Ep27. Krista Sande-Kerback: IBM & A.I.’s Promise

[00:00:00] Krista Sande-Kerback: We recently had some tech leaders call for the six month pause in the training of more powerful AI systems to allow for the creation of new ethics standards. And, you know, there’s good intentions there, but it missed a fundamental point, which is that these systems are already in our control today, as are the solutions.

With responsible training, together with an ethics by design approach, there’s a different way. You build that over the whole AI pipeline, you support it by multi stakeholder, diverse collaboration around AI. And that can make these systems better, not worse. AI is going to continue to evolve. Yes. There’s going to continue to be regulations.

No one is immune businesses of small and large, but there’s systems in use today. There’s systems coming online tomorrow. Training has to be a part of that responsible approach. We don’t need a pause. Prioritize responsible A. I. Because we have these frameworks.

[00:00:52] Helen Todd: Krista Sande-Kerback is marketing leader for open pages, which is IBM’s platform for governance, risk and compliance.

She’s working on the upcoming major launch of WatsonX.governance. Watson X is IBM’s recently announced generative AI platform that comes with a suite of tools for tuning large language models, a data store built on Lake House architecture and an AI governance tool kit. Aimed at mitigating risk associated with AI and protecting customers privacy.

Krista is a strategic adviser and marketing leader who has spent her career building and scaling marketing and transformation programs. I am providing critical insights to senior executives. She previously supported IBM’s acquisition of Brazilian RPA provider, WDG Automation in 2020, and has conducted market intelligence research on the latest technology trends, trained dozens of teams in agile methodologies, and scaled a startup.

Krista is an alumna of Dartmouth College and Columbia Business School. A former Fulbright scholar to Germany, she serves on the board of directors for the Fulbright Association’s New York chapter. She is also a council officer for the women in America, professional development and mentoring organization, where she is focused on increasing the proportion of women in the C-suite boardrooms and other prominent leadership roles.

Kristen, I met back in 2018 in New York City, and I was immediately impressed by her passion for supporting and lifting women up. You’ll hear more about this in our conversation as one of her goals is to demystify generative AI for women.
In today’s episode. You’ll discover how AI can help close the gender gap. In addition to IBM’s history with AI, environmentally responsible AI, the need for explainable AI, IBM skills, build program, and IBM’s industry leading efforts when it comes to AI ethics and governance. You’ll also hear what makes GenAI a game changer and why creativity may be the ultimate moonshot for AI.

Plus, why businesses of all sizes should be investing in artificial intelligence now. Krista shares that she knows that sometimes it feels like AI is promising the moon and the stars, but how do you actually make sense of all of it? Tune in to today’s episode to discover how. Enjoy.

Welcome to Creativity Squared. Discover how creatives are collaborating with artificial intelligence in your inbox, on YouTube, and on your preferred podcast platform. Hi, I’m Helen Todd, your host, and I’m so excited to have you join the weekly conversations I’m having with amazing pioneers in the space.

The intention of these conversations is to ignite our collective imagination at the intersection of AI and creativity to envision a world where artists thrive.

Helen Todd: Krista, it’s so good to have you on the show. Welcome to Creativity Squared.

[00:04:05] Krista Sande-Kerback: Helen. It’s so great to see you and a pleasure to be here today.

[00:04:08] Helen Todd: Yeah. Oh, Krista and I met through a mutual friend, both Krista and her friend are Fulbright scholars, and ever since we got coffee in New York City, we’ve been fast friends and you’ll learn a whole lot about Krista.

She works with IBM and in the governance, which I’m super excited about. And we’ll get a lot more into that. Today, but before we dive in, can you share with our viewers and listeners who are meeting you for the first time, who you are, what you do and your origin story?

[00:04:33] Krista Sande-Kerback: Well, you have an AI clone, but I have an identical twin.

[00:04:39] Helen Todd: I love that.

[00:04:39] Krista Sande-Kerback: And I studied geography, public policy, and German at Dartmouth college. So my path to a tech career was definitely not an obvious one, but it’s had some unifying themes. After college I moved to Germany and as you mentioned I did a Fulbright grant. And for quick context, that’s a competitive scholarship program run by the U.S. government with the goal to improve intercultural relations. Just, I was so fortunate to have had this opportunity. It’s still one of the best things that I’ve ever done. Then, I returned to New York City and I worked in management consulting for a few years. And among other projects, I did research and analysis on outsourcing and offshoring trends.
So, now that I’ve worked with robotic process automation and AI, there’s a common theme of business productivity through some of my work. From there, I went to work for a professional women’s network called 85 Broads, where I oversaw the modernization of the website, ran live and virtual events. And traveled around the world to help grow the brand to a 30,000 member organization.

It was very exciting. And I’ve always really loved supporting professional women in achieving their dreams. From there, I went to Columbia business school to take stock, get my MBA and having worked with a lot of women entrepreneurs, I was really keen to further my experience in technology marketing and strategy.

And at this point I’ve been at IBM for nine years and each opportunity has been better than the last. And I don’t regret that geography major because we need that breadth of experience among those of us working in AI. And so liberal arts majors can join us.

[00:06:12] Helen Todd: I love how you found the through line through all of your experiences.

