- 10 hours ago
Ivana Bartoletti, Global Chief Privacy & AI Governance Officer, Wipro, who is also the founder of Women Leading Artificial Intelligence, said the concern goes beyond participation in coding roles.
Category
🗞
NewsTranscript
00:00This is a time when we've just heard about a world of agents and you specialize on the
00:08subject of governance and privacy.
00:11I want to begin this conversation by understanding from you for the benefit of all our viewers
00:17and those who are here.
00:19In a world of agents where perhaps there will be multiple agents working for you and me
00:24in the future, what will privacy mean, if at all anything?
00:30Easy question.
00:34So AI agents is probably the most fascinating and interesting area.
00:40The idea that, as we just heard, we can have an agent to do things for us and we can
00:45say,
00:46book a trip and do this and do that for us.
00:48It's incredible.
00:50But AI agents bring enormous challenges.
00:53And the challenges are not just from a privacy standpoint, because, of course, once you start
00:58giving an agent access to your wallet, your bank account, what if something goes wrong?
01:04What if, for example, from a security and vulnerability standpoint, what if your money gets lost?
01:11So the challenges around AI agents are very much about security and resilience that have, of course, an impact on
01:21the protection of data.
01:23So we've got to be really careful about it.
01:26And there's also one other element.
01:29Agents, they bring together LLMs, all the things that you use on a daily basis.
01:34But they really make the challenges much bigger.
01:38So, for example, they hallucinate.
01:40But it's not just an hallucination like a normal LLM.
01:43It's hallucinations that go all the way down and pass from one agent to the other, from one.
01:50And so they can become quite risky.
01:53So what is the point here that I'm trying to make?
01:56Is that we have to build this tool with safety and resilience at the very heart.
02:05This is very, very important.
02:07And it's very much the heart of this AI summit, isn't it?
02:11The heart of this AI summit is about democratization, is about everybody having access, and is about governance.
02:22So governance doesn't mean that you stop using AI agents.
02:25It's the opposite.
02:26It means that you make them so safe that people can use them and trust them.
02:31It's not going to be an easy one.
02:32But this is where I'm really looking forward to leadership coming from India as well.
02:38So the leadership I'm after is, technically, technically, how are we going to safeguard the data people through agents?
02:47Now, I see young people in the room, and this is a challenge that I would like to throw to
02:51the young people.
02:53And say, come up with security-enhancing technologies, privacy-enhancing technologies, so that we can build this.
03:01Now, I do the best job in the world, which is to work with product managers and try to embed
03:07all these protections in the design.
03:09So I work on how do we design these agents, how do we create the trust stack so that these
03:14agents are well designed and people understand them.
03:19And, for example, I say to them, if the agent has access to particular information, let's escalate it to a
03:26human to say yes or no.
03:28This is very important because you've got to remain in charge.
03:31If the agent is going to access something very personal about you, then you may say, well, actually, on this,
03:38I am going to ask the owner to say yes or no.
03:44So it's all about design, see?
03:46Privacy, data protection, security, resilience, it's all about how we design this tool.
03:51What is the role of creating a governance architecture around this?
03:56The reason I ask you this, because you lead this function at one of India's largest information technology companies and
04:04for the benefit of our audience and for our viewers, she's Italian.
04:09She lives in Frankfurt, but she works out of a company that is headquartered in Bengaluru.
04:14So, you know, a global citizen if there was one, but can there be a governance framework for this future
04:24world, including whenever, you know, we have super intelligence, some experts say it's a few years away, somebody says sooner,
04:33somebody says later.
04:34But in that era, what kind of governance framework can we aspire to?
04:39I mean, I think governance is kind of a strange word to an extent, you know, because if I say
04:44to you governance, what is the first thing that comes to mind?
04:48What is the first word?
04:49I mean.
04:50Well, heavy handed state deciding the rules of what you can do or not do.
04:56Do you agree?
04:57See, that's not the case.
04:59So I'm here to really say it's not the case.
