WEBVTT - Smart Talks with IBM - Responsible AI: Why Businesses Need Reliable AI Governance 0:00:04.559 --> 0:00:12.760 Welcome to Tech Stuff, a production from iHeartRadio. Today, we 0:00:12.800 --> 0:00:15.760 are witnessed to one of those rare moments in history, 0:00:16.120 --> 0:00:19.360 the rise of an innovative technology with the potential to 0:00:19.480 --> 0:00:24.200 radically transform business in society forever. That technology, of course, 0:00:24.720 --> 0:00:28.240 is artificial intelligence, and it's the central focus for this 0:00:28.440 --> 0:00:32.440 new season of Smart Talks with IBM. Join hosts from 0:00:32.479 --> 0:00:36.199 your favorite Pushkin podcasts as they talk with industry experts 0:00:36.200 --> 0:00:39.800 and leaders to explore how businesses can integrate AI into 0:00:39.840 --> 0:00:43.160 their workflows and help drive real change in this new 0:00:43.280 --> 0:00:46.960 era of AI, and of course, host Malcolm Gladwell will 0:00:46.960 --> 0:00:49.280 be there to guide you through the season and throw 0:00:49.360 --> 0:00:52.279 in his two cents as well. Look out for new 0:00:52.280 --> 0:00:55.160 episodes of Smart Talks with IBM every other week on 0:00:55.240 --> 0:00:59.440 the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts, 0:00:59.640 --> 0:01:03.880 and learn more at IBM dot com slash smart talks. 0:01:06.400 --> 0:01:10.280 Hello, Hello, Welcome to Smart Talks with IBM, a podcast 0:01:10.280 --> 0:01:16.880 from Pushkin Industri's iHeartRadio and IBM. I'm Malcolm Gladwell. This season, 0:01:16.920 --> 0:01:21.280 we're continuing our conversation with new creators visionaries who are 0:01:21.319 --> 0:01:24.840 creatively applying technology in business to drive change, but with 0:01:24.920 --> 0:01:29.840 a focus on the transformative power of artificial intelligence and 0:01:29.920 --> 0:01:33.040 what it means to leverage AI as a game changing 0:01:33.160 --> 0:01:38.160 multiplier for your business. Our guest today is Christina Montgomery, 0:01:38.560 --> 0:01:42.640 IBM's Chief Privacy and Trust Officer. She's also chair of 0:01:42.680 --> 0:01:48.120 IBM's AI Ethics Board. In addition to overseeing IBM's privacy policy, 0:01:48.480 --> 0:01:53.480 a core part of Christina's job involves AI governance, making 0:01:53.520 --> 0:01:57.320 sure the way AI is used complies with the international 0:01:57.640 --> 0:02:03.480 legal regulations customized for each industry. In today's episode, Christina 0:02:03.520 --> 0:02:08.440 will explain why businesses need foundational principles when it comes 0:02:08.520 --> 0:02:13.240 to using technology, why AI regulation should focus on specific 0:02:13.440 --> 0:02:17.720 use cases over the technology itself, and share a little 0:02:17.720 --> 0:02:23.360 bit about her landmark congressional testimony. Last May, Christina spoke 0:02:23.440 --> 0:02:27.080 with doctor Lori Santos, host of the Pushkin podcast The 0:02:27.120 --> 0:02:32.600 Happiness Lab, a cognitive scientist and psychology professor at Yale University. 0:02:32.919 --> 0:02:38.280 Laurie is an expert on human happiness and cognition. Okay, 0:02:38.800 --> 0:02:40.079 let's get to the interview. 0:02:42.639 --> 0:02:44.720 So Christina, I'm so excited to talk to you today. 0:02:44.880 --> 0:02:47.080 So let's start by talking a little bit about your 0:02:47.160 --> 0:02:49.840 role at IBM. What does a Chief Privacy and trust 0:02:49.880 --> 0:02:50.920 officer actually do. 0:02:51.440 --> 0:02:55.200 It's a really dynamic profession and it's not a new profession, 0:02:55.760 --> 0:02:58.640 but the role has really changed. I mean, my role 0:02:58.760 --> 0:03:02.480 today is broader and just helping to ensure compliance with 0:03:02.600 --> 0:03:07.240 data protection laws globally. I'm also responsible for AI governance 0:03:07.280 --> 0:03:10.239 I co chair or AI Ethics Board here at IBM, 0:03:10.560 --> 0:03:13.600 and for data clearance and data governance as well for 0:03:13.680 --> 0:03:17.240 the company. So I have both a compliance aspect to 0:03:17.280 --> 0:03:19.960 my role, really important on a global basis, but also 0:03:20.520 --> 0:03:25.440 help the business to