เกี่ยวกับตอนนี้
In this latest episode, Mustansir Saifuddin delves into the ever-evolving landscape of Gen AI with Todd Kackley, Vice President and CIO of Textron. Todd shares how his team implemented their inaugural generative AI solution and the crucial role leaders play in crafting scalable, well-governed, and future-proof data analytics and AI infrastructure.
Todd shares invaluable insights into striking the balance between leveraging third-party capabilities and developing in-house models, shedding light on the dynamic interplay between proprietary solutions and open source technologies. Whether you're a seasoned leader or a budding enthusiast, understanding the nuances of Gen AI and its implications is key to driving innovation and staying ahead in today's digital landscape.
Todd A. Kackley is vice president and chief information officer for Textron Inc. In this role, he leads the business unit chief information officers and the Textron Information Services (TIS) organization. He oversees Textron's Information Management Council and manages Textron's information technology supplier and outsourcing relationships.
Prior to his current role, Kackley was executive vice president and chief information officer for Bell where he developed and executed IT and digital strategy, aligning business systems, infrastructure, cybersecurity and development capabilities to the needs of the business.
Connect with Us:
LinkedIn
Twitter: @Mmsaifuddin
YouTube
or learn more about our sponsor Innovative Solution Partners to schedule a free consultation.
Episode Transcript
[00:00:00.890] - Mustansir SaifuddinWelcome to TechDriven business brought to you by Innovative Solution Partners. Today, an old friend, Todd Kackley, Vice President and CEO of Textron, joins me to delve into the ever evolving landscape of Gen AI. Todd shares insights into implementing their inaugural generative AI solution and the pivotal role leaders play in crafting a scalable, well-governed, and future-proof data analytics and AI infrastructure.
[00:00:40.270] - Mustansir Saifuddin
So thank you for joining the session today. Today, we will be talking about any organization's readiness to embrace Gen AI. And I would like to get your perspective on this topic. So with that, I would like to get into our session.
[00:00:57.060] - Todd Kackley
Let's do it.
[00:00:59.280] - Mustansir Saifuddin
Awesome. Great. So let me start with, as you lead your organization in these times of quicker adoption, we see this across the board, across technologies. When we talk about AI-driven solutions, how are you and your team striking a balance between leveraging third-party capabilities versus developing in-house models?
[00:01:24.640] - Todd Kackley
Well, Mustansir I think that's a very good question. And this is also something very relevant for us at Textron. We recently implemented our first production generative AI solution, and that solution leverages a third-party proprietary generative AI model. And part of that decision making was certainly there's a lot involved in developing artificial intelligence models as a technology company as well at Textron. We've dealt with AI, and it's been part of our product offering and some of the things that we do around geospatial analysis and other things. And we built models in the past. So the decision to make versus buy, it wasn't necessarily one that we had to approach when it comes to generative AI, because I think this is an area where it's certainly a buy, leverage third-party models because of the capability and the massive amounts of compute and investments required to go off and build a model, the capability that we're We're seeing coming from third-party proprietary models that are secure. And certainly that's something that's very important to us, making sure that we leverage a third-party model that is secure and allows us to be able to work within the confines of our data tenant and be able to leverage the benefits of the OpenAI, but at the same time protecting our sensitive data.
[00:02:58.640] - Mustansir Saifuddin
Now that makes sense. I mean, let's fast forward, right? Are you even looking at in-house models sometimes in the future, or is this something that you think a third party will give you the leverage that you're looking for as far as technology as well as security? One of the big question marks right now with Gen AI.
[00:03:16.820] - Todd Kackley
In our defense businesses, it's something we're certainly possibly kicking around where there may be smaller scale opportunities for us to build lambda-based models or others using Compute technology that's not massive data center type of the computing horsepower that you need to be able to do something as large as an Open AI or an Azure AI type of experience. I contend that most of the use cases that we're seeing right now, I think there's a difference that people need to look at when they approach this question, and it's what What outcome are you trying to achieve? And if the outcome is, I want quick analysis or summarization or content generation and so forth, recreating that wheel and trying to build that model may not make sense where those third-party proprietary models already exist to do that. In fact, I'm seeing even in the ecosystem of development of workbenches and use cases around these third-party models, there are several developments already in place that have the turnkey copilot for a contract's evaluation or proposal generation or service centers and doing things like help desk. Those are accelerating the ability to implement quickly and get value quickly. And even the model itself, I read something yesterday that It just blew my mind that the evolution of ChatGPT 3.5 to 4.0 to 5.0 is at a pace that it's reinventing itself about every six months.