05:02So when we talk about governance, we talk something else.
05:05And I've learned this in my day job.
05:08I come from privacy.
05:10And actually, I come from sort of a legal background.
05:13And then I learned how to code and I went into the technical stuff.
05:16And I'll tell you what I learned on my own skin at Whipro.
05:20Okay.
05:21I learned that governance of AI has got nothing to do with compliance.
05:26It's got nothing to do with the law.
05:29It's got nothing to do with risk management.
05:31I mean, it's got something to do, but it's not that only.
05:36Governance is about making sure that you have long-term innovation.
05:42A lot of clients that I work with, they say, well, okay, this is great.
05:49You can do this, this, and this.
05:50But is it going to be effective?
05:52I mean, don't forget that a lot of AI is not bringing the ROIs that we expect.
05:59Why?
06:01Why?
06:02Because we are not, a lot of people are not choosing the right models, the right approach,
06:09the right solutions, the right use cases, the right risk appetite.
06:14It's really, so governance is not about strongholding and, and it's about enabling.
06:21So please change the approach.
06:24You've got to, I did this sort of massive rethinking within Whipro.
06:27And I said, actually, we're talking about strategic capability.
06:31We're not talking about compliance.
06:33That is not compliance.
06:35We're talking about how we go to an organization and we help them innovate in the long term.
06:42So this is really important.
06:43Now, of course, you see me as a European.
06:46And immediately you say the word governance.
06:48You think about European AI Act, legislation.
06:51They want to stop everything.
06:52They want to, is that true?
06:54Is that what you think?
06:55Well, yeah, yeah.
06:56I mean, the Europeans certainly lead in governing and regulating everything.
07:00And perhaps, you know, India follows pretty, pretty much a step behind.
07:05But you've just done a deal with the European Union.
07:07Well, we look forward to seeing the benefits come to both sides.
07:11And yes, and with the United States as well.
07:13Let me take this further from a point that you made.
07:17And this is, you spoke about hallucinations and a world of agents.
07:22I give some key tasks to my agent or to an AI engine.
07:26There is a multiplicity of engines to choose from.
07:29Right now, it's the Wild West.
07:31Many of them, especially in the Indian context, are offering free services, I imagine, to suck up the data and
07:37train their models.
07:38If something goes wrong, who owns the risk?
07:42Who is responsible?
07:43Where do I see redressal?
07:45If you go through the normal system, a contract goes bad or somebody doesn't deliver a service in time, you
07:52go to the courts.
07:54Where do I have recourse as a citizen today?
07:58Yeah, so, I mean, there are different kinds of arms, no?
08:01So, you can have a violation of your privacy sort of rights into this process.
08:05And then, you have the privacy legislation in India, which has been now operationalized through the rules.
08:12And it's great that this legislation means a lot because it really makes India a wonderful place to do innovation
08:21with.
08:21Because you have that certainty of the rule of the law from a privacy standpoint.
08:26Then, obviously, you have the responsibility and liability that these companies have.
08:31So, they have sort of these liabilities that are enshrined normally in contracts.
08:36And so, you can say, well, hey, so, there's a breach of contract here, you have done.
08:42But it's true that a lot of these things, you have to bear a lot of responsibility yourself.
08:49Like, we are hearing cases where people are using these tools and they're just copying and pasting it.
08:56Or they are using, and these systems, they hallucinate and they have put false clauses into them.
09:01And then, they present them.
09:02And even organizations that we've seen so recently in Australia, they were paid by a client to do a report.
09:09And they asked CHGPT to do it.
09:10And, of course, these systems hallucinate.
09:12And hallucination, by the way, is a strange word.
09:16Because hallucination is very much a human word.
09:19You know, like humans hallucinate.
09:20In this case, we're not talking about humans.
09:22We're talking about sort of statistical and mathematical sort of randomness.
09:27So, it's a strange word.
09:31But, basically, this company in Australia, they presented this to the client.
09:35And it was completely wrong because they, you know, if one of these tools, they don't know, they make it
09:42up.
