Decoding the Book "Can We Trust AI?" and How Teachers and Doctors Can Leverage the Technology for Better Outcomes with Dr. Rama Chellappa

On this episode of AI: Decodes the System, Dr. Rama Chellappa, author of Can We Trust AI?, and AI decode how Dr. Chellappa got involved in artificial intelligence (AI) research and why he decided to write a book on if we can trust AI? During the conversation, Dr. Chellappa breaks down AI and how, contrary to popular belief, the technology has been around for over 60 years and is ready to start receiving social security. Dr. Chellappa talks about how we work with AI today, will work with the technology in the future, and why we all need to understand how to work with AI just like we interact with the internet and cell phones. He also discussed the importance of responsible AI and training algorithms to be less biased. One of my favorite insights from the conversation is Dr. Chellappa's moonshot idea that classrooms should have one physical teacher and thirty virtual teachers. Can we trust AI? Check out the episode.

More about Dr. Rama Chellappa
Rama Chellappa, PhD (Potomac, MD), is a pioneering researcher and inventor in the fields of computer vision, artificial intelligence and machine learning. A Bloomberg Distinguished Professor in electrical, computer, and biomedical engineering, he is also the chief scientist at the Johns Hopkins Institute for Assured Autonomy and a member of the Malone Center for Engineering in Healthcare. A Fellow of both the Institute of Electrical and Electronics Engineers and the National Academy of Inventors, Chellappa holds eight patents. His work has been featured by the Associated Press, BBC, and The New Yorker. 

Follow Dr. Chellappa at the links below.

Dr. Rama Chellappa: Aritificial Intelligence for Engineering and Medical Lab website, LinkedIn, and Purchase Can We Trust AI?
AI: Decodes the System: Website, Instagram, Twitter, Facebook, and TikTok

Transcript

00:01.53
aidecodes
Dr. Rama shalapa or Rama um, for short is a pioneering researcher and inventor in the fields of computer vision artificial intelligence and machine learning a bloomberg distinguished professor in electrical computer and biomedical engineering. He is also affiliated with the Johns Hopkins Institute for assured autonomy the center for imaging excuse me and the center for imaging science the mathematical institute for data science and the center for language and speech processing. A fellow of 8 professional societies including the institute of like electrical and electronics engineers let me say that again. Um professor professional societies including the institute of electrical and electronics engineers the association for the advancement of artificial intelligence. And the association for computing machinery and the national academy of inventors shalapa holds 8 patents. He works his works have been featured by the associated press Bbc and national group geographic hello Dr Shalapa welcome to the show.

01:01.43
Rama Chellappa
Thank you Miss Ivy and thanks for having me on your show. Not at all that is that is nice and that is simple. Yes, please.

01:05.53
aidecodes
No problem. Do you mind if I call you Rama just for our conversation today. Okay, most definitely so Rama you've been studying artificial intelligence for a very long time. I read your story personally but can you tell the audience. What made you get into this industry.

01:25.12
Rama Chellappa
Yeah, in general I was interested in technology I think the way it worked in those days in India is if you're good at mad. They immediately say you are going to be in engineering I mean you really had.

01:36.75
aidecodes
Oh.

01:40.42
Rama Chellappa
Almost like you had no choice I mean you had choice but it was understood that if you are reasonably good at math. You become an engineer and so I I that's how it turned out but i. I was very interested about the moon landing thing as I say in my book in one sixty Nine I heard the landing live you know on the radio right.

01:54.26
aidecodes
We had.

02:01.98
Rama Chellappa
And a few weeks later my cousin who was lecturer at a local engineering college got 2 tickets to go to the us consular office in Chenai madraz India to watch a movie about it. So I watched the movie and I was thrilled that there is technology that would take a person.

02:06.60
aidecodes
A.

02:20.84
aidecodes
A.

02:21.82
Rama Chellappa
The moon and then bring the person back so I was hooked into engineering and technology. So I continued my studies in engineering and then I came to purdue and then I specialized in ah in ah 1 form of discipline which is. Ah, Subfield Of Artificial Intelligence so it is known as computer vision pattern recognition and so on so this is something I have been now working for almost forty two years

02:39.79
aidecodes
Yes, yes.

02:47.22
aidecodes
Oh wow, that's amazing I love that you started like as a child like thinking about these things and really looking at the future or what would eventually be your amazing career. Um, as you know this show is called Aideco the system because we break down different topics related to technology data policy and law. Um, and today we're really going to delve into technology and 1 technology in particular which is artificial intelligence. Why did you decide to write the book. Can we trust Ai. Like why right now and what what is this book about.

03:21.91
Rama Chellappa
Yeah, you know yai has been there for almost sixty five years there was a summer workshop in 1956 where some of the best minds in the country got together and discussed you know about this field.

03:27.29
aidecodes
And.

03:36.80
aidecodes
Is.

03:39.60
Rama Chellappa
So I like to say if Ai were a human being. It is ready to collect social security check and and and be on Medicare right? So Ai has had assumption down. It has been there for a while since 10 it has become a very active field again because now Ai knows how to mine data.

03:47.94
aidecodes
Yep.

03:52.47
aidecodes
So.

03:57.74
Rama Chellappa
Because we have powerful computers and we have a lot of data and so along with it. You know there are also concerns about Ai is is the Ai stealing. Our data is is it intruding into our privacy or does it show any bias towards 1 subgroup of population and so on.

04:07.13
aidecodes
And.

04:17.16
Rama Chellappa
And does it work all the time everywhere. So All of these concerns are there. So I thought people write a balanced book account of Ai pointing out the good things it has done. It is doing and it will. And at the same time acknowledge some of these concerns and point out technology always gets improved always gets refined. We are not driving model T anymore we are driving Tesla So. There's been a lot of improvement since Model t.

04:40.67
aidecodes
Right.

04:46.62
Rama Chellappa
So the same thing will happen with Ai with time. It will get better. So all of these concerns are real and they are being ah worked on and so I thought we should provide because people get the impression about Ai by looking at movies like you know I robot and the ter of terminate and all that thing. So.

05:02.20
aidecodes
Yep.

05:05.83
Rama Chellappa
And of course I was very much helped by my co-author Eric Neer and the editorial staff at Johns Hopkins University Press you know a book like this is a team effort because it has to be written in a nice way. So it is accessible to a lot of people you know.

05:12.17
aidecodes
Month.

05:23.45
Rama Chellappa
So so I was very much helped by my co-author and all the editors. So that's why we decided to write because trustworthy ai is very important people have to trust this technology that it will do good.

05:30.23
aidecodes
Okay.

05:36.90
aidecodes
Very true because everyone we're gonna get into this in a few minutes but I know right now like and like you said over the last decade or so people have like normal people right everyday people like me who I don't research this topic right? We've been. Getting inundated with Ai and hearing more about it. But you're right? It's been there for a very long time and in a lot of different applications. So I'm glad you talked about that I'm looking forward to talking a little bit more about that as we have our discussion. Um, but before we go there as an expert in this industry.

05:56.77
Rama Chellappa
Yes, sure. Yeah.

06:07.96
aidecodes
What do you believe are the benefits of Ai in the world today and how will it improve our lives.

06:11.56
Rama Chellappa
Yeah, one can look at many different sectors where Ai is being applied now from being a simple recommender system. You know you order a book in Amazon immediately. You get a response saying oh people who got this book also looked at these other books.

06:16.99
aidecodes
A.

