[00:00:40] Camille Morhardt: Hi, and welcome to Cyber Security Inside. I’m Camille Morhardt, and with me is Tom Garrison co-host of Live from the Green Room. Live from the Green Room is actually something we do where we join a conference, and then we grab thought leaders in between the talks that they’re giving so that we have a chance to chat with them and delve a bit deeper into the topics that they’re discussing while they’re at the conference.
Right now, we have with us Nufar Gaspar, Director of AI Everywhere within the IT Department at Intel. Welcome to the podcast, Nufar.
[00:01:14] Nufar Gaspar: Hi, thank you for having me.
[00:01:15] Camille Morhardt: So we’re really happy to have you on here. And, as Tom said, just a few seconds ago, “Wow! AI everywhere!” How actually do you deal with that? What is your goal and what are you actually trying to do?
[00:01:30] Nufar Gaspar: So I’ve been working on the iPhone 12 years. My entire team has been doing that and we wanted to really bring a lot of the learnings to others across Intel, and really scaling our usage and value across the company, and that’s when we initiated AI Everywhere. So the name kind of speaks for itself.
[00:01:46] Tom Garrison: Wow. You’ve been working on AI for 12 years. That just blew my mind. I was thinking of AI as something that’s a relatively new workload.
[00:02:01] Nufar Gaspar: So it’s little known fact that it’s been around for quite awhile, also in production in corporates for quite a while, but the hype is relatively new. So with that, you’re correct. (laughs)
[00:02:14] Camille Morhardt: That’s hilarious. So will you give us a sense generally in industry, why you think expanding AI within a company can fail? What are some of the key reasons that it doesn’t pan out?
[00:02:28] Nufar Gaspar: There are quite a few; many of them surprisingly comes to not properly selecting what to work on. I think many cases people are just kind of excited about the technology and say “let’s go!” and implement that and not really asking themselves whether it’s feasible, the idea that they have or whether they have all the business knowledge, the background, the agreement with the management that they’re working on something that is critical and that there is likelihood to really implement that.
And after that many other reasons–skills and tools and other considerations; but surprisingly, I think many of the fails can be nipped in the bud. If you do due diligence on the idea that you have and make sure that you really work on something that is going to be feasible and successful enough.
[00:03:15] Tom Garrison: You mentioned things that fail, but, you’re in IT now. And part of your charter is to work with various groups around the company to implement AI successfully, not fall into those traps you just said. For some of our listeners at other companies that work in IT, what are some of the things that you recommend that they do when they’re approaching their teams about AI? What are some things that they should be doing that maybe not so obvious?
[00:03:47] Nufar Gaspar: So, first of all, I think a lot of it goes back to training and making sure that people not just learn only the Intro to AI course, but rather learn the BKMs and the methodology and how to properly go after exploring a new idea with AI. There are pretty well established methodologies, some of them have been around for 20 years. So part of it is education and we do that a lot, making sure that people work on their ideas properly with the right tools. So that would be one.
The other is really to probably find the best responsibilities between the business experts and the AI expert and find the best way to collaborate between the two. If everyone brings to the table what they’re best at business experts learn AI a little bit, so they can talk to the AI expert and the AI expert learns the business to the extent where they’re meeting you halfway through. Good things happen when the collaboration is really effective and sharing a common language.
And of course I have many, many other things, but just what comes to mind because I give webinars on that so I can talk about that for 60 minutes. So I stop here. (laughs)
[00:05:00] Camille Morhardt: Well, I’m wondering, you’ve had tremendous success with this AI Everywhere conference that you set up and it’s internal to Intel. I know we’re on a public podcast right now, but I’m wondering, along the lines of Tom’s question, what kind of learnings, best-known practices did you figure out as you set up this conference that you feel other companies or organizations could also set up or structure within their group?
[00:05:30] Nufar Gaspar: First of all, we never dreamed that the conference was going to be that big; we thought we’re going to have a one- or two-day conference and a couple of hundred of people. And now we have four days and almost 5,000 people. And so I can’t tell you to dream big because that’s kind of something that happened in the process.