[00:05:12.810] - Todd Kackley
And that's 4xMaur's law. If you think about that, the pace of Maur's. As a technologist, that just blows my mind.
[00:05:21.990] - Mustansir Saifuddin
For sure. I think that leads me into my next question. When we talk about proprietary solutions versus open source, How does that affect your strategy?
[00:05:34.270] - Todd Kackley
Well, I certainly appreciate how open source really drives innovation, and it really OpenAI is a good example. Openai is a good example. The release of ChatGPT a little more than a year ago as an open source technology has really fundamentally changed the perception of artificial intelligence and pretty much put the technology technology in the hands of the everyday consumer. You don't need to have a PhD in neural network to go off and understand how to interact with artificial intelligence. In fact, I think that has really accelerated. It's the fastest adopted technology hitting the 100 million user mark in less than two months. But at the same time, as a technologist, I think we've seen over time that there are certainly risks with some of the open source technologies, whether they become proprietary at some point or they get consumed by some player and they now become a fee-based application, and you've already committed to the technology. We've seen that across some of the major players that consume other open source technologies, and they turn around and license it. That's certainly a risk. It's certainly something that you can't undermine or underestimate the amount of patching and keeping up with the vulnerabilities of open source technology, because I think they're a little more susceptible to one of the vulnerabilities.
[00:07:11.670] - Todd Kackley
But it's certainly something we have to weigh out when we make decisions around using proprietary technologies versus open source.
[00:07:21.460] - Mustansir Saifuddin
Yeah. I think what I'm hearing from you, it seems like a lot of that is dependent on your business model and your individual needs as an organization, right?
[00:07:31.620] - Todd Kackley
Absolutely. It's not just, let's go find the lowest cost solution always. You have to take into consideration many aspects not just is it the best solution? Is it a secure solution? Is it one that's going to endure, that's going to have a life cycle that is going to sustain? Because you don't make a technology investment. You hope not to make a technology investment and expect It's like that they only have a couple of years life cycle. You want it to endure a little bit. Sometimes what we've seen in the ecosystem, even in the larger players, in the consolidation of certain applications of the stacks and so forth, and the reinventing of those stacks, it's really forced us to stand back and look at all of these investments to make sure that there's going to be viability in the solution provider to continue to support, do investments, and continue to evolve that product set, whether it's a small individual capability or a larger enterprise-wide program.
[00:08:42.080] - Mustansir Saifuddin
Yeah, for sure. I think longevity and durability It seems like it needs to be built in the solutions that you're looking for, right?
[00:08:50.300] - Todd Kackley
Absolutely.
[00:08:51.310] - Mustansir Saifuddin
So thinking about from an experience perspective, you got years of experience, both in business and technology. What are some of the good use cases of AI-driven solutions that you've seen deliver tangible business value? I mean, it's all about providing value to business and then at the same time building trust with the business. What's your take on that?
[00:09:13.760] - Todd Kackley
Well, we go back a long ways Mustansir. I appreciate you not actually stating how long I've been doing this job, but it feels like much longer. But it's interesting. The whole generative, as I just said, the whole generative AI thing, if you think even a year and a half ago, this wasn't on the radar for most CIOs or most technologists in general. And now it's pretty much consumed most of our discussions. For sure. So we decided here at Textron to take a look at an applicable use case. And while there are lots of analysis on where these capabilities actually work well or maybe provide the best value, we focused on our service center area of opportunity. And when you think about a services business where you have large volumes of data, not just on how you support, sustain that customer's product and the history of sustaining that customer's product, but you also have the combination of data around all of your All of your technical publications, all of your engineering specifications, everything known. And many of our products are in a regulated business, like commercial jets and so forth, or helicopters. So you have a lot of data that you need to maintain, and having that in disparate sources or even trying to make it accessible to a maintenance technician or a service center, call center resource, sometimes is a challenge.
[00:11:06.350] - Todd Kackley
So we sought to try to accelerate the access to content internally for our maintenance technicians in our service centers, and it certainly provided an opportunity for them to have a capability, allow them to quickly either troubleshoot or identify content and point them to the the latest publication, latest operating procedure, latest set of instructions, and give them some insights as to how they should consider approaching either troubleshooting or diagnostics types of activities. We're certainly seeing that that's going to drive a measurable impact in the time that these maintenance technicians normally spend in front of a computer. They would rather spend time turning wrenches and doing the work to repair the aircraft and get our customers back in the air. And that's where the value is, having our customers' aircraft not sitting on the ground in our service centers, but in the air. So we see that not just in our aircraft industry, our business. But we also see that opportunities across anywhere we're providing a customer support type of solution. You think about even in IT, The help desks, we have... Most large organizations have a level one, level two, level three type of help desk.