06:30.63
Rama Chellappa
Or you see a Netflix movie and then immediately you say hey people who saw this movie also like you know and so on in fact, Netflix even will tell you how confident it is that it is matching your interest right? 97% you know it's matches and so forth right.

06:39.71
aidecodes
Right.

06:46.94
Rama Chellappa
So these are interesting applications but the application that we think will revolutionize our our in Medicine Um Andai can mine the continuous data that is collected. You know you know we we go and see a doctor and then we take a but lab work and there is a report.

06:51.50
aidecodes
A.

07:02.52
aidecodes
Right is.

07:05.15
Rama Chellappa
And there is you know sometimes you know x-rays and ct and mra and so on and even natural language conversations between the doctor and the patient all of them can be securedly kept someplace and Ai can work with them and kind of look for something you know.

07:10.24
aidecodes
A.

07:20.81
Rama Chellappa
Humans we can remember ah data a little bit. In fact, if you see a doctor the same doctor over a period of time when you go they will immediately take your file and kind of see where you were last time and they will ask you about that and so forth. They're not going to remember what happened eight years ago but a I can.

07:22.40
aidecodes
Right.

07:37.70
aidecodes
Um.

07:39.80
Rama Chellappa
Right? Ai can look for long-term things and what trends are and so forth and it will give some hints to the doctor. It will be like the Physician assistant it will be like a software assistant. Yeah right? So it's not going to send you to or and know it's not going to do that. The doctor is there.

07:52.38
aidecodes
Ah.

07:56.51
aidecodes
Right.

07:58.60
Rama Chellappa
Because you know that's important, but it will give the doctor some possibilities and so forth and and you know it will say maybe this person should take this booster now because it's been 10 years tetanus booster is needed. The doctor may not remember you may not remember.

08:12.77
aidecodes
Yeah.

08:15.48
Rama Chellappa
Um, and simple things like that and but then even more sophisticated things that it can look for and of course in the self-driving cars ai is present because this collects a lot of data from Cameras and Then. Ah, comes up with the ways to navigate traffic. You know and understands the rules and and all of that. So domain knowledge is being used and and so on so these are basically if a car has these sensors and then if it can alert you.

08:35.00
aidecodes
Who.

08:48.29
Rama Chellappa
About the cars and adjacent lanes then you'll be careful when you switch lanes. In fact, there's one car I test drove years ago if I tried to change lanes without giving the turn signal. It will bring me back to my lane I was kind of having fun with it on the freeway you know I'll go take it there and it'll bring me back and take it there bring me back.

08:50.60
aidecodes
Right? right.

08:59.26
aidecodes
A.

09:07.54
Rama Chellappa
And and the the person who was with me the the car salesperson said you know why don't you just go and you know axiliate and go near the car in front of you your car will automatically stop I said I don't want to take that risk yet I don't want it I have not tested that yet, but it does give me an alert my car.

09:17.93
aidecodes
Um I don't want to test that.

09:24.88
aidecodes
Um, right.

09:26.88
Rama Chellappa
It gives me an alert that I may hit so these are things that will reduce accidents these are things that will reduce the number of you know 30000 to 40000 people die on our highways accidents various reasons and so on so the safety of our livelihood you know is is going to be better and and health of course.

09:32.66
aidecodes
Yeah.

09:45.22
Rama Chellappa
It can also be used for an education. You know there are children with the different skills and you know, abilities and learning and ai can be a good tutor a personalized tutor. So what we think may happen eventually is personalized medicine. You are no longer an average.

09:48.10
aidecodes
Ah.

09:54.11
aidecodes
And. Oh Wow! yes.

10:02.40
Rama Chellappa
You are you know Miss Amber iv right so the Ai knows you they know they so next to time if you order oyster somewhere. It'll say thing you know, did you remember what happened the last time.

10:13.62
aidecodes
Right? last time you did this.

10:17.24
Rama Chellappa
Yeah, it can potentially hopefully I mean that's what so it'll be there as a friend and of course you know, even our best friend sometimes can be troublesome. We all have best friends like that. So Yeahi has to be watched like all technologies we have to refine it. We have to make it better.

10:20.40
aidecodes
Right. A.

10:35.76
Rama Chellappa
Eventually it will it will do good things. It's a good thing to have it.

10:39.92
aidecodes
I'm glad you brought up just the different ways that we're gonna work with Ai and I love that you say it's not an Ai versus human type world right? It's ai with humans and working together and I love the fact that you talk about how even the medical feel. Um, like.

10:49.32
Rama Chellappa
Yes, yes, yes.

10:56.16
aidecodes
Me personally I live with multiple sclerosis but I had to figure that out because I kept like asking my doctor and trying to like show indicators. But imagine if Ai could have read my um scans of everything that I had had seen my levels and how things were changing seeing seeing the different effects. And what I was dealing with and then ah looking at a system that had other like um like issues or symptoms and being be able to say to my doctor hey you should probably do these tests that would be amazing like the idea of personalized medicine. The idea of personalized tutoring. Um, it just makes me really excited. But I love the idea like you said of.

11:20.80
Rama Chellappa
Yeah, yeah, yeah, yeah, yes, yeah.

11:31.38
aidecodes
Ai and humans working together because I do believe that is the future. Um, but some people may be a little bit more skeptical or may not even know what the heck we're talking about like we say these words like Ai artificial artificial intelligence and all these other buzzwords seemingly buzzwords and technology. But.

11:33.21
Rama Chellappa
Yeah, yeah.

11:42.21
Rama Chellappa
Yeah, yeah, right? Yeah yeah, yeah, yeah.

11:48.49
aidecodes
Have its 2 year old and a 7 or 8 year old now. Niece um, how would you explain Ai to them and explain some of the simple ways ai shows up in their life. Yeah.

11:56.68
Rama Chellappa
Yeah I would start with the eight year old I will show them a tesla car and I will show them and and take them to a dealer and I'll show them and then I will take it for a test drive and you know I don't.

12:09.24
aidecodes
Um.

12:15.52
Rama Chellappa
Many kids know this there used to be a show called night rider talking car talking cars long time ago right? So flight flying cars right? and all of that. So every all their kids are excited by that. So.

12:20.94
aidecodes
Yep f.

12:28.37
Rama Chellappa
You can explain to the 8 year old how this car can potentially drive without the driver doing anything and the cameras are there and they collect information and they process them and they tell them where the other cars are so it can you know.

12:32.43
aidecodes
Oh.

12:43.53
aidecodes
Right.

12:44.40
Rama Chellappa
Ah Manu and know maneuver around those cars and then and it will know when to ah stop and go on so forth. So that is Ai living inside a car that is a good example for a 2 year old I was thinking little hard about that when I saw the question this is I don't know we can try it and see what you think.

12:54.91
aidecodes
Um.

12:57.30
aidecodes
Okay, yeah.

13:03.55
Rama Chellappa
There is a new thing called chat gbd you know? Yeah, there is everybody's excited so I will give a a nickname to that. So what is the gender for the 2 year old kid. She's a girl so I will give some friendly. You know the girl name I say sit down and talk to it.

13:14.30
aidecodes
She's a girl.

13:19.80
aidecodes
Nice I.

13:22.21
Rama Chellappa
And then she she will be amiss. He said wait a minute. It's not like a radio which just goes on I'm asking something what day is today and you know it's so this is ah you know Thursday oh okay, what do you like to do and so on and then she say wait. How is that this is happening like a computer is talkinging.

13:38.35
aidecodes
Yeah.

13:41.76
Rama Chellappa
Then you can explain to them that this is natural language processing somebody listened to a lot of such conversation and put together a system so you can almost imagine there is a person sitting inside the computer and talking to you right? It's not a can the thing it's not like you're putting a Cd and it.