And one thing that Kristen and I both share is a passion for helping women. Especially we were both at conferences yesterday that were all-female conferences. So can you share a little bit more about your passion and Why it’s so important to you to demystify AI for women.

[00:06:33] Krista Sande-Kerback: Absolutely. Empowering and connecting women has long been a passion of mine. 85 broads, which is now called elevate network is still an organization I’m deeply involved in, and I actually run IBM’s partnership with them. And I also sit on the alumni board of Women in America, which is a mentoring organization that’s empowering women to achieve positions of influence in business.

I want to educate as many women as I possibly can on generative AI. There’s a problem with too much homogeneity among the core group of people that have been developing AI to date, and AI needs to reflect the communities in which we serve. I also believe that if used correctly, AI has the power to close the gender gap.

And we know we have a real need to do that and a big opportunity. But coming from a place of humility, because I’m learning so much every day, really, I first want to have a conversation with both men and women. I want to learn about their pain points. I want to learn about their breakthroughs with AI as they’re getting up the learning curve and helped connect the dots.

As best as I can, every time I speak to a new community, I get the benefit of their perspectives. And if I can’t answer the questions, it motivates me to do the research. I can’t do my job well if I never leave the office, but I have to say it’s been exciting so far. And again, engaging with these communities is really the best part of it.

[00:07:53] Helen Todd: It’s so good to hear you say that. And knowing that you’re at IBM and helping. Amplify that message of diversity in all facets of these tools. One reason why I created the podcast is to force me to learn about AI too. So I definitely come from a place of humility as well. And you get to sit in such an interesting seat at IBM and there’s so many cool projects.

So what are you most excited about right now when it comes to creativity and AI?

[00:08:22] Krista Sande-Kerback: Creativity is really the must have generative AI skill. We like to talk about moonshots at IBM. And if you’ve been around for a while you’ve kind of heard this theme over the years. Creativity may be the ultimate moonshot for artificial intelligence.

And I think it shows how far we’ve come that we’re even talking about AI in these kinds of terms. I’m going to quote John Smith, who’s the manager of multimedia and vision at IBM research. He said, it’s easy for AI to come up with something novel just randomly, but it’s very hard to come up with something that’s novel and unexpected and also useful.
So we really still need people. The whole creative industry from film to advertising and marketing, my field is using these tools to test out new ideas and accelerate prototypes. But in spite of this. AI is never going to replace that human soul of creativity. Instead, we should be thinking about it as a smart efficient and inspirational assistant.

[00:09:19] Helen Todd: I love the John Smith quote about creativity and how you all define it, because it’s come up on the show a lot. How do you define creativity? So we’ll see if through that lens of novel, unexpected and useful if AI can meet that that benchmark, but IBM has been in the AI space for a long time and has done a lot of cool projects.
So can you give us a quick kind of cliff notes of just kind of the history of IBM and some of the cool projects specifically related to AI?

[00:09:48] Krista Sande-Kerback: I’d be happy to. Artificial intelligence, as I think we all know, is not a new concept. The birth of the AI conversation was denoted by Alan Turing, who’s often referred to as the father of computer science.
In a paper that he published in 1950 called Computing Machinery and Intelligence, he asks the question, can machines think? And we’ve probably heard of the Turing test, which challenges a human interrogator to try to distinguish an interaction between a computer and a human in a text response. So, while this test has undergone some scrutiny, it remains an important part of that early history of AI.

So that’s 1950. Going to 1956, John McCarthy coined the term artificial intelligence at the first ever AI conference at my alma mater, Dartmouth College. muCh later in 2004, he wrote a paper called, What is Artificial Intelligence? Where he offers a definition of AI as the science and engineering of making intelligent machines, especially intelligent computer programs.

And it’s related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable. Continuing on the line, 1967, Frank Rosenblatt builds something called the Mark I Perceptron, which is the first AI based on a neural network that learned through trial and error.

Down the road, by the 1980s, neural networks had become widely used in AI applications. 1997, big news for IBM, our deep blue supercomputer beats the world chess champion, Gary Kasparov, in a chess match and a rematch. And then just over a decade later, IBM Watson competed and beat Ken Jennings and Brad Rutter at Jeopardy in 2011.

So this was really our public demonstration of AI when no one was really doing it. Couple other more recent developments. In 2015, Baidu’s. Meanwhile, computer uses a special kind of deep neural network called a convolutional neural network and it their Alpha Go program beat the World, champion Go Player in a five game match.

Why this is important is that go is considered the most difficult game ever invented, and the number of possible moves as the game progresses is over 14.5 trillion just after four moves. Then, of course, 2022, I think everybody’s familiar with the Open AI Research Lab’s release of ChatGPT, which is, of course, the first public prototype of a generative AI system family.

So that’s a very exciting, big development. You know, everybody’s been talking about that. And then IBM has also, as I said, been working with traditional AI for a long time. But this year, we also announced a number of innovations to help businesses with generative AI. And it feels like every week there’s you know, there’s new news from you know, the other companies that we work without there in this space.

