet, 2018
“AI algorithms may be flawed. Datasets may be insufficient or contain biased information.
Inappropriate or controversial data practices by Microsoft or others could impair the acceptance
of AI solutions. These deficiencies could undermine the decisions, predictions, or analysis AI
applications produce, subjecting us to competitive harm, legal liability, and brand or reputational
harm.” – Microsoft, 2018
AI’s complexity and growing concerns about its potentially
harmful effects have inspired calls for AI Oversight.
AI GRC
AI &
Cybersecurity
AI
Governance
Legal
Governance
Data Governance