competitively differentiate because really trust is 0:03:25.440 --> 0:03:29.160 a strategic advantage for IBM and a competitive differentiator as 0:03:29.160 --> 0:03:33.080 a company that's been responsibly managing the most sensitive data 0:03:33.120 --> 0:03:35.720 for our clients for more than a century now and 0:03:35.880 --> 0:03:38.600 helping to usher new technologies into the world with trust 0:03:38.600 --> 0:03:42.000 and transparency. And so that's also a key aspect of 0:03:42.040 --> 0:03:42.480 my role. 0:03:43.240 --> 0:03:45.440 And so you joined us here on smart Talks back 0:03:45.440 --> 0:03:47.680 in twenty twenty one and you chatted with us about 0:03:47.680 --> 0:03:51.360 IBM's approach of building trust and transparency with AI, and 0:03:51.520 --> 0:03:53.520 that was only two years ago, but it almost feels 0:03:53.560 --> 0:03:55.880 like in eternity has happened in the field of AI 0:03:56.040 --> 0:03:59.040 since then, and so I'm curious how much has changed 0:03:59.080 --> 0:04:01.560 since you were here last time. Were the things you 0:04:01.600 --> 0:04:04.440 told us before you are they still true? How are 0:04:04.480 --> 0:04:05.120 things changing? 0:04:05.520 --> 0:04:09.000 You're absolutely right. It feels like the world has changed 0:04:09.080 --> 0:04:12.760 really in the last two years. But the same fundamental 0:04:12.800 --> 0:04:17.720 principles and the same overall governance apply to IBM's program 0:04:18.279 --> 0:04:22.640 for Data Protection and Responsible AI that we talked about 0:04:22.680 --> 0:04:25.360 two years ago, and not much has changed there from 0:04:25.400 --> 0:04:28.560 our perspective. And the good thing is we've put these 0:04:28.600 --> 0:04:33.039 practices and this governance approach into place, and we have 0:04:33.279 --> 0:04:36.920 an established way of looking at these emerging technologies. As 0:04:36.960 --> 0:04:40.360 the technology evolves, the tech is more powerful, for sure, 0:04:40.520 --> 0:04:44.560 foundation models are vastly larger and more capable, and our 0:04:44.640 --> 0:04:48.080 creating in some respects new issues. But that just makes 0:04:48.080 --> 0:04:49.880 it all the more urgent to do what we've been 0:04:49.920 --> 0:04:53.520 doing and to put trust and transparency into place across 0:04:53.560 --> 0:04:56.200 the business to be accountable to those principles. 0:04:57.400 --> 0:04:59.560 And so our conversation today is really centered around this 0:04:59.680 --> 0:05:02.960 need for new AI regulation and part of that regulation 0:05:03.080 --> 0:05:05.960 involves the mitigation of bias. And this is something I 0:05:05.960 --> 0:05:08.320 think about a ton as a psychologist, right, you know, 0:05:08.360 --> 0:05:11.120 I know, like my students and everyone who's interacting with 0:05:11.200 --> 0:05:14.279 AI is assuming that the kind of knowledge that they're 0:05:14.320 --> 0:05:17.320 getting from this kind of learning is accurate, right, But 0:05:17.400 --> 0:05:19.440 of course AI is only as good as the knowledge 0:05:19.440 --> 0:05:21.720 that's going in. And so talk to me a little 0:05:21.760 --> 0:05:25.000 bit about like why bias occurs in AI and the 0:05:25.080 --> 0:05:26.960 level of the problem that we're really dealing with. 0:05:27.800 --> 0:05:30.840 Yeah, well, obviously AI is based on data, right, It's 0:05:31.200 --> 0:05:36.200 trained with data, and that data could be biased in 0:05:36.200 --> 0:05:38.680 and of itself, and that's where issues could come up. 0:05:38.720 --> 0:05:40.600 They come up in the data, they could also come 0:05:40.680 --> 0:05:44.200 up in the output of the models themselves. So it's 0:05:44.320 --> 0:05:49.080 really important that you build bias consideration and bias testing 0:05:49.200 --> 0:05:52.600 into your product development cycle. And so what we've been 0:05:52.640 --> 0:05:55.200 thinking about here at IBM and doing we had some 0:05:55.279 --> 0:05:58.239 of our research teams delivered some of the very first 0:05:58.240 --> 0:06:01.599 toolkits to help detect bias years ago now right, and 0:06:01.640 --> 0:06:05.280 deployed them to open source. And we have put into 0:06:05.320 --> 