I just want to say one more thing, which is that the distinction between consumer and enterprise AI is important because ChatGPT has a lot of great consumer applications. Certainly, it’s lower risk for a lot of things. Many people are out they’re using it already. The current AI landscape also provides an opportunity for businesses to achieve significant breakthroughs.

With AI for Business, you want to think about it in the context of creating competitive edge, scaling it, and advancing trustworthy AI. And this is where we, as IBM, are focused within this landscape.

[00:13:09] Helen Todd: I love that background in the history. So thank you for sharing it. And I’m sure a lot of people who are listening and tuning in will remember the jeopardy part and the a recent interview that I’ll be sure to link to in the notes is with Ksenia who runs Turing Post and does a lot of history about Alan Turing as well.

So definitely want to invite all of our listeners. To listen to that show as well. And I know I’ve mentioned AlphaGo on the show before. It’s one of my favorite documentaries on AI and really goes into that game where the program beats Lee Sedol. There’s like my favorite scene in it. It’s kind of a little bit of a spoiler.

But Lee goes out to smoke a cigarette and the person who places the, I don’t know if it’s a marble or what it’s called on the board. And Lee comes back and has this like shocked face and on the back end, it’s like one and some million billion or whatever large number that anyone would select and he selects.

The move that beats the game. And then in the interview afterwards of, you know, like the odds against that selection were so, so minuscule. And when asked, like, how did you pick that move? He answered that’s the only move I saw and they call it the the God move, and it’s such like a great scene in that show.

Yeah, I was thinking actually before this, like, I should do like a film plus like a film club and all the films and sci-fi and documentaries and then talk about them too, for anyone who’s interested, let me know and we can get that started. But in terms of IBM has done more recently, I know there was something with the US open.

And so can you share some of the, some recent cool things that IBM has done as well.

[00:14:55] Krista Sande-Kerback: Yes, we’re working on all sorts of cool things. The U. S. Open is is one that I’m particularly excited about as a tennis fan. And the application there is AI for commentary, one of the applications. The digital experience of the US. Open is, of course, of enormous importance to our global fans. So IBM Consulting worked closely with the US Tennis Association to develop generative AI models that transform tennis data into insights and original content on the U. S. Open app and website. IBM Watson X, which is a next generation AI and data platform, builds and manages the entire lifecycle of the AI models that produce key app features such as match insights in the new AI commentary for US. Open highlight reels. The team then use what’s next data to curate connect and curate the USTA’s trusted data sources. This involves some de duping and filtering of the foundational data. And the process ultimately eliminates things like profanity and abusive language, which is super important.

And it manages data using compliance with privacy regulations like the General Data Protection Regulation or GDPR. There’s a lot more to it, but it, but, a couple more details are that the AI Commentary models trained to translate the metadata attached to video clips and generates dozens of different options before choosing the best sentence to describe the action and ultimately the operation of the model is monitored and managed using elements of what’s the next step governance.

Which, which ensures the AI is performed, compliant, and operating as expected. So that’s just one of the many the fun applications of IBM AI Technologies today.

[00:16:34] Helen Todd: I love that. And we’ve seen a bunch of examples. If you don’t have the guardrails with profanity and whatnot what can happen?

So, building that in the design is always wonderful to hear, and we’ll get to ethics by design in a little bit, but one thing that I’m really excited to talk to you about is environmentally responsible AI and this is something that we haven’t talked about on the show before. But I was talking to Paul Croner, who did the first AI art exhibit here in Cincinnati.

And when I was telling him about my clone, the first thing that he brought up was, well, what’s the environmental impact of all of this this new tech has environmental implications, and it’s so good to hear that IBM is thinking about that. So can you tell us what your initiatives are on this front?

[00:17:22] Krista Sande-Kerback: It’s such an important question, and it’s not one that I think is something that we might naturally think about as we’re playing with chat, GBT, your consumer applications. But, of course, I contributes to sustainability because there’s so much that we can do to it. make a difference. Managing buildings, utilities, robotic process automation that’s enabled by a I that can eliminate paper waste and automate all sorts of tasks supply chains.

There’s so much that it can do to enhance productivity. And we have a lot of this a I powered software that makes a huge difference. But on the flip side, It’s true. It’s I attended this conference that I attended this week featured Kate Crawford, who’s a Microsoft researcher, and she talked about how doing generative AI at scale could be 5 to 50 times more energy intensive than traditional models.

AI actually uses an enormous amount of water, and we really have to look at the whole supply chain with that. So I’m happy to say that IBM is doing a lot to address that. As I said, we have many sustainability solutions in place. It’s a very important very important for us strategically. But one of the things that we’re doing, for instance, is prototyping a brain like chip that could make AI more energy efficient.

Concerns have been raised about emissions associated with warehouses full of computers powering AI systems and our prototype could lead to more efficient, less battery draining AI chips for smartphones. So there’s a lot of research in this area. It’s really important. It’s good that people are asking the question and really thinking about this.

[00:18:57] Helen Todd: Yeah. Well, well, thank you for sharing that. Cause AI is here. You know, with the onset of chatGPT, it’s really captured, you know, the public imagination, but you know, as you so clearly and beautifully said about the history of AI with IBM and it’s only accelerating. And I know you do a lot of presentations on the enterprise front.

