Explore the successes, limitations, myths, and potential threats of artificial intelligence in this comprehensive lecture by Marc Mézard, Director of Ecole normale supérieure - PSL University. Delve into the recent breakthroughs in machine learning and deep networks, examining their applications in image interpretation, speech recognition, and game-playing. Gain insights into how deep networks are programmed to learn from data, understand their current limitations, and critically evaluate whether their achievements truly constitute "intelligence." Reflect on the foundations of scientific intelligence and consider the potential societal impacts of AI, including job displacement and autonomous decision-making systems. Engage with thought-provoking discussions on the future of AI, its role in scientific inquiry, and the ethical concerns surrounding its widespread adoption.
Artificial Intelligence: Success, Limits, Myths and Threats - Lecture 1
International Centre for Theoretical Sciences via YouTube
Overview
Syllabus
DATE: 06 January 2020, 16:00 to
Lecture 1 Public Lecture: 6 January 2020, PM
Lecture 2: Tuesday 7th January 2020, PM
Lecture 3: Wednesday 8th January 2020, PM
Artificial intelligence: success, limits, myths and threats Lecture 1
ICTS
ICTS Campus in Bangalore
What is the Goal of the ICTS?
Enabled by 3 interactive missions:
During the past decade ICTS has achieved some measure of success in all its three missions!
Programs:
Sample Programs...
Programs in Machine Learning
ICTS-Infosys Foundation Lecture series:
Research
ICTS People: Faculty
Students and Postdocs
ICTS as a platform for new initiatives:
Science Outreach
Public Lectures
Kaapi with Kuriosity
Abdus Salam Memorial Lectures
Einstein Lectures
Vishveshwara Lectures
D.D. Kosambi Lectures
Mathematics of Planet Earth MPE 2013, Bengaluru
Bangalore Area Science Habba
Mathematics Circles
ICTS Organization
ICTS Resources
Thank You!
Artificial Intelligence: Success, Limits, Myths and Threats
Chapter One - Myths and Reality
The new era of AI
ImageNet Database and Challenge
Convoy of self-driving trucks completes first European cross-border trip
The new era of AI
2 - Language understanding
AlphaGo
July 2019 : Pluribus
Chapter Two - Machine learning
Machine Learning
Test phase=present new picture, that the machine has not yet seen
Chapter Three - The Machines: Artificial neural networks
Everyone recognizes
Artificial neural networks
Frank Rosenblatt's perception
What is new since Rosenblatt's perceptron?
Neural network reading digits
Performance on handwritten digits
Deep neural networks
Bigger networks, more parameters. Larger database!
New computing paradigm. Collective representation of information, going to larger scales. Robust.
Chapter Four - Why deep networks are not yet? a panacea
Three main problems:
1- Huge amount of labelled data is necessary for learning in deep networks
Oh, look at ko bamoule! Do you see ko bamoule?
Chapter Five - About scientific Intelligence
Quote from Chris Anderson -The end of Theory: The data deluge makes the scientific method obsolete
Thought experiment :
We are still very very far from General Artificial Intelligence
Conclusion - So, what is going to happen?
Predicting the future
Predicting the future ?
A major concern for the present:
In 2018:
Take-home message
The End
Q&A
Taught by
International Centre for Theoretical Sciences