0:06:08.640 place for our developers here at IBM and Ethics by 0:06:08.680 --> 0:06:11.880 Design playbook that's sort of a step by step approach 0:06:12.640 --> 0:06:19.160 which also addresses very fully biased considerations, and we provide 0:06:19.200 --> 0:06:22.400 not only like here's a point when you should test 0:06:22.440 --> 0:06:24.840 for it, and you consider it in the data you 0:06:24.880 --> 0:06:26.599 have to measure it both at the data and the 0:06:26.640 --> 0:06:30.160 model level or the outcome level, and we provide guidance 0:06:30.160 --> 0:06:32.839 with respect to what tools can best be used to 0:06:32.880 --> 0:06:36.360 accomplish that. So it's a really important issue. It's one 0:06:36.400 --> 0:06:39.360 you can't just talk about. You have to provide essentially 0:06:39.480 --> 0:06:42.719 the technology and the capabilities and the guidance to enable 0:06:42.760 --> 0:06:43.920 people to test for it. 0:06:44.640 --> 0:06:47.320 Recently, you had this wonderful opportunity to head to Congress 0:06:47.360 --> 0:06:50.400 to talk about AI, and in your testimony before Congress, 0:06:50.440 --> 0:06:53.440 you mentioned that it's often said that innovation moves too 0:06:53.480 --> 0:06:56.080 fast for government to keep up, and this is something 0:06:56.080 --> 0:06:58.240 that I also worry about as a psychologist. Right our 0:06:58.279 --> 0:07:01.160 policy makers really understanding the issues that they're dealing with, 0:07:01.640 --> 0:07:03.880 and so I'm curious how you're approaching this challenge of 0:07:03.920 --> 0:07:06.360 adapting AI policies to keep up with the sort of 0:07:06.440 --> 0:07:08.960 rapid pace of all the advancements we're seeing in the 0:07:08.960 --> 0:07:10.320 AI technology itself. 0:07:11.280 --> 0:07:14.520 I think it's really critically important that you have foundational 0:07:14.560 --> 0:07:20.000 principles that applied to not only how you use technology, 0:07:20.040 --> 0:07:21.840 but whether you're going to use it in the first place, 0:07:21.840 --> 0:07:23.520 and where you're going to use and apply it across 0:07:23.600 --> 0:07:27.280 your company, and then your program from a governance perspective, 0:07:27.360 --> 0:07:29.880 has to be agile. It has to be able to 0:07:29.960 --> 0:07:35.680 address emerging capabilities, new training methods, etc. And part of 0:07:35.720 --> 0:07:40.240 that involves helping to educate and instill and empower a 0:07:40.280 --> 0:07:44.080 trustworthy culture at a company so you can spot those 0:07:44.120 --> 0:07:46.360 issues so you can ask the right questions at the 0:07:46.440 --> 0:07:49.640 right time if you try. We talked about during the 0:07:49.680 --> 0:07:53.920 Senate hearing, and IBM's been talking for years about regulating 0:07:54.200 --> 0:07:57.360 the use, not the technology itself, because if you try 0:07:57.360 --> 0:08:01.200 to regulate technology, you're very quickly going to find out 0:08:01.600 --> 0:08:04.560 regulation will absolutely never keep up with that. 0:08:05.040 --> 0:08:07.280 And so in your testimony to Congress, you also talked 0:08:07.280 --> 0:08:10.480 about this idea of a precision regulation approach for AI. 0:08:11.120 --> 0:08:13.320 Tell me more about this. What is a precision regulation 0:08:13.400 --> 0:08:15.400 approach and why could that be so important. 0:08:15.680 --> 0:08:18.640 It's funny because I was able to share with Congress 0:08:19.280 --> 0:08:23.080 our precision regulation point of view in twenty twenty three, 0:08:23.440 --> 0:08:26.240 but that precision regulation point of view was published by 0:08:26.360 --> 0:08:31.040 IBM in twenty twenty So we have not changed our 0:08:31.160 --> 0:08:36.079 position that you should apply the tightest controls, the strictest 0:08:36.120 --> 0:08:40.360 regulatory requirements to the technology where the end use and 0:08:40.480 --> 0:08:44.400 risk of societal harm is the greatest. So that's essentially 0:08:44.440 --> 0:08:47.440 what it is. There's lots of AI technology that's used 0:08:47.440 --> 0:08:51.080 today that doesn't touch people, that's very low risk in nature. 