So can you share with us kind of the, from the seat that you sit, some of the it. Are the trajectory of where we’re going with.

[00:19:29] Krista Sande-Kerback: Absolutely. AI is no longer a business experiment. It’s for businesses. It’s becoming an integral part of strategy, large and small. Here are a few stats that I think are exciting.

More than 80 percent of enterprises are working with or planning to leverage foundation models and adopt generative AI. This comes from a report from Scale Zeitgeist 2023 AI Readiness Report, and they shared that with the companies that they interviewed, 21 percent have AI models in production, about 29 percent are experimenting with generative AI, and another 31 percent are planning to work with generative AI models.

So that’s huge. That’s really a paradigm shift and according to Goldman Sachs, another stat that I like, generative AI has the potential to raise global GDP by 7 percent in 10 years.

[00:20:19] Helen Todd: That’s amazing. And you mentioned foundational models, and that might be a new term for a few of our listeners. So can you do very kind of high level, what’s the difference or how foundational models enable the new genAI just for our listeners who may be hearing that term for the first time?

[00:20:37] Krista Sande-Kerback: Absolutely. And as I said, it’s a paradigm shift. So I’ll try to break this down a little bit. Traditional AI requires individual siloed models. So for every task, a particular AI model is typically built on a set of labeled data. And then there’s task specific training for each model. This training is not generally transferable without a lot of retraining with a completely different set of data, and consequently can be very time consuming when there’s a lot of tasks, even if they’re somewhat similar.

So traditional AI requires lots of human effort and cost just on things like labeling the data. And given that an enterprise will require hundreds of thousands of automated tests, this is very expensive and often why a I adoption stalls from incubation to production. You can see that it being implemented in some discrete parts of the business.
But how do you bring that all together? Unlike traditional AI, foundational models don’t begin with labeled data, but they’re pre trained on massive amounts of data. Specifically, they’re multitasking models that can be applied to many different tasks. They’re adaptable with little or no training, and they are pre trained and require no supervision.
Now, back to the environmental side of things. Yes, they can require a lot of graphical processing units or GPUs and memory. They require some energy, but they don’t require the same amount of large and costly human capital to come up with specific labeled data for tasks. Going into generative AI, it refers to a set of AI algorithms that can generate new outputs, such as text, images, code, or audio based on the training data, unlike traditional AI systems that are designed to recognize patterns and make predictions.

Sometimes the AI that powers these solutions are referred to as decoders. IBM has been investing in foundation models and generative AI is one of the ways to bring these models to life. So I hope that just kind of, sums that up quickly for you.

[00:22:27] Helen Todd: Yeah, that was perfect. And it’s just so fascinating to me that why, you know, they’re so popular right now is the just the natural language and how for all the tech and how technical the back end is that enables it, that Why it’s so popular is that people can interact with it in such a natural language way.

But one thing too, that I love about how you think about AI on the enterprise level is the AI plus ladder. So can you share with us for our business listeners how you think about adding AI into their business businesses and that. Ladder.
[00:23:03] Krista Sande-Kerback: Absolutely. AI has really gone from a world where companies are thinking about running their business plus doing some AI to help it to a world where AI Is first to help the business. The leading companies for the next decade or two are going to be the ones that decide that they will be AI first. And this. Decision will dictate so many things from how they operate, how they work with employees and work with their customers and suppliers.
The AI plus ladder is a term that IBM used a few years ago, and at the time it was primarily focused on data and data life cycles. The plus AI model is It’s then how clients collect data, organize, and grow their data. And that creates the bottom rung, starting with that good, well organized, collected data.

Today, the latter is about how clients add AI to their applications, automate their workflows, replace their workflows, and get to the ultimate point that AI does the work. And this is what AI First means.

[00:23:57] Helen Todd: I remember being at a conference. This was pre pandemic. It was one hosted by the information which is a fabulous subscription tech publication where they said AI is going to be a layer in every business and help with decision making.

And one thing I find fascinating with what you said is eventually AI will be doing the work. So can you kind of give a few examples of like the type of work that AI will actually be doing? At that level of the ladder.

[00:24:26] Krista Sande-Kerback: Yes, absolutely. With AI employee productivity is expected to be the primary driver of economic value, and I’ll caveat that any of this is my point of view, although, of course, our CEO Arvind Krishna regularly talks about this.

AI can definitely help remove repetitive tasks to boost productivity and growth and also keep people safe. So a few examples that I really like are within HR and talent management. During the talent acquisition process, you can use AI to automate all sorts of repetitive tasks. Think about the onboarding emails that you send to an employee or the process of recruiting.

And then, you know, you’re trying to bring in a candidate. You free up time for humans to. to deliver more white glove treatment to candidates. Another example within the finance function, AI can be performed to all of the transactional work with increased speed and accuracy. It has this ability that we’ve discussed to learn from large data sets, and then it can really improve accuracy in areas such as budgeting and forecasting, and this really enhances company wide decision making.

Then, finally, here’s what I’m excited about, is that There are so many jobs that require people to be unsafe for instance lifting heavy things within a factory setting or working someplace where there’s potentially unsafe chemicals or unstable conditions. You can now use AI for instance, with our maximal visual solution on drones.