13:43.17
aidecodes
In here.

13:52.75
aidecodes
All right.

13:58.81
aidecodes
A.

14:00.65
Rama Chellappa
Does the same thing every time she talks to it will be different. The questions will be different and ah so after a while let's say you you give ah some you know, nickname you know, call call it I don't know um Cindy or whatever you call it so you want to talk to Cindy or to it will say. yeah yeah I want to please switch on the laptop.

14:12.58
aidecodes
Okay.

14:18.10
aidecodes
Oh yeah.

14:20.70
Rama Chellappa
And you know then you can slowly get her to understand. There's no person sitting inside the laptop but it is what we call as artificially intelligent system because it talks like an intelligent system right lot of you. You may have already seen.

14:23.16
aidecodes
Right.

14:32.34
aidecodes
Okay.

14:34.67
Rama Chellappa
This thing has gone wide a lot of people are saying oh it just this and say things right? that right? and everybody is excited. In fact, when you try to get on to it. It says I'm sorry there's just too many people talking to me, you know, sometimes yeah, right? yeah.

14:42.76
aidecodes
Yeah, it shuts down I was using it yesterday and it shut down multiple times when I was playing with it I think so too because she's already like she's familiar with at least like Alexa and things like that. So she asked a less.

14:50.10
Rama Chellappa
So I think that may be appropriate for a 2 year war I don't know this is an experiment.

14:58.52
Rama Chellappa
Yeah, yeah, yeah, yeah, yeah, right? Yeah yeah, yeah.

14:59.34
aidecodes
Alexa questions at the age of 2 which is wild to me and she's always like typing on a computer thinking. She's doing something. Um, she knows her alphabet already like she can count. She's like really advanced for her age. Um, so I think she would definitely understand that I'm with that.

15:13.30
Rama Chellappa
Yeah I think she's going to be a super ai researchers 2020 years from now. Ah.

15:19.52
aidecodes
So for what are some other ways other than like the teslas of the world. Um, or even some of the stuff happening with chat gbt which is now like in the news all over the places where everyone has access to it through openai. What are some of the other things that. We use in our day-to-day lives that most people don't even realize are Ai.

15:38.48
Rama Chellappa
Um, I think the recommended systems I think people may not realize that because every time you go to Amazon you know you get that so there is a version of Ai doing that and then you will notice with time it changes what it recommends.

15:40.88
aidecodes
Um, yeah.

15:51.69
aidecodes
Um, yeah.

15:55.60
Rama Chellappa
If you continue to see certain types of movies. Netflix will just bring them up all the time because it just says no you just what you like and you say wait a minute. No I like other things too right? So you know so that is the thing that they can relate to and I think you know right now Ai is present in.

16:00.79
aidecodes
Um, right.

16:12.33
Rama Chellappa
Um, and self-driving cars is where it's dominantly present medical it is slowly there. You know at least 120 or so medical procedures based on Ai ah apparently have been approved by Fda so they will hear it more and more and I think.

16:14.31
aidecodes
A.

16:27.57
aidecodes
Um, oh Wow. Okay.

16:30.14
Rama Chellappa
Ah, more and more what will happen is a doctor may tell them you know this is you know my opinion based on my training I also have a confirming opinion from this Ai you know sitting next to me and that I think is going to be an interesting dynamic in medicine. They'll be a triplet a patient.

16:42.70
aidecodes
And. Right.

16:49.93
Rama Chellappa
Ai and a doctor right now you know before Ai came we trusted. Whatever the doctor said right? The doctor says you know based on the report this what I recommend because the doctor has been trained well you know in this country they go through various very very serious training. My 2 children are doctors. So I know.

16:59.56
aidecodes
Right is that.

17:06.17
aidecodes
Out. Nice.

17:08.83
Rama Chellappa
How good you know they're they're trained and so we trust them and now there is another thing sitting there ai and the patient may ask well doctor I know you went to a good medical school. But what about this little box here is telling you something and what about it then so we need to have the trust.

17:11.33
aidecodes
Social and like.

17:25.55
aidecodes
For that size and thats what delete.

17:27.40
Rama Chellappa
That the machine knows what it is doing and it's confident enough and and all that thing. So I think they will see more and more of that you know it's it's still is slowly happening and I think that's going to happen. But right now if you ask me where Ai is present I will say natural language models.

17:36.76
aidecodes
Students still be happy, but.

17:43.58
aidecodes
Connected models. Yeah that England So that.

17:46.16
Rama Chellappa
Chat gp is a great example and there are many versions of that I'll say siri is an example there is a version of Ai sitting there Alexa version of Ai is sitting there. For example, 1 of my friends said that he was casually discussing buying mattresses and then.

17:53.83
aidecodes
Yeah, like Edward one of good.

18:01.36
Rama Chellappa
And he logs into the computer all these commercials from matttrass companies come right? So that that is a because somebody is ah figuring something about you right? So I think that's where it is right now, but it's getting more and more visible. Yeah.

18:02.84
aidecodes
Exactly.

18:09.73
aidecodes
Nothing you right? So that.

18:18.82
aidecodes
Yeah, because I know um I was talking to someone and they were like I don't understand why I say something in the ad comes up I'm like is slitchling is literally processing what you're saying to get you that ad um and other behaviors as well. So it's pretty cool that it's already here. We're already working with it and it's also.

18:27.77
Rama Chellappa
Yes.

18:36.18
aidecodes
Cool about the future of Ai and what it's going to look like in the next ten years I'm really excited about that I know in reading your book. Um, you discussed some of the more controversial applications like self-driving cars and we already started to hint out hint at that a little bit.

18:38.58
Rama Chellappa
Right? Yes, yes. Yeah, yeah, yeah.

18:50.96
aidecodes
Um, I know a lot of people are very afraid of Tesla's ability to um, let a car go full ah full self drive and they've seen the videos where people are sleeping cars dogs are driving cars and all these things are happening.

19:04.56
Rama Chellappa
It's yeah yeah, yeah, um I think you know in every technology we have what we are as real Pi in the sky idea.

19:06.30
aidecodes
What are your thoughts on letting Ai fully take over our roadways.

19:17.19
aidecodes
To by and for the occupation.

19:18.76
Rama Chellappa
I'm not calling fully automation as spy in this guy but it is a worthy goal for technologies to think about because we have to imagine so if we don't build things for tomorrow or next week we build things for 10 years out right? So fully automated car. You know is very exciting.

19:21.90
aidecodes
What people? yeah because we have.

19:29.83
aidecodes
Right.

19:36.83
Rama Chellappa
You know people say oh I can just you know make a sandwich and you know I can have a little kitchen and my I personally would like to have a restruo machine and make coffee. So just put it there and drink can read a book and kind of okay as a hopkins come maybe I can get down but my my view is.

19:38.00
aidecodes
Right. Same.

19:49.57
aidecodes
Exactly.

19:54.60
Rama Chellappa
Actually we computer vision people worked on this problem since the mid 80 s mid 80 s in fact, my men mentor suggested project known as autonomous land vehicle more for a military in those days because these things were bulky and you know and the sensors were bulky and and so on. So um.

19:56.31
aidecodes
So okay, just. 6.

20:08.53
aidecodes
Right? so.

20:14.42
Rama Chellappa
The way I see it is that to get to the fully automation you still have to have a lot of other component technologies done as an example alerts about cars and adjacent lanes alert that you're going. You may hit the car in front of you or when you reverse you know what's out there in the back and so on. Okay.