0:08:51.600 --> 0:08:55.040 And even when you think about AI that delivers a 0:08:55.160 --> 0:09:00.520 movie recommendation versus AI that is used to diagnose cancer, right, 0:09:00.640 --> 0:09:04.120 there's very different implications associated with those two uses of 0:09:04.160 --> 0:09:09.000 the technology. And so essentially what precision regulation is is 0:09:09.000 --> 0:09:12.400 apply different rules to different risks, right, more stringent regulation 0:09:12.679 --> 0:09:15.800 to the use cases with the greatest risk. And then 0:09:15.960 --> 0:09:20.120 also we build that out calling for things like transparency 0:09:20.720 --> 0:09:24.280 you see it today with content right, misinformation and the like. 0:09:24.880 --> 0:09:28.760 We believe that consumers should always know when they're interacting 0:09:28.840 --> 0:09:31.800 with an AI system, So be transparent, don't hydra AI 0:09:32.880 --> 0:09:36.480 clearly define the risks. So as a country, we need 0:09:36.520 --> 0:09:39.600 to have some clear guidance right in globally as well 0:09:39.920 --> 0:09:43.520 in terms of which uses of AI or higher risk 0:09:43.600 --> 0:09:47.920 where will apply higher and stricter regulation and have sort 0:09:47.960 --> 0:09:50.880 of a common understanding of what those high risk uses 0:09:50.920 --> 0:09:54.480 are and then demonstrate the impact in the cases of 0:09:54.559 --> 0:09:59.280 those higher risk uses. So companies who are using AI 0:09:59.440 --> 0:10:03.839 in space where they can impact people's legal rights, for example, 0:10:04.440 --> 0:10:08.680 should have to conduct an impact assessment that demonstrates that 0:10:08.760 --> 0:10:11.920 the technology isn't biased. So we've been pretty clear about 0:10:12.360 --> 0:10:15.720 apply that the most stringent regulation to the highest risk 0:10:15.800 --> 0:10:16.480 uses of AI. 0:10:18.200 --> 0:10:20.479 And so so far we've been talking about your congressional 0:10:20.480 --> 0:10:23.320 testimony in terms of, you know, the specific content that 0:10:23.360 --> 0:10:25.840 you talked about, But I'm just curious on a personal level, 0:10:26.160 --> 0:10:28.280 you know what was that like right Like, right now, 0:10:28.280 --> 0:10:30.559 it feels like at a policy level, like there's a 0:10:30.640 --> 0:10:33.000 kind of fever pitch going on with AI right now. 0:10:33.240 --> 0:10:34.679 You know, what did that feel like to kind of 0:10:34.760 --> 0:10:37.200 really have the opportunity to talk to policy makers and 0:10:37.240 --> 0:10:40.240 sort of influence what they're thinking about AI technologies like 0:10:40.320 --> 0:10:41.280 in the coming century. 0:10:41.320 --> 0:10:44.240 Perhaps I was really an honor to be able to 0:10:44.320 --> 0:10:47.280 do that and to be one of the first set 0:10:47.320 --> 0:10:51.400 of invitees to the first hearing. And what I learned 0:10:51.440 --> 0:10:54.600 from it essentially is, you know, really two things. The 0:10:54.640 --> 0:10:58.480 first is really the value of authenticity. So both as 0:10:58.679 --> 0:11:02.640 an individual and as a company, I was able to 0:11:02.679 --> 0:11:05.240 talk about what I do. You know, I need a 0:11:05.240 --> 0:11:08.680 lot of advanced prep right I talked about what my 0:11:08.800 --> 0:11:12.240 job is, what IBM has been putting in place for 0:11:12.360 --> 0:11:16.600 years now, so this isn't about creating something. This was 0:11:16.800 --> 0:11:19.240 just about showing up and being authentic. And we were 0:11:19.240 --> 0:11:22.360 invited for a reason. We were invited because we were 0:11:22.400 --> 0:11:26.120 one of the earliest companies in the AI technology space. 0:11:26.760 --> 0:11:31.920 We're the oldest technology company and we are trusted and 0:11:32.240 --> 0:11:34.640 that's an honor. And then the second thing I came 0:11:34.679 --> 0:11:37.400 away with was really how important this issue is to society. 