People can become drone inspectors rather than having to, you know, go into the field. They’re no longer in dangerous situations. And these are all a win for my book.

[00:25:56] Helen Todd: Yeah, that’s very cool. And I think just kind of punctuates the point of AI not replacing people, but The jobs evolving and an example of what that looks like. So, I love that.

[00:26:11] Krista Sande-Kerback: It’s absolutely true. I mean, I’m not gonna pretend that like the previous discourse around robots taking our jobs has completely gone away But it’s way more nuanced than that. If you look at across history There have always been fears about the latest waves of technologies replacing workers, but we’re still here This wave is interesting though, because it’s gonna affect larger portions of white collar jobs, but there’s also a need for this I mean There, the need for boosting productivity is growing quickly because there’s a future of work study by the Global Talent Crunch that estimated that a global talent crisis could cost nations to the tune of 8.

5 trillion in unrealized annual revenues with about 85 million jobs unfilled by 2030 due to lack of skilled workers. And for reference point, that’s equivalent to the GDP combined of Germany and Japan. So there’s actually a huge need for this. But I also think that it just makes for a more rewarding experience at work.

[00:27:04] Helen Todd:Yeah, well, I know I was mentoring a high schooler the other day, and he actually asked, you know, what he should be thinking about for a career. And it was right after we had talked and I shared that there’s going to be this huge gap in skills needed. And also when I went to, or when we were in college, social media marketing was not an industry that did exist.

So I went to school and. You know, the past 14 years of my life have been with the industry that was a there when I was in college. So there’s going to be a whole new slew of jobs and job titles and industries that are popping up. Well, when one thing I want to ask, and this came up. In I’m actually going to read the question because I moderated a panel yesterday and all female panel on AI here in Cincinnati, which was really great because we hear about productivity and efficiencies all the time, but I’m going to read the question that I asked on stage.

So the greatest economist of the 20th century john Maynard predicted the biggest challenge of the 21st century would be how we’d spend. all of our leisure time that the more productive we become, the more leisure time we’d have. In fact, the Jetsons predicted or have a two hour work days in the future.

And it seems like in a lot of ways, the Jetson sci fi is coming into reality, but the industrial revolution proved to be the opposite of this. The information age is no different, especially, you know, in the States, we work during our vacations, even if we take, you know, That is, if we even take vacations and now we’re in the era of AI how do we ensure more productivity is balanced with human well being and just not expecting more and more in less time. So I’d love to get your thoughts on this, too.

[00:28:54] Krista Sande-Kerback: I’m a strong believer in workplace well being and. I believe that AI frees up more time for strategic thinking, and we need to take that, and we can harness this.

I think back to when I was a very young management consultant, the number of extremely late nights that I pulled trying to get an Excel model, and the formula is just right. We’re trying to get a PowerPoint presentation ready for the client, and I was there manipulating things. It’s little things like that.

I’m so excited. I don’t have to do any of that anymore or not. I still have to do some of it, but not nearly as much. But I really think that you know, to be able to continue moving the dial on this technology to be able to make the big strategic decisions that we do about our businesses and of course, how to bring AI into them.

You need to free up that time. It’s critical. There’s so many distractions out there. I think that it can really help. And part of well being also is inclusivity. I heard a stat this week at the MetLife Triangle Tech X conference that I attended that employees are six times more likely to innovate in supportive, inclusive environments.

So we also need to think about harnessing change and engaging the diverse communities that have skin in the game. And I think that is such a critical part of well being too.

[00:30:08] Helen Todd: Well, I’m a big supporter of well being too. So it’s really good to hear you say that. And I definitely invite all of our listeners and viewers, especially who lead teams to also be thinking about this too, of, you know, I think there’s a lot of pressure to be extremely competitive and to get.

To really offer more and more, but to like, really think about what you’re asking of your team and your employees and your staff and like, how can you streamline it and make sure that there’s space to offer up that time for the creative and strategic thinking, because I know even just talking to women at the conference yesterday, it can be so overwhelming of like, Even taking the time to learn the tools and then figuring out the workflows.

And I think this kind of leads me to, to the next question is that, you know, AI is promising the moon and the stars, and this is something that you said in your opening, but how do you make sense out of all of that? So I’d love to kind of dovetail into that of, you know, what are the considerations for enterprises?

And then, you know, we’ll get to the governance, which I know we, we promised at the top of the show.

[00:31:15] Krista Sande-Kerback: I know, it sometimes feels like AI is promising the sun, the moon and the stars, but how do you make sense of this? I know that I want to be a part of demystifying this for the audiences that I speak to, and it’s also about approaching it myself with humility and beginner’s mind.

First, I think we can recognize that globally we’ve made enormous strides. AI is unlocking productivity, it’s uncovering valuable insights across a large range of applications. And broad adoption of AI systems will require humans to trust their output. When people understand how technology works, and when we can assess that it’s safe and reliable, we’re far more inclined to trust it.

Many AI systems to date have rightly been deemed black boxes, where data’s fed in, results come out, but we don’t really know how. To build trust, we need to be able to look inside these AI systems to understand the rationales behind the algorithmic outcomes and to even ask questions. How did it come to this decision?