20:20.95
aidecodes
Right.

20:33.92
Rama Chellappa
Are extremely useful to reduce accident to reduce the number of death so full automation is a great technology goal but for it to happen this this is what I was thinking the other day. Let's assume there are 10 people who are willing to go full automation.

20:49.13
aidecodes
Okay.

20:52.90
Rama Chellappa
And then there are 10 people like me who say no I like all these safety features but I still want to drive. Okay, then what happens is you have a mix of things on the road. The full automation ah car automated cars. You can think of as ai because it's fully automated.

20:55.30
aidecodes
Right? they. The mix that's.

21:07.81
aidecodes
Ah.

21:09.14
Rama Chellappa
It it does everything on its own. It has rules it and and so on but the ones that I drive you can consider them as humans because I am driving it although there are safety features immediately. This brings into discussion how humans and Ai can work together now imagine.

21:15.40
aidecodes
Right.

21:23.56
aidecodes
Right? okay.

21:27.50
Rama Chellappa
I'm going on a freeway sixty five miles an hour there's not much a space for making mistakes. So unless everything is automated. There's going to be some kind of confusion when you have some automated cars and some semiautomated cars right.

21:40.72
aidecodes
Right.

21:45.20
Rama Chellappa
So it's not clear how it's going to shape up but there are some domains where full automation is is being thought of. For example, they want to have a fully automated 18 wheeler carrying stuff on the freeways because they say freeway driving is more organized you stay on the lane.

21:56.18
aidecodes
Um.

21:58.57
aidecodes
Right.

22:01.63
Rama Chellappa
And you don't change unless whereas you know if you go into the city. It's a little different. The traffic is a little bit more complicated so there are some applications like that where full automation is warranted in fact Dr Tom Strad a friend of mine who was in darpa and managed the darpa grant challenge. He says in the book.

22:05.74
aidecodes
Right? If not.

22:14.99
aidecodes
A.

22:20.34
Rama Chellappa
The way to really test my confidence in the automated car is how long can I close my eyes and drive. He says few seconds and he is a technologist of of you know extreme capability I mean he he lives and breeds technology in computer vision technology in particular.

22:25.28
aidecodes
Are right.

22:36.42
aidecodes
In a.

22:40.21
Rama Chellappa
So that's what he says right? So it all depends on how much how comfortable we are not everybody is is going to be comfortable as he says you know he can only close his eyes for a few seconds. So what? what? I think you know is is a fully automated car should we give up I.

22:46.90
aidecodes
Esther.

22:50.61
aidecodes
E.

22:59.10
Rama Chellappa
Don't think so there are people with different preferences for driving some people like to drive and even if you want to completely let the car drive I think it's individuals preference. That's what I I'm going to say there may be some applications where a convoy are fully automated.

23:02.22
aidecodes
That's fair.

23:10.30
aidecodes
A.

23:18.60
Rama Chellappa
Ah, vehicles you know can go and so on maybe in Battlefields and things like that you know that is potentially possible and but in in our lives day-to-day lives. There's going to be a mix.

23:25.49
aidecodes
Um, okay, do you think at any point um that insurance companies when they start to see how effective automation is will. Charge more for your insurance If You don't have those features.

23:42.59
Rama Chellappa
Is possible but you know I think they're going to look at their risk and and and see if for example even now.

23:47.62
aidecodes
A.

23:55.35
Rama Chellappa
Ah, you know they can they for different age groups. They have different rates right? for? yeah you know folks younger than 25 the charges are higher if you're an urban area so they calculate based on the risk so for them to come to that they should be enough number of automated cars so they can really see the impact.

23:56.45
aidecodes
And right.

24:13.10
aidecodes
Yeah, ah.

24:14.91
Rama Chellappa
And then they can you know revise it. But I think if my rates go up that's fine I may still prefer to dry because I like to have control of the steering wheel just a way I despite being an ai researcher you know, but some in fact, there was a poll some years ago

24:20.74
aidecodes
Um, right.

24:34.70
Rama Chellappa
They asked people whether they would be happy to fly in a plane without the pilots you know, okay now. But I think 60 to 70% of germans apparently said yeah we are fine but now in fact, 90

24:40.27
aidecodes
Um, a.

24:46.25
aidecodes
Ah, okay, this autopilot is heavily driving them now. But.

24:51.38
Rama Chellappa
Yeah, now. Yeah but 90% of the long flights are done by computers. The pirates are you know watching it and then they go take off landing. You know they they they do that. But and once they put the coordinates. You know it kind of takes care of the flights right? but still.

24:56.62
aidecodes
Um, yeah, right.

25:08.47
aidecodes
Um, man.

25:10.93
Rama Chellappa
I am happy to see a greyhaired ah ah pilot when I am sitting in my seven seven seven the big boy you know air corps I said okay so this guy was stuff driven for some time flown for some time so he probably knows where all the buttons are so I feel that so again, the comfort for the passengers right? so.

25:12.12
aidecodes
Same yeah. Right.

25:28.45
aidecodes
Um, yeah I Definitely agree I I am not as comfortable with um, full self-driving I'm ah I'm okay with like some of the I Love the features I think it's great. Um, like I've driven teslas and I love the features and how it like helps support me.

25:29.84
Rama Chellappa
Yeah.

25:44.34
Rama Chellappa
Yep.

25:44.45
aidecodes
But I don't think I'll be to the point at least in the next ten to fifteen years where I'm like full um give over full control I just can't and I agree with you like airplanes have been driving us for a while now. Um, but it's still something about seeing that human in the cockpit that makes me feel. Okay I don't think I could walk in there. There is no human there.

26:02.28
Rama Chellappa
Yeah, yeah, right? Yeah, in fact I used to joke that the Airlines pay pilots to color their head white so that the.

26:04.32
aidecodes
Or even if it's a robot that looks like a human like someone has to be there that gives me some form of comfort and what I'm used to.

26:14.78
aidecodes
I Think so I have never seen one. This hair isn't white now that you said it. So I think it does make you feel that oh this is the expert and makes us feel all feel safe. So I think there's something there. Yeah.

26:16.66
Rama Chellappa
The the passengers feel more comfortable I don't I don't I don't I don't think that's happening but I'm just joking.

26:27.47
Rama Chellappa
Yeah, right? Yeah yeah, you want to be comfortable when you are when you are at yeah yes, yes.

26:34.46
aidecodes
And that's important with technology right? The whole building trust and um, like sharing that and I know in the last week is been a little crazy out here with the Ai artists conversation particularly around things like linza.

26:45.51
Rama Chellappa
Yeah, yeah, yeah, yeah, yeah.

26:47.70
aidecodes
Um, and other applications that are using Ai and then I know chat Gp has been all the buzz now is real and the topic ah Ai is bringing a lot of people now into the conversation who never thought about it. What are your thoughts on how Ai might be misused in the future. Some of the ethical challenges or questions that come up for you. And how do we actually safeguard so we can move towards the future where we're working with Ai.

27:07.70
Rama Chellappa
Yeah, right right? I think you know there are some applications of Ai where the algorithm may not be doing the best thing possible face recognition. You know is is often discussed in this context.

27:18.39
aidecodes
Yeah.

27:26.55
Rama Chellappa
It's possible if the data is not properly curated. We don't use the proper last functions and so on the performance may be different for different groups of people if you train using you know, let's say Caucasians and then you test it on you know Asian faces. You know it's going to be.

27:29.98
aidecodes
You had.

27:38.97
aidecodes
And then a.

