You are viewing the chat in desktop mode. Click here to switch to mobile view.
X
Emerging Technologies - Session 3
powered byJotCast
Prev Page View Full Chat
Megha Anilkumar Nair
3:16
Nandan Sudarsanam starts with an introduction to Artificial Intelligence and Machine Learning.
3:17
AI is an umbrella term. It is all about making machines behave intelligently, as per him.
3:19
Machine learning is a subset and application of artificial intelligence. One way of behaving intelligently is by learning from data. Artificial intelligence learns from past instances. Machine learning is all about algorithms that allow artificial intelligence to learn data.
3:25
After a brief description of the concept of Artificial intelligence and machine learning, Dr. Sudarsanam goes on to explain the history of Artificial Intelligence. It began in the mid 1950s and started proliferating into various industries.
The prerequisite for AI is primarily machines, and the next would be improved algorithms, as per Dr. Sudarsanam.
He considers the move to digital platforms to be the most important reason for the boom of Artificial Intelligence.
3:31
3:32
Citing problems with traditional learning methods, he explains machine learning as a method of supervised learning. With examples like detecting spam and filtering house prices, he illustrates the working of artificial intelligence.
3:33
Relating the concept to journalism, Prof. Sudarsanam goes on to talk about natural language processing and generative models using the same examples he cited before.
3:36
When talking about supervised learning, unsupervised learning must also be explored. "How do you get the labels of spam or not spam in the first place?" asks Dr. Sudarsanam. The concept of unsupervised learning is about the probability of something happening without an authoritative entity.
3:38
The next concept is reinforcement learning. It is all about "learning from doing". It is an interactive way of gathering data while learning in a supervised environment, he explains.
3:43
On the role of humans in AI, he says, "Humans play an important role in AI." A big part of the working of the technology is the "human in the loop".
3:44
He poses important questions about the future of the technology. They are:
"Can AI be made more safe?"
"Can it be made such that if a human is missing something, can it compensate for that?"
3:47
What is your view on security of AI?
It should be increased (66.7% | 2 votes)
 
It is fine as it is (33.3% | 1 vote)
 
Total Votes: 3
Talking about the AI revolution impacting the job sector, he accepts that its effect on any commodity is bound to show. Any automation will take away jobs.
Prof. Sudarsanam opines that any automation has created more jobs than it has taken away.
3:48
Answering a question on AI taking over human existence, he says that the whole concept of AI and consciousness is still at a nascent stage to image. "Consciousness is still an emerging property," as per him.
3:49
Dr. Sudarsanam: "It is like preventing overpopulation in Mars without getting there."
3:53
On the topic of ethical dilemmas in AI, he cites the moving train example to say that these are philosophical questions. "From a utilitarian perspective, it is safer and better. It is following safer rules."
3:57
Prof. Sudarsanam draws on the environmental impact of AI to say that the carbon footprints of 'data farms' is huge. The economic benefit from this technology may not arrive at sustainability. This is a consequence of capitalism.
4:05
4:06
Answering a question about male bias in AI, he attributes the web component of AI to be a challenge for bias in algorithms. He also mentions that research on lessening and preventing bias in artificial intelligence is an effort in progress.
On the question of morality in artificial intelligence, Dr. Sudarsanam says that AI should not perpetuate a bias. "It hard to say it can get rid of bias because it is present both in the data and in the human component."
4:10
The moderator for the session, Dr. Jai Asundi poses a question on how AI can be used in India for the betterment of the society. Prof. Sudarsanam cites classroom education as an example. Taking tests can be governed by AI.
4:13
4:14
That was a wrap for session three of the colloquium. Thanks for being with us. Stay tuned for session 4.
Connecting…