2. Research focus areas
Ethics of AI
Safety of AI
Robustness of AI
Mitigating
Antisocial Online
Behavior
Big Data Cloud/Edge
Computing
Wireless
Networks
ICTD
Artificial Intelligence/
Machine Learning
Ethics and
Technology
Learning/
Pedagogy
Cognitive
Networks
Robotics
Adversarial ML IoT/ Smart Cities Algorithms
Engineering
Education
Priority research areas
Secure, Robust,
Ethical AI and ML
Using ML/AI for
social good
ICTD
…
Research Funding
FB Research Ethics in AI Research Initiative
for the Asia Pacific Award Winner (2020)
Culturally informed pro-social AI regulation and persuasion framework
Dr. Junaid Qadir, Dr. Amana Raquib
3. Workshop Road Map
1 What is AI? Types of AI? History of AI
3 Big Ideas and Important Concerns Related to AI/ML
2 Machine Learning (ML)
6. What is Artificial Intelligence?
Artificial intelligence is a field renowned for
its lack of consensus on fundamental issues
7. A melting pot of ideas, techniques, and insights that
relate to artificial (computational) intelligence
Cognitive Hexagon
8. Different takes on “what is intelligence”
The ability to solve problems
The ability to learn from experience
The ability to reason about things
The ability to recognize patterns
The ability to figure out causes of things (or to understand)
The ability to pursue objectives and purpose.
The ability to generalize and abstract (and make analogies)
9. AI focuses on building “intelligent machines”
Intelligent agent is a system that
perceives its environment and
takes actions which maximize its
chances of success. (Russell & Norvig)
11. “An attempt will be made to find how to
make machines use language, form
abstractions and concepts, solve kinds of
problems now reserved for humans, and
improve themselves. We think that a
significant advance can be made if we
work on it together for a summer.”
John McCarthy and Claude Shannon
1956 Dartmouth Workshop
AI’s official birth
13. History of AI—Knowledge Based Approaches
Computer
Data
Program
Output
Traditional Programming
1988—93: Expert systems industry busts
1970—90: Knowledge-based approaches
1969—79: Early development
1980—88: Expert systems industry booms
14. Problem space grew too quickly
Complexity of the world made it hard to encode all rules
AI Winter and Underwhelming Results
15. 1990— 2012: Statistical approaches + subfield expertise
• Resurgence of probability, focus on uncertainty
• General increase in technical depth
• Agents and learning systems…
“AI Spring”?
2012 onwards: Lots of Excitement
• Big data, machine learning, deep learning
• AI used in many industries
History of AI—Data Driven Approaches
16. What is behind modern success of AI?
Modern AI and Data-Driven Methods
Model Uncertainty
Use data to learn
from experience
1997
Deep Blue
2005
Stanley
2011 Watson
19. Difficulties in teaching concepts to computers
In Arthur Samuel’s classic 1962 essay "Artificial Intelligence:
A Frontier of Automation", he wrote: “Programming a
computer for such computations is, at best, a difficult task, not
primarily because of any inherent complexity in the computer
itself but, rather, because of the need to spell out every minute
step of the process in the most exasperating detail.
Computers, as any programmer will tell you, are giant
morons, not giant brains.”
Traditional Programming
Problem space grew too quickly
Complexity of the world made it hard to encode all rules
AI Winter and Underwhelming Results
20. Let’s try Machine Learning
Learning
Algorithm
Output
Input Program/
Model
Machine Learning
Essence of Machine Learning:
– There is a pattern in data, however it is complex and difficult to articulate
– We cannot pin it down mathematically (i.e., it is too complex for that)
– Let the computer learn a model for the concept itself using algorithms.
21. “Learning” in Machine Learning
Arthur Samuel
How do we learn the right weights (model parameters)?
Use of mathematical
optimization to reduce
the discrepancy between
predictions and actual
labels (which is computed
by loss function)
23. Supervised learning: the machine experiences a series of inputs: x1, x2,
x3, x4, … along with the correct labels y1, y2, … and it aims to learn a
mapping so that it can make a correct prediction for a new input
Supervised learning
Fraud Detection
Toxic Comment Detection
Binary Class Classification Multi-Class Classification
{0, 1, 2, … 9}
24. Deep Learning and Deep (Neural) Networks
M. Mitchell Waldrop PNAS 2019;116:4:1074-1077
26. Unsupervised learning
Unsupervised learning: the goal of the machine is to build a
model of x that can be used for various tasks such as reasoning,
decision making, predicting things, communicating, etc.
Applications: Clustering, anomaly detection, etc.
29. Reinforcement Learning
Basic idea:
Receive feedback in the form of rewards
Agent’s utility is defined by the reward function
Must (learn to) act so as to maximize expected rewards
All learning is based on observed samples of outcomes!
36. Despite the AI optimism, ML is no panacea
Many people find the lack of rigor and sound
analytical understanding of the internal working
of ML troubling
Despite the AI optimism, ML is no panacea
37. Bias (Can the ML model learn our biases?)
Interpretability (Can we understand ML models’ “reasoning”?)
Privacy (Is the ML model protecting our privacy?)
Security (Is the ML model secure against adversarial attacks?)
Open questions related to ML/AI
Trustworthiness/Reliability
(Can I trust the prediction of an ML model?)
38. Overfitting and the importance of generalization
Clever Hans (pictured) “solved”
arithmetic problems by simply
following the cues that inadvertently
emanated from his trainer’s body
language.
42. “Every new technology will bite back.
The more powerful its gifts, the more
powerfully it can be abused.”
Kevin Kelly
Looking at technology critically
Ledger of Harms
44. Developing Safe & Trustworthy AI/ML systems
THERE ARE ETHICAL
CHOICES IN EVERY SINGLE
ALGORITHM WE BUILD
“
45. Ability To Act Ethically and With Wisdom
Wisdom
Understanding
Knowledge
Information
Data
An ounce of information is worth a pound of data.
An ounce of knowledge is worth a pound of information.
An ounce of understanding is worth a pound of knowledge.
An ounce of wisdom is worth a pound of understanding.
Good for your self/tribe is
not necessarily good for
common good of humanity