0:11:37.440 --> 0:11:41.920 I don't think I appreciated it as much until following 0:11:42.040 --> 0:11:46.360 that experience. I had outreached from colleagues I hadn't worked 0:11:46.400 --> 0:11:49.280 with for years. I had an outreach from family members 0:11:49.280 --> 0:11:51.960 who heard me on the radio. You know, my mother 0:11:52.120 --> 0:11:55.000 and my mother in law, and my nieces and nephews 0:11:55.040 --> 0:11:57.439 and my friends of my kids were all like, oh, 0:11:57.440 --> 0:11:59.360 I get it. I get what you do. Now, Wow, 0:11:59.440 --> 0:12:02.840 that's pretty you know. So that was really probably the 0:12:02.960 --> 0:12:05.520 best and most impactful takeaway that I had. 0:12:06.000 --> 0:12:09.640 The mass adoption of generative AI, happening at breakneck speed, 0:12:10.000 --> 0:12:14.000 has spurred societies and governments around the world to get 0:12:14.120 --> 0:12:20.400 serious about regulating AI for businesses. Compliance is complex enough already, 0:12:20.600 --> 0:12:23.880 but throw an ever involving technology like AI into the mix, 0:12:24.320 --> 0:12:30.600 and compliance itself becomes an exercise in adaptability. As regulators 0:12:30.600 --> 0:12:35.160 seek greater accountability in how AI is used, businesses need 0:12:35.240 --> 0:12:40.160 help creating governance processes comprehensive enough to comply with the law, 0:12:40.400 --> 0:12:43.480 but agile enough to keep up with the rapid rate 0:12:43.520 --> 0:12:48.880 of change in AI development. Regulatory scrutiny isn't the only 0:12:48.960 --> 0:12:54.120 consideration either responsible AI governance. Of business's ability to prove 0:12:54.160 --> 0:12:59.000 its AI models are transparent and explainable is also key 0:12:59.040 --> 0:13:04.080 to building trust with customers, regardless of industry. In the 0:13:04.120 --> 0:13:08.319 next part of their conversation, Laurie asked Christina what businesses 0:13:08.360 --> 0:13:12.760 should consider when approaching AI governance. Let's listen. 0:13:13.600 --> 0:13:16.040 So it's a particular role that businesses are playing in 0:13:16.120 --> 0:13:18.920 AI governance, Like why is it so critical for businesses 0:13:18.960 --> 0:13:19.840 to be part of this? 0:13:20.480 --> 0:13:24.959 So, I think it's really critically important that businesses understand 0:13:25.480 --> 0:13:28.400 the impacts that technology can have, both in making them 0:13:28.400 --> 0:13:32.240 better businesses, but the impacts that those technologies can have 0:13:32.720 --> 0:13:37.839 on the consumers that they are supporting. You know, businesses 0:13:37.960 --> 0:13:42.560 need to be deploying AI technology that is in alignment 0:13:42.720 --> 0:13:44.560 with the goals that they set for it and that 0:13:44.720 --> 0:13:47.000 can be trusted, I think for us and for our 0:13:47.040 --> 0:13:50.839 clients A lot of this comes back to trust in tech. 0:13:51.280 --> 0:13:56.960 If you deploy something that doesn't work, that hallucinates, that discriminates, 0:13:57.440 --> 0:14:01.600 that isn't transparent, where decisions can't be explained, then you 0:14:01.720 --> 0:14:05.360 are going to very rapidly erode the trust at best 0:14:05.520 --> 0:14:08.839 right of your clients and at worst for yourself. You're 0:14:08.880 --> 0:14:11.560 going to create legal and regulatory issues for yourself as well. 0:14:11.600 --> 0:14:15.520 So trusted technology is really important, and I think there's 0:14:15.559 --> 0:14:17.760 a lot of pressure on businesses today to move very 0:14:17.840 --> 0:14:20.520 rapidly and adopt technology. But if you do it without 0:14:20.600 --> 0:14:24.080 having a program of governance in place, you're really risking 0:14:24.280 --> 0:14:25.240 eroding that trust. 0:14:25.360 --> 0:14:27.560 So this is really where I think a strong AI 0:14:27.640 --> 0:14:31.200 governance comes in. Talk about from your perspective, how this 0:14:31.280 --> 0:14:35.160 really contributes to maintaining the trust that customers and stakeholders 0:14:35.200 --> 0:14:36.360 have in these technologies. 0:14:37.080 --> 0:14:39.640 Yeah. Absolutely. I mean you need to have a governance 0:14:39.680 --> 0:14:43.680 program because you need to understand that the technology, particularly 0:14:43.720 --> 0:14:48.280 in the AI space, that you are deploying, is explainable. 