[00:12:38.140] - Todd Kackley
And as questions go from level one to level two, and level two to level three, the cost of that answer continues to get higher. So bringing the information and the ability to solve that person's question closer to level one, or maybe not even needing level one at all, if you're bringing it to the user, it And yourself, most people that I know, when you have a general household issue, you go to YouTube. How do I solve this? How do I fix this? And before you call the service technician, everything, you try to figure out. And I think users are more inclined to do that if they have access to the right information at their fingertips, and it's useful and current and timely. So I think those are really applicable use cases.
[00:13:30.620] - Mustansir Saifuddin
No, I think for sure, what I'm hearing from you is time to delivery, right? And the example you use of YouTube is so real. Every day we do this, we look up things and we are trying to find things. How can we apply the same approach in a business scenario, in an organization which has got all these different levels of teams and support systems in place? The time to delivery to a customer, either internal or external, makes a lot of sense. I And this whole use case, especially when you talk about what Gen AI can do, seems like it's just helping get to the answers much faster than we ever comprehend it in the past.
[00:14:13.560] - Todd Kackley
I agree. I do think it's going to transform how knowledge workers and beyond that, even the example of the shop floor of workers that we're having, the maintenance technicians, they're very excited because if you think about it, we have a... Most organizations have a challenge around keeping talent and skills and the continued evolution of having to provide them with the training and the knowledge. And the generative AI opportunities that allow us to use content generation for training or just the accessibility to the information levels the field a little bit from somebody that maybe 20, 30 years of experience to somebody that has less. So they all are leveraging the same resource and getting that expert information to help them accelerate their jobs. So I'm excited about that. I think that's going to help, particularly help us bring in the next workforce in the future and prepare them for the skills.
[00:15:25.600] - Mustansir Saifuddin
No, totally. I totally agree on that. So taking a little segue over here, on On a personal note, where do you go do your research? What are some of the readings that you're currently doing? Would you like to share?
[00:15:39.140] - Todd Kackley
So particularly on the topic of generative AI, there's a lot to stay on top of. I certainly follow most of the leading business journals and tech journals, and stay on top of that. I've got a lot of flags in my search criteria to be able to pop up new things that are coming up, and certainly look for those articles. And I follow several podcasts and tech leaders, spend a lot of time talking to systems integrators as well. And then I have a very large network that I belong to a number of professional organizations with technical leaders, CIOs, and we have a lot of discussions on this topic as well. It's an evolving topic, and I find that not just generative AI, there are several disciplines when leading an information technology organization that one needs to stay on top of. And then I try to find time also to read self-development books, things certainly leadership and books on organizational change and leading large organizations and working across generations. So there are tons of resources. I think the most valued resource is in your peer network, having an opportunity to sit with peers in the same type of role that you have and have those discussions, whether it's a chief information officer, Chief Information Security Officer, CTO, even an analyst in a business function or a technical function, having that ability to work with their peer group and understand what's going on day to day helps them develop their self-awareness and their knowledge of their skills.
[00:17:48.750] - Mustansir Saifuddin
Yeah. It seems like your time is filled with all these different areas where you're pulling in the information and making sure that you're applying those your day to day operations as well as your learning that you're going through.
[00:18:05.910] - Todd Kackley
I can't profess all of it sticks. I mean, there's probably more content going in that actually I can retain. But it's one of these things I find that having conversations about it and checking your understanding, this gener AI, going back to that quickly, my having to understand how all of this works and feeling comfortable with the solutions that we're looking to bring forward and how those operate safely, securely, and understanding how we explain that to the users and leaders as well. That's a big piece of that, being able to understand the capabilities and technology enough to assure key stakeholders and leaders that we're making the right decisions as Absolutely.
[00:19:01.290] - Mustansir Saifuddin
Absolutely. So let's talk about from a Gen AI perspective, right? It is so new. It's a developing area for both business and technology. How do you see leaders such as yourself establishing a data, analytics, and AI infrastructure that is both scalable, well-governed, and at the same time, future-proof? It's a lot of things in one, but I-Yeah.