27:45.43
Rama Chellappa
Lower so we have to make sure we train it properly now what I tell miss I v is that it's still an algorithm I can probe it I can give examples I can see how it behaves I can evaluate the results and so on so well Ai bias.

27:46.77
aidecodes
Okay.

27:56.80
aidecodes
And.

28:03.23
Rama Chellappa
Is something that we should be concerned about but we can also reduce mitigate Bias Maybe not to zero but we can manage right? So I tell people so you compare with human Bias you know.

28:12.60
aidecodes
Um.

28:14.70
aidecodes
Who. Um, right.

28:20.40
Rama Chellappa
That's that's so we can talk about that for a couple of hours right so what are you comparing with? that's what I asked them you know and some humans have biases about other humans and how they look how they speak where they come from and so and there is no algorithm to mitigate that hopefully only by.

28:30.60
aidecodes
Okay.

28:39.81
Rama Chellappa
Educating from Childhood onwards you know when a child is born a child has no bias child doesn't know Bias who some child may be shy. Some child may be outgoing but you know but they're not going to I mean they're always very of strangers as they should be.

28:40.36
aidecodes
Right? right.

28:55.98
aidecodes
Okay.

28:59.11
Rama Chellappa
Right? Ah, but I don't think they carry a by but then over a period you know it kind of goes this way or that way. So I always think tell people let's think of what we are comparing with I can't even measure. Anything with human bias I can't even you know I can only educate them. We have all these retraining this and that we do sometimes but ai bias can be measured I can probe it and I can figure out a way to reduce it and I cannot guarantee hundred percent it will go away.

29:12.30
aidecodes
Right.

29:23.86
aidecodes
Um, and.

29:30.25
Rama Chellappa
And there is a tradeoff between Bias and the performance and so forth. So so this is all things that we can analyze and look at so that is the comforting thing that we have about because it's after all a technology you know it's It's not magic carpet. It's it's in your computer and we have a code we can look at it We can.

29:32.40
aidecodes
Right.

29:40.89
aidecodes
And it's.

29:50.19
Rama Chellappa
Fix it. So most technologies don't do well hundred percent you know this on day one but they get better.

29:58.14
aidecodes
I definitely agree with you I know? Um I've had conversation with folks around predictive analytics in particular um and different algorithms and how you can use those actually to identify that bias and to make sure it takes that into account. And I think a lot of people don't think about that that you can train it and help it understand what the current state is so it can keep that in mind and I love that you said the pieces that you were talking about around. We can test it we can um make modifications. It's not like we're putting 1 thing out and we stop there like you said model t to tesla. Um, we have the ability to.

30:17.69
Rama Chellappa
And yeah.

30:25.81
Rama Chellappa
Yes, yeah, right.

30:31.81
aidecodes
Play around with things and at the end of the day. It is never going to be his bias as humans. Um, and even with folks who are a little bit more concerned around like self-driving cars and accidents with self-driving cars still never in most situations will not have as many accidents as humans. Um.

30:34.91
Rama Chellappa
Yeah, yeah, yeah.

30:46.10
Rama Chellappa
Yeah, because humans I see people carrying a cell phone and then driving with one hand I think we have become addicted to the ding sound from the cell phone the moment it is there. We have to see what the message is right? Yeah, so and then you know.

30:50.44
aidecodes
Oh yeah.

30:58.25
aidecodes
Ah, immediately.

31:02.59
Rama Chellappa
Ah, we did some study for department of transportation I don't know in the mid 90 s we put a camera dashboard camera to see driver glance and it was startling to us that only 20% of the time people are looking in the road they're looking here. They're looking. They're looking down and you know yeah I mean it was really crazy and.

31:05.70
aidecodes
Are a.

31:16.80
aidecodes
Um, who oh my gosh me.

31:21.12
Rama Chellappa
And and so so we gave that report and so so you know accidents are happening like that. So anyway I think the point is we can fix. We can address the bias in an algorithm.

31:26.70
aidecodes
Right.

31:32.24
aidecodes
Right now that's super helpful and back to your point about folks are looking at the road 20% we definitely need Ai to support us in this future we're we're living in for sure.

31:44.23
Rama Chellappa
Yeah, yeah I think we can think of having a camera looking at say hey hey look look straight down the road. What are you doing I mean as is possible is possible. Yeah, yeah, yeah, it's doable. Yeah.

31:53.58
aidecodes
Right? Or that or but tells you you've been looking down too long look up. Um, and I know we're talking about biases now. But what are some of the biases.

32:01.57
Rama Chellappa
Is doable. Yeah.

32:07.14
aidecodes
Um, that are inherent in the software like things that you just expect to come up I know we start digging into a little bit but like more specifically what are things that you're concerned about and that you see come up.

32:16.67
Rama Chellappa
Yeah I think you know we built a face recognition system and then we we probed it and we found it had some bias to gender and it it did better for males and females and then it had bias towards skin tones and so on.

32:20.71
aidecodes
A and.

32:33.95
aidecodes
Are um.

32:34.40
Rama Chellappa
Um, and then we you know developed a couple of algorithms to reduce the biases. So we published it so we are telling people this is possible. You know, but you have to explore you have to probe the algorithm. In fact, there is this 2018 study that came from mit that.

32:42.20
aidecodes
Um, okay, great and.

32:52.16
Rama Chellappa
Looked at you know, phases of light skinned males and females from Netherlands and dark skin males and females from some african countries took 3 existing commercial face recognition algorithms and did gender classification now.

32:57.36
aidecodes
A.

33:03.18
aidecodes
Here.

33:06.47
Rama Chellappa
Some careful study was done about that by one of my friends at Caltech a professor pietro perona of course the first people thought you know it is being sensitive to know gender and and and skin tone and it was not you know doing well for dark skin males and females and so forth. But what happened in that particular dataset I think is the the hairstyle was also very similar because the conservative parliamentarians from netherlands and their faces from the african countries they belong to politicians and newscasters and so forth so that everybody had a short close hair.

33:32.68
aidecodes
In.

33:44.70
Rama Chellappa
So the machines can get confused. It can make mistakes so you know what do we attribute to the mistakes. Do we attribute to the fact that they all have very similar hair style or you know it's actually a skin tone and the gender and so on so they're all kind of in the mix so that made us look at our own.

33:52.66
aidecodes
And.

34:01.63
aidecodes
Okay.

34:02.66
Rama Chellappa
System to see if there is a bias and then we told how we can reduce the bias and developed ah versions that have less bias than the ones before. So I think ai in the good old days was based on domain knowledge. For example, if an Ai plays chess. It understands the rules for the chess is programmed and then it looks 30 moves ahead and that's why it's able to beat grandmasters now it's all data. So if the data is not of high quality properly collected curated and used.

34:21.40
aidecodes
Right. A and a.

34:37.49
Rama Chellappa
You're going to see issues. In fact, it's the statistics you know theory that when you have noise in your data. You're going to have bias in your estimates. It's it's you know is Prun We can prove that right? So so how you know we have to be in in face recognition you you kind of.

34:45.58
aidecodes
Okay.

34:56.83
Rama Chellappa
Can figure out what you may be biased to but in general Ai it's hard to figure out so but there are methods to probe. You know what the output depends on and then how much they are expressed.

34:58.74
aidecodes
And.

35:09.99
Rama Chellappa
In the features that we learn for the ai to work and then how we can kind of moderate them and so on. So no, not 1 thing becomes very dominant. You know? For example, you know if you have 5 people and one of them is very strongly opinionated and you know.