0:14:48.320 --> 0:14:52.680 You need to understand why it's making decisions and recommendations 0:14:52.720 --> 0:14:54.200 that it's making, and you need to be able to 0:14:54.200 --> 0:14:56.600 explain that to your consumers. I mean, you can't do 0:14:56.680 --> 0:14:58.760 that if you don't know where your data is coming from, 0:14:58.760 --> 0:15:01.240 what data are you using to train those models, if 0:15:01.240 --> 0:15:05.040 you don't have a program that manages the alignment of 0:15:05.080 --> 0:15:08.640 your AI models over time to make sure as AI 0:15:08.880 --> 0:15:13.360 learns and evolves over uses, which is in large part 0:15:13.880 --> 0:15:17.680 what makes it so beneficial that it stays in alignment 0:15:17.720 --> 0:15:21.200 with the objectives that you set for the technology over time. 0:15:21.840 --> 0:15:25.680 So you can't do that without a robust governance process 0:15:25.680 --> 0:15:29.160 in place. So we work with clients to share our 0:15:29.200 --> 0:15:31.680 own story here at IBM in terms of how we 0:15:31.720 --> 0:15:35.320 put that in place, but also in our consulting practice 0:15:36.080 --> 0:15:40.600 to help clients work with these new generative capabilities and 0:15:40.600 --> 0:15:43.560 foundation models and the like in order to put them 0:15:43.560 --> 0:15:45.600 to work for their business in a way that's going 0:15:45.640 --> 0:15:48.280 to be impactful to that business but at the same 0:15:48.360 --> 0:15:49.440 time be trusted. 0:15:49.840 --> 0:15:51.720 So now I wanted to turn a little bit towards 0:15:51.720 --> 0:15:56.200 watsonex governance, and so IBM recently announced their AI platform, watsonex, 0:15:56.440 --> 0:15:59.600 which will include a governance component. Could you tell us 0:15:59.640 --> 0:16:01.960 a little more about watsonex dot governance. 0:16:02.320 --> 0:16:04.560 Yeah, I mean before I do that, I'll just back 0:16:04.640 --> 0:16:08.240 up and talk about the full platform and then lean 0:16:08.280 --> 0:16:11.160 into Watson X because I think it's important to understand 0:16:11.440 --> 0:16:17.360 the delivery of a full suite of capabilities, to get data, 0:16:17.440 --> 0:16:20.360 to train models, and then to govern them over their 0:16:20.400 --> 0:16:25.920 life cycle. All of these things are really important from 0:16:25.960 --> 0:16:28.800 the onset you need to make sure that you have, 0:16:29.440 --> 0:16:33.680 you know, for our watsonex dot AI for example, that's 0:16:33.720 --> 0:16:38.280 the studio to train new foundation models and generative AI 0:16:38.360 --> 0:16:44.000 and machine learning capabilities, and we are populating that studio 0:16:44.360 --> 0:16:49.640 with some IBM trained foundation models which we're curating and 0:16:49.720 --> 0:16:53.480 tailoring more specifically for enterprises. So that's really important. It 0:16:53.520 --> 0:16:55.880 comes back to the point I made earlier about business 0:16:55.880 --> 0:17:00.200 trust and the need you know to have enterprise is 0:17:00.280 --> 0:17:05.600 ready technologies in the AI space. And then the watsonex 0:17:05.600 --> 0:17:08.800 dot data is a fit for purpose data store or 0:17:08.840 --> 0:17:12.560 a data lake, and then watsonex dot gov. So that's 0:17:12.960 --> 0:17:17.600 a particular component of the platform that my team and 0:17:17.680 --> 0:17:21.080 the AI Ethics Board has really worked closely with the 0:17:21.119 --> 0:17:24.680 product team on developing, and we're using it internally here 0:17:24.680 --> 0:17:27.880 in the Chief Privacy Office as well to help us 0:17:28.320 --> 0:17:33.240 govern our own uses of AI technology and our compliance 0:17:33.280 --> 0:17:39.359 program here. And it essentially helps to notify you if 0:17:39.400 --> 0:17:42.520 a model becomes biased or gets out of alignment as 0:17:42.560 --> 0:17:45.440 you're using it over time. So companies are going to 0:17:45.520 --> 0:17:48.040 need these capabilities. I mean they need them today to 0:17:48.119 --> 0:17:53.080 deliver technologies with trust. They'll need them tomorrow to comply 0:17:53.200 --> 0:17:55.280 with regulation, which is on the horizon. 