09:43And, Ivana, this is not an issue that is, you know, alien to us.
09:47The Supreme Court of India, no less, has just spoken about the fact that it's concerned about AI being used
09:53in legal petitions, in paperwork.
09:56And I think it's a great concern.
09:58What can a country like India, where you have this massive, massive number of users, many of the proprietary technologies
10:08overseas, those companies operate in India,
10:11what can we aspire to do in terms of a regulatory framework for AI?
10:17What should be the best case scenario?
10:19So, I would say don't do like what other countries are doing.
10:22You don't have to, right?
10:24It's really important.
10:25I mean, I saw when I was driving here this morning, I saw a poster, one of the many posters.
10:31I mean, you've invaded Delhi with posters about the AI Summit.
10:34It's quite nice to see.
10:36So, I saw one of the posters and said, the only one I saw with that, and he said, we're
10:44sort of dealing with AI in a unique way.
10:48And I felt, well, yes, exactly.
10:51That is what matters.
10:51No, you're not going to do things like in Europe, because Europe has got the European AI Act, which doesn't
10:57suit India.
10:58It's different.
11:00I mean, the European AI Act is an ex-ante legislation based on a culture that is completely different.
11:05That's okay.
11:06And the United States, they're acting in a different way.
11:10They're acting, I mean, don't think there is no regulation.
11:13It's regulation in another way.
11:14I mean, even imposing tariffs is a way of regulating.
11:17Yes.
11:17And so, and states, every single state in the U.S. is regulating in its own way, especially when it
11:24comes to the use of AI in the labor market, so algorithmic systems that can hire and fire individuals.
11:32So, in this doing, I feel the right thing.
11:36And first of all, because, and I mean, we can't underestimate, I've seen this to all the commentators from America,
11:44from the Europe I've been talking to, that I'm really annoyed that they haven't, it's not covered as much as
11:49it should.
11:49This is the first summit happening in the global south.
11:54This is incredible.
11:56Incredible.
11:57The first one happening in the global south.
11:59Yeah.
11:59And in a place, in a place, and I feel that being a European, working for an Indian company as
12:10vice president, I feel that gives me a very different perspective on things.
12:15And so, when I talk to them, I say, I could, you know, it's the first summit happening in a
12:21country that combines two elements.
12:24First, the ability to create digital infrastructure and scale up.
12:31I mean, I look, I look at the fact that you've done things like UPI.
12:38And I'm thinking, this is a country that can scale up digital infrastructure.
12:42Second, this is a country that is hungry to use technology.
12:47Hungry to use technology.
12:49So, when you align these two factors, that's really big.
12:54When you, on top of that, you add talent.
13:01Talent.
13:01Talent.
13:02And I know, because if I look at how many engineers and data scientists and people who have studied technology
13:08I have in Wipro, it's amazing.
13:10And the scale of expertise.
13:13So, when you add all of this and you put it together, and you also look at the fact that
13:19you have the ambition, and I say this to the Western audiences that do not often understand because it's a
13:26different culture, and it's important to do that.
13:29And I say, the ambition of creating growth with AI to grow as a country, yeah?
13:38And the ambition to say, hey, we create 20% of the world data.
13:42We also want to leverage tools and use tools where we can speak the languages that we speak, that can
13:49reflect not just the culture of, for example, that is embedded into American-created or elsewhere-created tools, but also
13:58the ethical norms that are in this country.
14:01This is where I think, that is where all the potential of India comes from.
14:07So, I wouldn't worry too much about these big LLMs.
14:12What I think is very important is the investment in data centers, the openness to the large corporations, obviously, and
14:20the investments that we're seeing, the tax relief for data centers is important because AI is not just technology.
14:27AI is earth.
14:28AI is materials.
14:29AI is conductors.
14:30People think AI is just technology.
14:32No.
14:33So, investment.
14:34What is important is the creation, so the models, smaller models that can be trained and tuned to really with
14:44India data.
14:46So, models, applications.