35:17.12
aidecodes
E.

35:28.85
Rama Chellappa
Talks very loudly and maybe the conversation kind of may all go towards what that person is thinking right? So so we have to make this activation map. We call it.

35:31.81
aidecodes
Correct.

35:37.37
aidecodes
A.

35:40.80
Rama Chellappa
Neural Networks So we have to kind of watch them and make them behave equally well for all populations and things like that. So that that technique is available.

35:46.33
aidecodes
Um, that's super helpful. Thank you for breaking that down because I know people the first thing that always comes up is bias and Ai and a lot of people bring that up another issue that comes up often is.

35:53.92
Rama Chellappa
Yeah, yeah. Yes, Yes, Oh yeah, I think what has happened now is that we have to be very savvy. We have to know what information we are sharing and what is the outcome of that. The example I give.

35:59.49
aidecodes
Privacy what are your thoughts on the privacy front as it relates to Ai.

36:08.99
aidecodes
Um, okay, um.

36:15.69
Rama Chellappa
And I go to supermarket I give my telephone number. There's a bonus savings they call it I get you know 2 strawberry boxes and 1 blueberry box free $2 of shaving cartridges and things like that. So here I exactly know what's happening I'm giving my phone number I'm getting some discounts.

36:18.25
aidecodes
Um, yeah.

36:22.40
aidecodes
Right.

36:28.14
aidecodes
E.

36:34.21
Rama Chellappa
But there are other situations. You know when I use a platform social Network platform and this and then I don't know what exactly So we have to be careful if you put something up there. You have to assume that it will go somewhere and you know, etc now what I'm saying is that some of these.

36:43.91
aidecodes
Yeah.

36:50.84
Rama Chellappa
Platforms. You know they give you maps. They give you gmail and all of those things so they have to support that and so they like to use the information you know and sell it to the advertisers and so forth So you have to be very careful What you give it suppose if I just say I'm sabi can I have your social security number. Please.

36:59.66
aidecodes
You have a.

37:08.17
aidecodes
Immediately? No, but we give up almost the equivalent.

37:09.40
Rama Chellappa
You're not going to give it to me. Yes, right? Yes, exactly So you have to you know right away right? Where are some things you have to be careful. Okay, so privacy is something depends on an individual's comfort what you put out.

37:15.75
aidecodes
So yeah.

37:23.98
aidecodes
Sure.

37:27.49
Rama Chellappa
I mean younger generation seems to put everything up I would not put any of my crazy dancers on Tiktok I will not right I mean my generation is there be oh my god what is this? but then yeah, exactly yeah, first of all, it won't look pretty so why should I put it up so you know, um.

37:35.91
aidecodes
I Like no right.

37:44.73
Rama Chellappa
Ah, so we have to make a smart choice. Nobody is going to yank thing from you if you don't give it to them. So so you have to be careful. You don't give out your credit card number right? You don't give your bank account number suppose you get an email.

37:53.64
aidecodes
Um, yeah, estra.

38:01.87
Rama Chellappa
Well we have just come to $200000000 and we don't know how to spend it and if only you give me your bank account social security number drivers license I'm going to deposit 50000000 into account if you respond to that then well no, you should not you so we have to be savvy.

38:13.46
aidecodes
Exactly.

38:18.23
Rama Chellappa
Because there are always people who are trying to exploit our vulnerabilities and and and so on so we have to be savvier. So what I tell you is that watch it be careful. Be alert and and you know be careful about what you share.

38:21.67
aidecodes
Okay.

38:27.23
aidecodes
Okay.

38:35.30
aidecodes
So I definitely agree with where you're coming from around like 1 a lot of the apps right? now we were talking I was talking about this with some folks around some of the apps that are more popular with the ai artists that are getting folks to download their images and things like that into them. Um, and folks are concerned right about where their data is going and things like that. In particular I know you have a lot of experience and face recognition. What are some of the privacy concerns with that.

38:56.65
Rama Chellappa
Yeah, the the way the data is collected for training phase recognition system must be about board must be transparent I actually served in a committee that was constituted by a state senator in Maryland.

39:06.29
aidecodes
Um, okay.

39:12.34
aidecodes
And.

39:14.98
Rama Chellappa
Mr. Senator Sidor and there is a discussion about that in chapter 3 and he was trying to get some regulations passed because of privacy concerns and you know whether it will be abused misused and so forth.

39:19.17
aidecodes
And.

39:27.51
aidecodes
Yeah.

39:29.71
Rama Chellappa
And that that team had you know police chiefs it had Aclu people. It had attorney state attorney office representatives and so on and we we came up with some kind of a bill but still it didn't pass this time I think because there were still some issues whether you know.

39:41.46
aidecodes
Okay.

39:47.90
Rama Chellappa
Auditing is allowed or not in terms of how you know police will use a face recognition system. So so I think it has to go through a few trials you know. So yes, we have to there are some systems that are working now where they have collected face from.

39:51.17
aidecodes
E.

40:05.82
aidecodes
Right.

40:06.21
Rama Chellappa
Web everybody without really playing any concern about whether they give consent or not so those systems run into trouble sometimes because you know they can get into legal difficulties. So we have to be able to.

40:10.94
aidecodes
So.

40:23.26
Rama Chellappa
Say how we trained our system. What dataset we used how we you collected the data all of those things have to be transparent so people can trust that nothing unto was done here in designing this system. You know if you look at various drugs that we are all you know. Taking in behind a very such drug There is a whole set of protocols. They will tell you how they tested it and you know what happened? What are the side reactions. Ah so that kind of ah auditing has to be done for any Ai algorithm so people will know exactly.

40:50.42
aidecodes
Ah, okay.

40:56.92
aidecodes
A.

41:00.95
Rama Chellappa
What happened here and that is that is the way that that may hopefully you know address some of these concerns I think face is very personal so we are very careful about it right? So yeah.

41:10.66
aidecodes
It is yeah because I know I remember um when I one time I was in my iphone I looked at the section that said had grouped all my pictures by face and I'm like I didn't ask for this. Sure I agreed to something in some terms and conditions. But I didn't ask for it and even I know on us like sites like Facebook same thing they're like is this so-and-so so or what have you and they're doing a lot of um, looking at your face to to recognize you. Some people are okay with it I'm personally not so I do understand where you're coming from and I I agree.

41:33.47
Rama Chellappa
Yeah, yeah.

41:38.94
Rama Chellappa
Yeah, yeah.

41:45.66
aidecodes
Wholeheartedly that there should be um, like guidelines around how you explain to the public. What's happening with these different devices and I'm glad you call that out.

41:52.58
Rama Chellappa
Yeah, right? Yeah because all you see sometimes is that you know you are being videotaped here. You know, just a little note right? So maybe we should then all wear a mask and and and they walk around right? So ah.

41:58.31
aidecodes
Right? I mean.

42:06.67
Rama Chellappa
But but they do announce they have to announce that they do. But I think when they design a system and they put it out for ah use I think more details should be provided about how the data was collected.

42:12.78
aidecodes
Yeah.

42:20.66
Rama Chellappa
Going to be happening but right now there is no federal regulation for face recognition. So some cities have taken some actions and some you know cities are wondering what to do with it and so on so it's in a kind of limbo right now.

42:22.83
aidecodes
Right? so.

42:32.94
aidecodes
Yeah that's the interesting part is like I feel and I could be wrong, but I feel like at 1 point government was way ahead of innovation on things like the internet going to space and all that and I feel like now government is more so chasing. Innovations as it relates to how we regulate or think about those things and in the absence of that a lot of technologies are popping in which is great I'm all here for innovation like I love this stuff. But then the end of the day is I'm wondering like how then do we make sure like you just said if if.