0:17:55.520 --> 0:17:58.399 And I think compliance becomes even more complex when you 0:17:58.440 --> 0:18:02.560 consider international data protection laws and regulations. Honestly, I don't 0:18:02.600 --> 0:18:05.160 know how anyone on any company legal team is keeping 0:18:05.240 --> 0:18:07.480 up with this these days. But my question for you 0:18:07.600 --> 0:18:11.200 is really how can businesses develop a strategy to maintain 0:18:11.240 --> 0:18:14.040 compliance and to deal with it in this ever changing landscape. 0:18:14.119 --> 0:18:18.040 It's increasingly more challenging. In fact, I saw statistic just 0:18:18.119 --> 0:18:22.720 this morning that the regulatory obligations on companies have increased 0:18:22.720 --> 0:18:26.600 something like seven hundred times in the last twenty years. 0:18:26.400 --> 0:18:31.000 So it really is a huge focus area for companies. 0:18:31.160 --> 0:18:34.040 You have to have a process in place in order 0:18:34.080 --> 0:18:36.600 to do that, and it's not easy, particularly for a 0:18:36.680 --> 0:18:40.560 company like IBM that it has a presence in over 0:18:40.560 --> 0:18:43.600 one hundred and seventy countries around the world. There is 0:18:43.640 --> 0:18:48.520 more than one hundred and fifty comprehensive privacy regulations. There 0:18:48.560 --> 0:18:53.080 are regulations of non personal data, there are AI regulations emerging, 0:18:54.040 --> 0:18:58.119 so you really need an operational approach to it in 0:18:58.240 --> 0:19:00.280 order to stay compliant. But one of the things we 0:19:00.320 --> 0:19:02.480 do as we set a baseline, and a lot of 0:19:02.480 --> 0:19:05.919 companies do this as well, So we define a privacy baseline, 0:19:05.960 --> 0:19:10.760 we define an AI baseline, and we ensure then as 0:19:10.800 --> 0:19:13.200 a result of that that there are very few deviances 0:19:13.280 --> 0:19:16.359 because it incorporates in that baseline. So that's one of 0:19:16.400 --> 0:19:18.560 the ways we do it. Other companies, I think are 0:19:18.640 --> 0:19:23.840 similarly situated in terms of doing that. But again, it 0:19:24.000 --> 0:19:26.679 is a real challenge for global companies. It's one of 0:19:26.680 --> 0:19:30.640 the reasons why we advocate for as much alignment as 0:19:30.720 --> 0:19:36.080 possible on the international realm as well as nationally here 0:19:36.119 --> 0:19:39.920 in the US, as much alignment as possible to make 0:19:40.119 --> 0:19:45.080 compliance easier for easier, and not just because companies want 0:19:45.119 --> 0:19:48.280 an easy way to comply, but the harder it is, 0:19:48.520 --> 0:19:52.399 the less likely there will be compliance. And it's not 0:19:52.520 --> 0:19:58.600 the objective of anybody, governments, companies, consumers to have to 0:19:58.760 --> 0:20:02.280 set legal obligation that companies simply can't meet. 0:20:02.520 --> 0:20:04.560 So what advice would you give to other companies who 0:20:04.560 --> 0:20:07.640 are looking to rethink or strengthen their approach to AI. Governance. 0:20:07.720 --> 0:20:11.040 Think you need to start with, as we did, foundational principles, 0:20:11.680 --> 0:20:15.119 and you need to start making decisions about what technology 0:20:15.119 --> 0:20:17.480 you're going to deploy and what technology you're not, What 0:20:17.480 --> 0:20:18.600 are you going to use it for, and what aren't 0:20:18.600 --> 0:20:20.000 you going to use it for? And then when you 0:20:20.040 --> 0:20:24.400 do use it, align to those principles. That's really important. 0:20:24.480 --> 0:20:29.000 Formalize a program, have someone within the organization, whether it's 0:20:29.040 --> 0:20:33.320 the chief privacy officer, whether it's some other role, a 0:20:33.400 --> 0:20:37.840 chief AI ethics officer, but have an accountable individual and 0:20:37.920 --> 0:20:42.080 accountable organization. Do a maturity assessment, figure out where you 0:20:42.160 --> 0:20:44.919 are and where you need to be, and really start 0:20:46.000 --> 0:20:50.480 putting it into place today. Don't wait for regulation to 0:20:50.560 --> 0:20:53.200 apply directly to your business, because it'll be too late. 0:20:54.440 --> 0:20:57.840 So Smart Talks features new creators, these visionaries like yourself 0:20:57.840 --> 0:21:01.320 who are creatively applying technology in business to drive change. 