But what we really try to make sure is that we cater to a vast range of skills and whether people are just getting started with AI, but also cater for the practitioners that are experienced. We made sure that we bring diverse content like technical talks of people talking about the work that was already done.
And also fortunately, we’re able to bring all the top management of Intel to come and share their point of view. And that really helps when you have a lot of support from management that really acknowledge the importance of AI. So with that, I think was the kind of a transform success because we got a lot of endorsement from top management, as well as more together and many of the practitioners and they use cases and the learnings that caters for everyone and not just for some possibilities.
[00:06:34] Tom Garrison: It occurs to me that from a workload standpoint, maybe there are unique characteristics of the AI workload that IT has to deal with–unique from standard sort of business process or business infrastructure backend that you would normally expect. Is that true? Does AI introduce unique challenges for IT to be able to support that workload in the environment?
[00:07:04] Nufar Gaspar: Probably the answer is yes and no, because a lot of the enablers for us is essentially data. Having good data is not just for AI, it’s also for analytics and automation. And actually, a lot of the work that is usually been done by IT is to redesign the business process, really to structure the data. So once that is done for it, even if it’s done for other purposes, usually it will be easier to introduce AI on top of just IT coming and making some organization in the business processes and the data in the IT systems. So that’s the part where it’s similar.
The part with it’s sometimes different is that first of all, it’s a very high-end technology that sometimes requires, even more high-end than ideal kind of workloads or it systems, it requires a lot of faith from the business partners. So it’s not just that you can come as an outsider, say, “Hey, take my algorithm, change everything that you work on, trust it and go about your day.” You really often have to be much more inclusive to the business and to have a stronger partnership and collaborations so that they really will be willing to adopt something that sometimes is in between intimidating and transformational or disruptive the way businesses run. That’s the part where it might be a little bit different than just your regular IT work.
[00:08:22] Camille Morhardt: I hear a lot about some of the main kind of transformations happening right now between IT and business units out in the world along the lines of automation of tools. What kinds of things are you seeing?
[00:08:37] Nufar Gaspar: I think many of the ideal organizations are focused exactly on the two things that we’re talking, that is a digital transformation, but also AI. We see both of those topics being discussed sometimes together, sometimes separately between the two. In many organizations, IT are the ones bringing the AI to the edge of the technology. Maybe there are some other organization doing that.
Another very big trend that is coming sometimes from IT, also very classical IT thing that how can you do machine learning or AI at scale and low total cost. So those are a set of best practices and methods and tools to really help you not just solve your problem with AI, but also we need production in a sustainable way. And so I see a lot of that in the chatter and in the work being done in many organizations.
[00:09:22] Tom Garrison: Nufar, I wonder if you could give us a glimpse into your day now, you got a big 5,000 person conference going on. You’re talking to us for a few minutes. Where do you go next?
[00:09:38] Nufar Gaspar: So the conference is firefighting and small heart attacks before each session (laughs) and a big win once the lights go on and everything seems to be normal. Fortunately I work with a very dedicated team– my team and others that volunteer to help; we’re kind of just making sure that that would from here on. But just today I deliver them a tutorial. Later on I’m going to deliver an incubation session where we bring to light some new AI ideas and discuss them. There will be many technical talks happening live. I think some of them are happening as we speak. So it’s a very hectic and everything is happening twice to cater both for at the DMR Geo and later to the European and the Asian region. So a very busy day, but extremely fulfilling as more and more people come and join us.
[00:10:42] Tom Garrison: Yeah, we certainly appreciate you spending time with us.
[00:10:45] Camille Morhardt: Thank you so much for joining us. Nufar Gaspar who is director of AI Everywhere for IT at Intel. She’s also head of this conference, which is this massive internal 5,000 person spanning every geography and multiple time zones. We pulled her out of her day. I don’t even know what time it is in her world right now, but thank you so much for joining us and giving us some insight into what it is to try to structure the expansion of AI across a Fortune 100 company.
[00:11:19] Nufar Gaspar: Thank you, Camille. Thank you, Tom.