It’s about people and technology working together, but beyond the lofty language. And I know as a marketer, we actually have to educate consumers and provide a roadmap for deploying and utilizing this technology with expertise that’s customized to their business. And we can do that today. With the state of the technology that we have and where we’re going at a very concrete level at IBM.

We have a model for a pilot that begins with kind of the following framework setting the exploratory strategy. Launching pilots to establish proof of value, transitioning those top pilots to implementation, building the foundational capabilities to then leverage generative AI at scale, and finally, to scale value capture by transforming multiple production workflows and experiences across the enterprise, say, over a 12 to 18 month horizon.

So we want to bring it to the from the ultra high conceptual level really down to that practical level. And there is a framework for that.

[00:33:13] Helen Todd: Yeah, I love that. So and one thing I just want to point out too, because I know our listeners and viewers come from a lot of different backgrounds. If you need enterprise support, I would be happy to connect you with Krista and everything that she’s saying.

Also, you know, yeah. Is applied to all business sizes of, you know, strategically of taking that conceptual of how to layer AI into your company and workflows at all levels. And if you have more than enterprise needs, I have some people that I can point you to so I want to make sure to open up that invitation because this really impacts every business, whether you’re a solopreneur to a major enterprise.

When it comes to figuring out and I think one thing that’s really important to you that Chris has said is the horizon that there’s a change management and it’s not going to be for as fast as AI is, you know, just all the news announcements and tech deploying and hitting our feeds on a regular basis that it does take time to actually implement.
So I, I don’t know if you want to expand a little bit on your pilot program, but. Or just the change management of how you came up with the 12 to 18 months. But I think that’s important that, you know, it does take time to layer these into your systems because I’m sure a lot of people feel pressure to need AI built into their company yesterday.

[00:34:35] Krista Sande-Kerback: Yes, I definitely want to hone in on that return on investment part of it. And what’s top of mind for almost everybody that I speak to in a business content is how do we make this make sense financially? Historically, it’s been a mixed legacy. Fewer than a quarter of organizations have been able to achieve a return on investment for AI above the average cost of their capital.

Financial returns are improving. I think we’re going to see some real changes in those numbers, especially with the Explosion of generative AI, which, as I said earlier, so many companies are experimenting with today. It hasn’t been enough to fully justify the capital expense. That’s really reliable.

Representational consensual data is foundational. And then that’s of course, that’s a very important part of the picture. And to make A. I. Investments more cost effective, companies need flexible, reusable models that can be applied in a variety of ways. This is also more sustainable and less energy intensive, including generated new content.

So you need to have a coherent strategy. You need to integrate people and processes. And I just really wanted to mention that ROI point because again, that’s so much of what’s on the minds of the leaders that I speak to today.

[00:35:50] Helen Todd: I’m so glad that you mentioned that because we’re all in business. So ROI is so important and even, and one thing I’ll just add to that too, that came up on the panel and has come up in conversations on the show too is also asking the question of like, why, what’s the purpose?

Why are we adding it? And I’m sure you work with businesses on this the strategic point of the strategy too, but at the end of the day, it’s going to impact the bottom line as well. So let’s talk about governance because it’s really important. And I know IBM is doing a lot of really great things on the governance front.

So can you can you tell us, I guess, from a high level of like how you would define or think about AI governance and then what IBM is doing specifically on this front?

[00:36:34] Krista Sande-Kerback: I’ll start with a definition and then talk more about some of what IBM is doing. First of all, AI governance is defined as the overall process for directing, managing, and monitoring the AI activities of any organization.

This includes processes that trace and document the origin of data, models, and associated metadata and pipeline for audits. So now kind of with that definition established, whether your organization is considering adopting AI or is already further down the journey, establishing that governance framework should absolutely be part of your strategy.
Organizations that stay proactive and infuse governance into their AI initiatives can help minimize risk while strengthening their ability to meet ethical principles and government regulations. In particular, the leaders of organizations and highly regulated industries such as banking and financial services are legally required to provide certain levels of transparency to satisfy regulators.

There’s a lot more I could go into on this, but it’s actually so critical that we pay attention to this. Given what’s going on in the world, the AI EU act is in is in draft form right now. We’ve had GDPR regulations for a long time. Minimizing risk is just, is hugely important.

[00:37:48] Helen Todd: That’s really great.

And one thing that I would just add to that too, is like, while there’s not adequate. Regulation or any regulation right now in the U. S. Around a I that companies should really be thinking about this now and adopting these to set them up for success for because it’s inevitable that it will be coming.

And especially as you mentioned in heavily regulated industries that already require it. But I do know Because you shared with me that IBM has an ethics board for AI governance, so I’d love for you to share more about that and other initiatives that IBM is doing on the governance front.

[00:38:23] Krista Sande-Kerback: Absolutely.

IBM has an AI ethics board and its mission is to support a centralized governance review and decision making process for IBM’s AI ethics policies, practices, our research, our products, as well as our services. This importantly includes a very diverse set of stakeholders from across the company and is supported by community of IBM employees who serve as AI focals and ethics advocates.