0:21:01.640 --> 0:21:04.320 I'm curious if you see yourself as creative. 0:21:04.480 --> 0:21:07.800 You know, I definitely do. I mean you need to 0:21:07.880 --> 0:21:12.520 be creative when you're working in an industry that evolves 0:21:12.560 --> 0:21:17.040 so very quickly. So you know, I started with IBM 0:21:17.320 --> 0:21:20.440 when we were primarily a hardware company, right, and we've 0:21:20.520 --> 0:21:23.960 changed our business so significantly over the years and the 0:21:24.040 --> 0:21:28.480 issues that are raised with respect to each new technology, 0:21:28.520 --> 0:21:32.160 whether it be cloud, whether it be AI now where 0:21:32.160 --> 0:21:33.760 we're seeing a ton of issues, or you look at 0:21:33.760 --> 0:21:37.919 emergent issues in the space of things like neurotechnologies and 0:21:38.000 --> 0:21:44.600 quantum computers. You have to be strategic and you have 0:21:44.720 --> 0:21:47.960 to be creative and thinking about how you can adapt 0:21:48.600 --> 0:21:53.840 agilely quickly a company to an environment that is changing 0:21:53.880 --> 0:21:56.040 so quickly and with. 0:21:56.080 --> 0:21:59.000 This transformation happening at such a rapid pace. Do you 0:21:59.040 --> 0:22:01.320 think creativity is a role in how you think about 0:22:01.359 --> 0:22:05.520 and implement specifically a trustworthy AI strategy. 0:22:06.600 --> 0:22:10.640 Yeah, I absolutely think it does, because again it comes 0:22:10.680 --> 0:22:13.840 back to these capabilities, and there are ways, I guess 0:22:14.040 --> 0:22:17.800 how you define creativity could be different, right, but I'm 0:22:17.840 --> 0:22:21.040 thinking of creativity in the sense of sort of agility 0:22:21.080 --> 0:22:25.160 and strategic vision and creative problem solving. I think that's 0:22:25.440 --> 0:22:28.560 really important in the world that we're in right now, 0:22:28.600 --> 0:22:33.159 being able to creatively problem solve with new issues that 0:22:33.240 --> 0:22:36.320 are rising sort of every day. 0:22:36.960 --> 0:22:38.520 And so, how do you see the role of chief 0:22:38.560 --> 0:22:42.200 privacy officer evolving in the future as AI technology continues 0:22:42.240 --> 0:22:45.120 to advance, Like what stuff should CPOs take to stay 0:22:45.119 --> 0:22:47.040 ahead of all these changes that are come in their way. 0:22:48.320 --> 0:22:52.240 So the role is evolving in most companies, I would 0:22:52.280 --> 0:22:57.240 say pretty rapidly. Many companies are looking to chief privacy 0:22:57.280 --> 0:23:00.800 officers who are ready understand the data that's being used 0:23:00.800 --> 0:23:04.320 in the organization and have programs to ensure compliance with 0:23:04.440 --> 0:23:08.240 laws that require you to manage that data in accordance 0:23:08.240 --> 0:23:10.880 with data protection laws and the like. It's a natural 0:23:10.920 --> 0:23:16.920 place and position for AI responsibility, and so I think 0:23:16.920 --> 0:23:19.600 what's happening to a lot of chief privacy officers is 0:23:19.640 --> 0:23:23.119 they're being asked to take on this AI governance responsibility 0:23:23.160 --> 0:23:26.680 for companies and if not take it on, at least 0:23:26.720 --> 0:23:30.119 play a very key role working with other parts of 0:23:30.160 --> 0:23:33.800 the business in AI governance. So that really is changing. 0:23:34.040 --> 0:23:38.040 And if chief privacy officers are in companies who maybe 0:23:38.119 --> 0:23:41.879 haven't started thinking about AI yet, they should, so I 0:23:41.920 --> 0:23:45.760 would encourage them to look at different resources that are 0:23:45.800 --> 0:23:50.159 available already in AI governance space. For example, the International 0:23:50.200 --> 0:23:54.640 Association of Privacy Professionals, which is the seventy five thousand 0:23:54.680 --> 0:23:59.760 member professional body for the profession of Chief Privacy officers, 0:23:59.840 --> 0:24:04.520 just recently launched an AI Governance Initiative and an AI 0:24:04.560 --> 0:24:08.520 Governance Certification program. I sit on their advisory board. But 0:24:08.640 --> 0:24:11.320 that's just emblematic of the fact that the field is 0:24:11.400 --> 0:24:12.840