43:01.63
Rama Chellappa
Um, yeah.

43:06.42
aidecodes
Cities are popping up cameras on the corner and we see that in a lot of different places DC or and a lot of other places where literally all these recordings are happening what's happening. Um, how are they stored what's happening to that data like I don't know personally um so it's just interesting to see where this is going.

43:08.64
Rama Chellappa
Gift.

43:18.10
Rama Chellappa
Yeah I think in Ai private sector is and I had right there. They're just going through. Yes yes, and.

43:27.17
aidecodes
Yeah, there they were government I used to work for government we are behind hundreds of years I hope.

43:36.63
Rama Chellappa
Hopefully tens of years but but private companies are going. You know going ahead and you know government entities are catching up and and so on so it's it's ah.

43:42.59
aidecodes
Yeah.

43:52.53
Rama Chellappa
Well, it's been like that I mean boeing is a private company build great planes and sorts lockheed Martin and so and then you know company government buys planes from them.

43:53.68
aidecodes
Oh yeah.

44:01.77
aidecodes
Yes, so.

44:04.48
Rama Chellappa
So I think armmind will buy a systems from these various companies that are doing you know cutting edge research in. Ai yeah.

44:12.93
aidecodes
Yeah I definitely in Greeks I know even folks who are doing things in predictive they're using private sector technologies I think maybe one county I've worked with they created their own but that's rare um is more often. The private sector is leading in this space for sure. Um.

44:16.24
Rama Chellappa
Yeah.

44:23.94
Rama Chellappa
Right? yeah.

44:27.62
aidecodes
1 of the reasons why let's ship the conversation a bit because I know people are listening and they're like this ai stuff sounds cool. It sounds like it's the future. You're telling me on Dr Rama that we're gonna be working together with Ai.

44:28.85
Rama Chellappa
Yeah, yeah, yeah.

44:41.30
aidecodes
What would you say to people who want to dive into this fill and how can they get involved in it if they see it as a career or a place where they want to learn more. How do they do that.

44:47.42
Rama Chellappa
But yeah, definitely we actually are working on bringing a high school unit 2 units to discuss Ai in the context of. Ah, self-driving cars and so on and the lessons are being attend now and things like that. What kind of questions should be asked and so forth I think if anybody who has computer science background.

44:59.95
aidecodes
Who.

45:09.37
aidecodes
A.

45:11.32
Rama Chellappa
Can easily you know, get into that. It's It's very easy. It's a start in undergraduate curriculum and computer science and engineering in in most places. Um and you can read you know a lot of books and of course many of them are technical So I don't want to be advertising my book but it.

45:18.66
aidecodes
A.

45:24.30
aidecodes
Yeah, advertise it. That's what you're That's what you can do that here.

45:30.70
Rama Chellappa
Yes, it's it's something. Yeah that's they can read and there are other books like that there's actually another book written by Henryri Kissinger and they're one of my colleagues in computer vision Dan Hutton Locker at mit and and so on they have written a book about ai um. So there are lots of of course these days so much is available in the open literature if we just just have to you know, go and and there may be even courses given in and community colleges and so forth to learn about ai. They can do that? um.

45:51.79
aidecodes
Oh yeah.

46:00.86
aidecodes
Ah.

46:04.23
Rama Chellappa
And think those are the places and then of course we come to graduate school and we give you a lot of courses to take Ai yeah, it's Ai is everywhere now to to learn and it's it's a fascinating feel.

46:15.15
aidecodes
And it sounds like you're not saying it's impossible. It sounds like you're saying It's very possible for a listener to go learn more about this.

46:19.60
Rama Chellappa
Oh yes, oh yes, very much. So I mean I never I Never think that anybody cannot learn anything they want once they set their mind to it. That's all it's simple. The focus attention.

46:29.83
aidecodes
A a.

46:38.38
Rama Chellappa
And spend some time with the books are podcasts a lot of video lectures available. A lot of podcasts on a various aspects of they can ah learn and it's you know it's not. You know it is It is learnable just like any any computer science course.

46:41.38
aidecodes
Oh yeah.

46:57.33
Rama Chellappa
Engineering course it's learnable and if you are interested in. You know, not an engineering person but you are interested in ethics and so forth ai also gives tremendous opportunities now to explore you know Ethical ai has become a field by itself.

47:07.83
aidecodes
Um, yeah.

47:11.94
aidecodes
And.

47:15.90
Rama Chellappa
So you can do that if you are in medicine and you want to apply Ai to medicine. You know they do Md ph d the ph d component can be an Ai and they can do Md are if you want to do business financial intelligence. You know that's also possible wherever there is data.

47:24.46
aidecodes
Um, a and.

47:33.29
Rama Chellappa
Now Ai can jump into it so you know and data is everywhere for example, legal things Ai can summarize it for you. Yeah, if you have a hundred page you know thing if you don't have time to read. It may just ready it to 2 pages so it will useful in law.

47:40.95
aidecodes
Yeah, oh yeah, yeah yeah.

47:53.20
Rama Chellappa
Useful in business useful in Medicine useful in engineering design and so on and so forth. So anybody you know, maybe we can even ah train an ai to do your taxes. Yeah.

48:03.30
aidecodes
Texas I think we can at some point I believe that? yeah, but yeah, but.

48:07.76
Rama Chellappa
I Mean we do have turbotax and all of those things but they are not thinking you have day enter and they kind of verify. But yeah I can say you know what given this? Maybe you can take this detection. How about that it.

48:18.33
aidecodes
I would support that and even on the legal side. Um I have my JD and one of the things that was very sad for me when I was in school um because I'm a math person as well. But a lot of lawyers will say.

48:21.25
Rama Chellappa
Yeah, yeah, yes. Yeah.

48:32.12
aidecodes
I don't do math. That's why I'm a lawyer but I started I was telling them like hey like Ai and other technologies are literally I think there was a a I um tool that beat a contract employee and was able to go through way more contracts than any of us will be able to read or legislation and things like that and help um with decision making and things like that.

48:48.91
Rama Chellappa
This.

48:51.84
aidecodes
So it's just very interesting. How some feels I'm a little nervous they're they, they're gonna have to get on board with technology when they're not necessarily looking at it right now.

48:55.99
Rama Chellappa
Yeah, you know I'm at this interesting I always felt to be a great lawyer. You have to be a great Mathematician because because because you know logical reasoning. You know that? and.

49:06.50
aidecodes
They don't believe that Yeah, the logic piece is there but they don't equate it with it. Yeah to me, it's the same thing.

49:15.96
Rama Chellappa
Yeah, right that that that is something to do the mathematics exactly if that law says XYAndZ and the the conditions are a v and c yeah, exactly where it is applicable. Where is it not so I actually had on meeting I was there and.

49:21.43
aidecodes
Take away see it That's math or algebra.

49:32.39
Rama Chellappa
A person who was raising questions about privacy and face recognition and ethical considerations is Ph D was in mathematics I said Wow That's good. So I said yeah you you can point precise. It can be very that That's what mathematics mathematics teaches you to be precise in what you say and so definitely.

49:35.75
aidecodes
Is the issue of thread.

49:42.95
aidecodes
Site exactly that's but at the benefit just to leave site.

49:52.25
Rama Chellappa
Ah, to be a good liar. You have to be very good in mathematical reasoning.