As new technology emerges, foundation models, for instance, the AI ethics board is actively gauge and supporting alignment with these pillars and principles. And it’s going to evolve to address new AI ethics issues. The reason that it’s needed is that people view AI as the top technology to address humanity’s challenges.

And that comes from a, an IBM Institute of business study, value study that we ran in 2021. But 88%. Participants said it’s important for organizations to address the ethical and responsible use of AI address personal data and information. 89 percent are asking for organizations to be transparent about how data and AI models are built, managed, and used.
And 87 percent want explainable AI. They want organizations to be able to Explain how a I arrived at a decision. So it’s clear that organizations need to take a principal stance on current concerns, but then also follow through with meaningful actions. And organizations, AI teams, as I said earlier, they tend to be significantly less diverse than their enterprise workforces.

So we need to have we need to close these gaps across gender, race, and orientation, and we need to have these diverse stakeholders. I think that we’ve done that by the way that we’ve structured our ethics board.

[00:40:05] Helen Todd: That’s amazing. And I love the points that you’ve made about a black box, because we’ve seen this with social media, that all the algorithms, like we don’t know how they make decisions.

And that the larger the large language models are that the more complicated it is, and it’s all black box. And then the You know, the creators of these tools are like, well, we don’t know. And then it’s like, how do you regulate that? So it’s so good to hear that IBM is having that. And one film that I also recommend everyone watching is the A.I. Dilemma by Tristan Harris and Aza Raskin is more of a presentation that’s on YouTube about. The how problematic it is with that black box. So it’s really great to hear that IBM is addressing that. Cause that’s a really big concern within the AI landscape. And another thing that I loved in looking at your framework ahead of our conversation is IBM does ethics by design.

And I’d love for you to expand to our listeners and viewers what that means to IBM.

[00:41:12] Krista Sande-Kerback: Ethics by design is a structured framework with the goal of integrating tech ethics in the technology development pipeline, which includes, but it’s not limited to AI systems. It enables AI and other technologies as a force for good by embedding these tech principles throughout product services and broader operations.
So, kind of, per what you’re saying about Tristan Harris, per, I think everybody’s aware that we recently had some tech leaders call for the six month pause in the training of more powerful AI systems to allow for the creation of new ethics standards. And you know, there’s good intentions there. It missed a fundamental point, which is that these systems are already in our control today, as are the solutions.

With responsible training together with an ethics by design approach, there’s a different way. You build that over the whole AI pipeline. You support it by multi stakeholder, diverse collaboration around AI, and that can make these systems better, not worse. AI is going to continue to evolve. Yes, there’s going to continue to be regulations.

No one is immune. Businesses are small and large. But there are systems in use today. There are systems coming online tomorrow. Training has to be a part of that responsible approach. We don’t need a pause to prioritize responsible AI. Because we have these frameworks.

[00:42:25] Helen Todd: This gives me the most hope so far out of all of the interviews and that we’ve had on the show of an actual solution and not just talking about it conceptually or academically.

So that, that is so great to hear. And one question too, on this front, cause I did talk to a gentleman who is in AI. He has government contracts for the military and they have to make sure that Everything is explainable and the models that they build have much fewer decision key points so that it is explainable.

And how he explained it to me is with models like chatGPT, the dimensionality of all of the points that go into consideration for the output is just so enormous that the complexity of it makes it, I don’t know if you use the word impossible, but impossible to understand the decision making.

And that’s why it’s a black box. So I don’t know if you can speak to the difference of the chat GPT and the dimensionality of the decisions going into it and IBM’s approach or the differences between those two. Cause when I heard, when I hear stuff like chatGPT is a big black box. It’s like, well, why?

And I know Ezra Klein in another interview, I think on Hard Fork, or I forget which show. I know Ezra Klein has mentioned before that it should be on the onus of the companies to be able to have the explainable ai. So it’s so good to hear that IBM is doing that, but I didn’t know if you could speak to the difference between the impossibility argument and then IBM’s approach of explainable AI and the difference.

[00:44:02] Krista Sande-Kerback: One thing that comes to mind that I think is exciting that we’re working on is we’ve, as I mentioned earlier, we’ve announced a new. Innovations this year. The introduction of what’s next dot AI and what’s next dot data. And recently actually September, we announced the general availability of our first models in the Watson X granite model series.

And this is a collection of generative AI models that advanced. Generative AI into business applications and workflows. I want to mention this because businesses could bring their propriety proprietary data to IBM base models, and they can build a model that’s unique to their business and use cases.

It’s not necessarily about, you know, the biggest model is the best. It could be something that’s very custom, and it’s an opportunity for them to work with us in a way that IBM Provides the standard contractual, intellectually property protections for IBM products. This will apply to these Watson X AI models, and it’ll this will really enable businesses to be value creators.

So I think that’s an exciting example of of an innovation that, you know, I think is going to really help businesses.
[00:45:11] Helen Todd: And when we hear data is the new oil and data, this is really a great example of how a company can leverage their own data. And is that how I understand it?

The IBM solutions is to really help companies make the most out of their own data with AI, correct?

[00:45:29] Krista Sande-Kerback: That’s right. I mean, it’s a mix of they can bring their data in. There’s also all of this open source data. Another application that I’m super excited about is IBM and NASA open source, the largest geospatial AI foundation model on the hugging face.