[00:19:29.970] - Todd Kackley
That's a really good question. One of the things that we're still having a discussion on is there's been several years in the data and analytics, and I know you and I go back several years in the data analytics space originally. And organizations have spent a lot of time and money and effort putting together large data lakes or a focus on structured data for the purpose of data and analytics, for the purpose of dashboards, for the purpose of possibly maybe even more impactful data sciences and driving more value out of predictive and prescriptive type of data analysis and so forth. And that's all part of this evolution, I think. And we're still getting our arms around this. There's And maybe I'll just say this, and whether people think I'm right or wrong, we'll let them be the judge of that. But I think there's a Venn diagram here where data sciences and data analytics overlaps with the generative AI, and there's something in the middle of that Venn diagram. But the difference is when you're really focused on the predictive and the prescriptive repeatable model of a data science model where I need this machine learning algorithm to drive a very predictable output, understanding whether or not my machine on the shop floor is performing predictably, and I can identify when it's varying off of its function versus the generative AI type of experience where I need very comprehensive analysis with structured and unstructured data versus the data science, which may be more structured and curated data.
[00:21:40.160] - Todd Kackley
I've often described it to people. I see data sciences as I'm looking for the needle in the haystack, and I see generative AI is what's in the haystack? And oh, by the way, there's a needle. So there are two different ways to look at it. I think when it comes to making decisions around the whole data side of this, data governance and data ownership is going to be a key piece of that, and it always has been, because ultimately, neither side works well with bad data. So that's certainly going to be something that's going to continue to drive the discussion. But I think a bigger piece of this is going to change the The way we think about architecting our data, whether or not it needs to be as structured as we've had in the past. We were just having this conversation the other day with our senior leaders in the past to build something like a chatbot or to build something like a very large search engine, you had to spend time creating the database, creating the indexes, creating all the special keys, doing all the tables and everything you to do, and then do all the programming around it.
[00:23:02.920] - Todd Kackley
With generative AI, it doesn't matter really what structure your data is in. It's really about the prompt and how you ask the question and interface with the model to get the answer. So I do think we have to make a clearer distinction on what tool sets are used for what purposes, because you can't always use one or the other to solve the same thing. But that's going to be the challenge for technologists as they look at, do I need a data mart? And one of the risks here also When you're dealing with data, this is ultimately replicating all this data over multiple places and paying for it in multiple places and so forth. So you have to think about that. What's the architecture? Does it sit natively in the application and AI works around it. So we'll see. It's still too early to tell, but I think those are things that we have to directionally think about.
[00:24:09.750] - Mustansir Saifuddin
Yeah, definitely. I think it's an evolution, and we are just getting into it, finding ourselves right in the middle of it. Great discussion so far. But based on all that we have covered so far, what is that one key takeaway that you want to leave with our listeners today?
[00:24:30.030] - Todd Kackley
Well, if we're talking about, since the theme has been largely generative AI, I think the key takeaway is that There may be a few, but I think when you're having the discussion around generative AI, it's important to understand that many organizations have been doing AI for a while. Data sciences is part of that evolution, building an algorithm and building a model and so forth. So I've seen peers in my industry and others, even not in my industry, their immediate reaction was, we've got to shut this genervate eye down or we need to go slowly. And the reality is, I think it's an evolution of what we've been normally doing with It's things like machine learning and algorithms and so forth. And it's something that regardless of whether or not, and I'm finding here, regardless of whether or not you have a position on to go fast and forward or to be more risk averse and hold back, the users are going to find a way to use this capability. And that's probably the big message here. If you're a technologist listening to this and you think you've got ChatGPT blocked or you're shutting down and saying, We're going to go slow.
[00:26:05.320] - Todd Kackley
Trust me, your users are finding a way to use this technology to help them with their jobs. And you're going to have to figure out how to make it part of your business processes going forward in a secure manner that meets the requirements of the business and the need. And and embrace it because it's not going to go away.
[00:26:34.360] - Mustansir Saifuddin
Thanks for listening to Tech-Driven Business brought to you by Innovative Solution Partners. Todd shared valuable insights on how organizations can transform business with generative AI. His main takeaway, Gen AI is here to stay. If you are a technologist, know that users are finding a way to use this technology to help them with their jobs. We would love to hear from you. Continue the conversation by connecting with me on LinkedIn or Twitter. Learn more about Innovative Solution Partners and schedule a free consultation by visiting isolutionpartners.com. Never miss a podcast by subscribing to our YouTube channel. Information is in the show notes.