14:49We need applications that have real uses every day and that can be scaled across the country.
14:56We trust because this is where adoption happens and with adoption growth.
15:01And the last element on the regulatory side, the techno-legal approach chosen by India is a clever one.
15:12Very unique.
15:13It basically says, we're not going to do overarching regulation.
15:20We are going to do all-round accountability of AI, looking at the real impact in the world rather than
15:30ex-center, leveraging technology to solve some of the governance issues, which is good.
15:38And there is a lot that we can learn from this around the world.
15:41And this techno-legal approach is fundamental in a country like this because, as I said before, your strength is
15:50to be able to scaling up this digital infrastructure across the country.
15:53So, I think this is a good way of doing things.
15:57I would add the cooperation and global cooperation is also very important.
16:03I am glad that I live in Germany where there is sort of strong cooperation with India.
16:09I see a lot of openness to India's talent and vice versa.
16:12It's really good.
16:13This cooperation is important because we have to find technical and techno-social and techno-legal solutions to the future.
16:24And I want to close on one thing, if you allow me.
16:26A lot of people talk about AI capabilities.
16:29And it is important because they will increase and increase.
16:33We will be here probably or another part of the world in one year and we will have even more
16:37AI capabilities.
16:38So, I would suggest we've got to stop the focus, of course AI capabilities is important, but on the capacity
16:46that we have as businesses, as country, to absorb these capabilities and to govern them.
16:53Because this means, how are we going to govern the fact that you have deepfakes?
16:57And deepfakes are terrible for misinformation, but also they are terrible because they penalize the vulnerable people the most and
17:03women the most.
17:05Yes.
17:06So, how are we govern the step by step?
17:09So, we see harms, we intervene, and we see if existing legislation is sufficient.
17:15If not, we come up with soft and hard legislation.
17:18We don't have to immediately come up with hard legislation also.
17:21I mean, in India to legislate takes a lot of time.
17:23I mean, it's not like it's a democratic process.
17:28So, it's step by step with the ambition that we want to leverage AI to make India stronger.
17:35And if we make India stronger, we make the rest of the world stronger too.
17:38Absolutely, Ivana.
17:40In fact, the fundamental change that AI is bringing to society, to governance, to policy making, to every single field
17:50of human endeavor is humongous.
17:52And I would completely agree with you that learning to grapple with this, both in terms of the policy architecture,
18:00the governance factors, is a tremendous one.
18:03As I let you go, one quick 30-second kind of answer.
18:08Is there one risk that you think is being underappreciated in the Indian context?
18:15So, you know, 400,000 people registered to visit the AI summit in India.
18:19You made that very important point.
18:21This is a country hungry for technology.
18:23We are an aspirational country.
18:24But is there one risk that is, according to you, underappreciated?
18:29Quick answer.
18:30I think the one risk is the impact how AI can be is on women.
18:39And I'm going to say this because I'm also the founder of an organization called Women Leading in Artificial Intelligence.
18:45And it's not just about women participation into the coding side, because India's got 43% of women in coding.
18:52Yes.
18:53And in Europe 19, so we've got a lot to learn from you.
18:59But it's how that workforce present translates into power.
19:04And I think artificial intelligence has got enormous potential for women, but it also has the potential to reinforce existing
19:14stereotypes and inequalities into the systems.
19:17We've got to be very, very aware of this.
19:20And this is where I would like to see a lot of women participation into the discussion, not just coding,
19:26but also taking the lead on the policy side.
19:28That's a very important point that we closed this session on.
19:31Just yesterday, we did a session.
19:34We were partnering with one of India's largest groups with 1,600 rural women.
19:39And they ran live pilots on utilizing various AI technologies to do stuff.
19:46And I was amazed.
19:47And these were women from, you know, deep rural India and the kind of work that happened.
19:53So I really take that point as one of the most important ones that you have said.
19:57Ivana, we are completely out of time.
19:58This conversation could have gone on for longer, but thank you very much for your time with us today.
Comments