49:58.29
aidecodes
I'm glad you said that because now every lawyer listening you've heard it from a doctor a Ph D who knows this stuff you all can do its logic. It's the same thing. So thank you for saying that out loud because it's truly important. Um I want to talk about really quickly. Normally I ask on ask every.

50:09.60
Rama Chellappa
Yes.

50:15.99
aidecodes
Guests on the show like how do you think your particular topic will impact the world. So I'm gonna ask the same of you. Why do you think it's important for listeners to understand how Ai will affect our lives.

50:25.45
Rama Chellappa
Yeah, let's think back and look at some of recent Technologies Internet suppose you choose. Okay I'm not going to have to anything to do with internet think about what your life would be are.

50:30.13
aidecodes
Victim Okay, internet you so again I'm not going what.

50:41.58
Rama Chellappa
Okay, cell phones. No I Only like the ones that are tethered to the wall. Okay, so what would it be right? So we had you know web web technology right? Okay so we can think of all these technologies and we all adopted it and we learn how to use it.

50:43.61
aidecodes
And.

50:49.81
aidecodes
Bad technology so we can.

51:00.26
Rama Chellappa
To improve our efficiency to improve our knowledge and to improve other things right? So Ai you should view it as another technology which is now able to process the data that's available that's being collected.

51:07.93
aidecodes
That is not.

51:15.87
Rama Chellappa
And then it'll provide some and help with some making some informed decisions right? and depending upon various stages of your life. It can help you when you're young with education especially for students who have.

51:20.48
aidecodes
Very sure about.

51:27.36
aidecodes
Yeah education. Yeah, right? and.

51:32.83
Rama Chellappa
Learning disabilities it. It can be patient. It can go over the material again and again and teach and so forth years ago. Somebody asked me what my moon shard would be in education I said I want one physical teacher and 30 virtual teachers.

51:37.66
aidecodes
I get.

51:45.12
aidecodes
Of your teachers that right page with the again.

51:47.40
Rama Chellappa
Because a class has 30 students and the teacher cannot pay equal attention to all 30 students. They always focus on the middle so to speak and there are some very smart ones and they may get bored and there are some who are struggling to catch up and who are left behind now Ai can kind of get her to personalized you know Medicine Personalized education

51:52.13
aidecodes
Yeah, better. Yeah.

51:59.70
aidecodes
Yep.

52:03.17
aidecodes
And and.

52:06.67
Rama Chellappa
And so on so all of these things are are are going to be there and so we should adapt the technology you know it's ah, a new. It's not a totally new technology as I said Ai is you know ready to collect social security hence medicare it's it's it's an old thing but it's using, no no.

52:19.69
aidecodes
Is.

52:25.86
Rama Chellappa
Techniques that have come up because we have data and we have computing power. We are understanding how to mine the data and and and make it work for us. So I think they should embrace it. They should be cautious so we all use internet and we we are. We are told.

52:39.50
aidecodes
Here.

52:44.33
Rama Chellappa
When you get a suspicious email don't click on it. So we we listen to that if we don't listen to it then we are going to be in trouble and somebody will hack into our machine and so on so that doesn't mean that you'll never stop using your computer.

52:46.17
aidecodes
Right? e.

52:51.66
aidecodes
Oh.

52:59.58
Rama Chellappa
You won't use internet or you won't use web and so on no we have to use all these technologies and Technologies somebody will abuse it for their own personal benefits and so on so we have to and watch for it and be careful. Be alert.

53:06.93
aidecodes
Okay.

53:15.30
Rama Chellappa
But let's let's say explore this technology. Let's embrace ai and let's see what it can do for us in various ways in various portions of our lives if you are elderly we may like to have a bunch of variable sensors telling us you know how things are going for us. So I tell people what about this wearable sensor I say well just like a Tesla has cameras and abs and all those things we can have fitbits. We can have Apple watch. We can have other kinds of set senses that can watch for what's happening and tell us you know if things are okay, you know so that is potentially possible in the last you know.

53:36.58
aidecodes
Yeah.

53:52.86
Rama Chellappa
Few decades of our lives and so forth. So you can see I can you know work with you for the entire you know life segment at various places it can I think it's actually interesting to think about it.

54:09.71
aidecodes
It is I'm actually getting very excited in this conversation. Yeah, right? yeah.

54:09.94
Rama Chellappa
For your 2 year old. We talked about yeah Chad from Chad Gbt onwards 8 year old for you know playing chess maybe a I can play chess with a kid and tell learn no moves no moves minimo and all. And and and so on and then education and then you know various new technologies can be tar and you know it just goes like that. Yeah, so it will it is something to understand to learn and I think that will be good.

54:39.56
aidecodes
And I definitely love the idea of the education piece especially as we talk about education in this country to be able to give a child an Ai that can focus on them their learning styles their ability and even work with them outside the classroom. Um, that would be amazing and I just think there's so many potential with it. So I'm gonna ask you the question on the cover of your book and want to hear your answer. So Dr Shalapa can we trust Ai what awesome I love it. So.

54:57.60
Rama Chellappa
Yeah, yeah.

55:07.89
Rama Chellappa
Qualified s.

55:11.69
aidecodes
People are going to want to know where they can get your book where they can get in contact with you feel free to do any promotions and talk about the book more where can we find you? What do you have going on. Yeah.

55:20.30
Rama Chellappa
Yeah, the book is available in Amazon and and and other places you can order books and John Sopkins University Press also can send you these books? Ah my I'm not present in Twitter.

55:35.46
aidecodes
Yeah.

55:35.89
Rama Chellappa
Tiktok and those sorts of things Facebook I have a small presence once in a while I'll look at what my other colleagues say years ago when I wanted to get in to understand what Facebook is all about I asked my son whether he'll be my friend. He said yes I asked my daughter she said absolutely not. So. I understand and so you know so they can go to my website. A iemdotjh you dot adu. It's for Ai for engineering and medicine.jh you for edu and as we talk about education since Ai can be global.

55:58.10
aidecodes
Yeah, awesome. Did you.

56:11.10
aidecodes
A a.

56:11.16
Rama Chellappa
Just imagine how much of learning and education can be done across the globe. You know after all I was a skinny kid in Chennai and you know I came to know of United States and I was in third grade because my headm mistressres.

56:26.10
aidecodes
Is.

56:30.62
Rama Chellappa
Coming to United States to take a tour of the schools here and that's the first time I heard United States of America right? I was eight years old and the reason why I remember it is there was a big function in the school and the teacher you know the headminters came and said she's going and they put garlands on her and then you know and so forth I said oh this must be something. Ah.

56:31.79
aidecodes
Just.

56:44.80
aidecodes
Oh wow.

56:50.37
Rama Chellappa
You know, cool place because if they're making such a big deal about that and then when I was ten years old I wrote to the us embassy in Delhi asking for a book about United States and two months later I got a beautiful book. So what I'm saying is this is before when Ai was not.

56:51.47
aidecodes
And.

57:02.21
aidecodes
I Love it. Thanks.

57:08.10
Rama Chellappa
Everywhere now Ai can be everywhere so kids everywhere can dream and and then think of technologies and build new technologies and so forth I think Ai can can be a force for good.

57:21.50
aidecodes
I love it. I think there's no better way to end than that. Um folks listening will make sure to drop everything Dr Shalapa said in the ah show notes so you can get his book make sure you check it out is a great read. Thank you so much for joining me today and I look forward to seeing what you're doing in the future happy holidays that you too.

57:33.77
Rama Chellappa
Yeah, thank you very much have a great day happy holidays. Yeah Bye bye.