One of our partners with access to NASA satellite data, and that’s being used to accelerate climate related discoveries, huge number of applications, monitoring natural disasters, et cetera. So of course the inner geography and nerd in me was very happy when this announcement was made, but it’s really a mix.

There’s these models are being provided open source. You’re bringing companies data. We’ve got we’ve got options.

[00:46:05] Helen Todd: That’s so cool. And open source ethically done. And so always love hearing that again too. Well, I know we’re getting close to time with our interview. So, you know, I’d like to ask this question on the show.
If you want our listeners and viewers to remember one thing, what is it? That you want them to walk away with from our conversation today.

[00:46:29] Krista Sande-Kerback: There’s no better time to learn about and get your hands on AI. As I said earlier, I come from a liberal arts background. You’ve come from all sorts of different backgrounds, industries.

That’s a blessing. That’s we should see our differences as opportunities. It’s a chance to stand out, build diverse teams, and make this technology the best that it possibly can be. On the flip side, and just, I’m going to put my governance hat on for a moment, regulations are going to go crazy at some point.

There are a lot of instances where AI can go wrong, where we can suspect litigation issues. So please, we do have to put proactive governance.

[00:47:13] Helen Todd: And then, and I know you mentioned a lot of different Watson and links and cool tools. So I’ll be sure to put all of these links and everything that you mentioned in the dedicated blog post that accompanies this interview. And you can, all of our listeners and viewers can find the link to that.

blog post in the description. But before we sign off, is there anything else that you want to share? Krista?
[00:47:38] Krista Sande-Kerback: You know, I know that there’s a lot of information about AI out there. And it can feel probably overwhelming at times. I hope that this interview has. It’s done its job in spurring some new ideas without adding to that overwhelm.

But here’s some specific ways that we can offer to help to get you past information overload and hopefully to some new value. IBM has a program called Skills Build. And in September, we committed to training 2 million learners in AI by the end of 2026 with a focus on underrepresented communities. And this coursework is available for free online.

It includes things like prompt writing, getting started with machine learning, improving customer service with AI and generative AI and action. So there’s lots of opportunities to get that training. And then You know, as I mentioned as well, governance is so important. We are, we’re feeling good about our upcoming launch.

The fact that you can govern AI models and machine learning models together through our technology is a big deal. We believe that we have something that’s different. In any event, make sure to be paying attention to this as you continue along on your AI journey. And there’s a few different ways to get started with IBM.

Should you you know, choose to explore further one is through a free trial. Second, we can offer client briefings where it’s an opportunity to get a discussion and custom demonstration of IBM’s generative A. I. What’s next point of view and capabilities. And through that, you can learn how generative AI can specifically be leveraged for impact in your business. And then finally, there’s a more in depth pilot program where you can engage with an IBM multidisciplinary team to jointly innovate and rapidly prove the business value of generative AI solutions using Watson X. So those are some of the things that IBM as a company offers.

I as an individual, I’m just so happy to be here today and to share ideas. I look forward to exchanging more. I know I’m going to keep learning and I look forward to hearing from this audience.

[00:49:37] Helen Todd: I love that. And and just one clarifying question. And this is just really, you know, for my own knowledge to the tools that you mentioned, like the WatsonX.governance, are these all enterprise level?

Or can any business small to medium size, like use these tools as well?

[00:49:55] Krista Sande-Kerback: It really runs the gamut. We’re working with clients, large and small. IBM is focused on AI for business will continue to be that way, but there’s a need for small organizations to be paying attention to all of these things as well.

And we offer solutions for both.

[00:50:12] Helen Todd: That’s so great to hear. Cause I know I’ve mentioned, or we’ve mentioned enterprise a lot throughout this conversation. But for everyone, who’s not enterprise listening and wants to do the trial I think that’s great. And. I know I will be using it after our conversation too, and can report back after that.

Well, Krista, it has been so wonderful having you on the show and I just love your mission to help, you know, women and everyone demystify AI. So. Thank you so much for your time and your insights today.

[00:50:45] Krista Sande-Kerback: Pleasure speaking with you. Helen. As always.

[00:50:50] Helen Todd: Thank you for spending some time with us today. We’re just getting started and would love your support. Subscribe to Creativity Squared on your preferred podcast platform and leave a review. It really helps and I’d love to hear your feedback. What topics are you thinking about and want to dive into more?

I invite you to visit creativity square.com to let me know, and while you’re there, be sure to sign up for our free. So you can easily stay on top of all the latest news at the intersection of AI and creativity. Because it’s so important to support artists, 10 percent of all revenue Creativity Squared generates will go to ArtsWave, a nationally recognized nonprofit that supports over 100 arts organizations.

Become a premium newsletter subscriber or leave a tip on the website to support this project and ArtsWave. And premium newsletter subscribers will receive NFTs of episode cover art and more extra content. To say thank you for helping bring my dream to life. And a big thank you to everyone who’s offered their time, energy, and encouragement and support so far.

I really appreciate it from the bottom of my heart. This show is produced and made possible by the team at Play Audio Agency. Until next week, keep creating.