1 00:00:02,560 --> 00:00:06,280 Speaker 1: As technology progresses, AI becomes more routinely a part of 2 00:00:06,320 --> 00:00:10,240 Speaker 1: people's everyday interactions in the world. From self checkout machines 3 00:00:10,280 --> 00:00:13,920 Speaker 1: to generative AI search engines, AI has many offerings to 4 00:00:14,000 --> 00:00:16,960 Speaker 1: how people pursue solutions in their day to day lives. 5 00:00:18,120 --> 00:00:21,200 Speaker 1: I myself have been using AI powered copilots to help 6 00:00:21,239 --> 00:00:24,720 Speaker 1: me with my coding projects. This AI assistant not only 7 00:00:24,800 --> 00:00:28,560 Speaker 1: understands my chosen coding language, but also plain English text 8 00:00:28,560 --> 00:00:32,120 Speaker 1: as well. It automatically creates code for my prompts. It's 9 00:00:32,159 --> 00:00:36,000 Speaker 1: been a massive help for me these AI driven coding assistants. 10 00:00:36,200 --> 00:00:39,000 Speaker 1: They're always there and available. It's like having a senior 11 00:00:39,040 --> 00:00:42,919 Speaker 1: developer by my side without the grumpiness. This was my 12 00:00:43,040 --> 00:00:46,879 Speaker 1: first experience working directly with an AI copilot, and for me, 13 00:00:46,960 --> 00:00:49,440 Speaker 1: it has reinforced my view that AI is a tool 14 00:00:49,479 --> 00:00:53,800 Speaker 1: meant to help, not hinder our endeavors. I, like many 15 00:00:53,800 --> 00:00:57,600 Speaker 1: developers using these assistants, can now focus on creative problem 16 00:00:57,720 --> 00:01:02,480 Speaker 1: solving and working on harder and more impactful solutions. This 17 00:01:02,640 --> 00:01:06,760 Speaker 1: synergy empowers myself and other developers to tackle complex projects 18 00:01:06,800 --> 00:01:11,039 Speaker 1: with confidence, leading to faster innovation and higher quality software. 19 00:01:11,840 --> 00:01:15,479 Speaker 1: SAI continues to evolve so as the potential for groundbreaking 20 00:01:15,480 --> 00:01:22,640 Speaker 1: advancements in the tech industry. I'm all in, Hey there, 21 00:01:22,800 --> 00:01:26,640 Speaker 1: I'm grain class and this is technically speaking, an Intel podcast. 22 00:01:27,240 --> 00:01:30,600 Speaker 1: The show is dedicated to highlighting the technology is revolutionizing 23 00:01:30,720 --> 00:01:34,560 Speaker 1: the way we live, work, and move. In every episode, 24 00:01:34,600 --> 00:01:37,800 Speaker 1: we'll connect with innovators in areas like artificial intelligence to 25 00:01:37,840 --> 00:01:41,920 Speaker 1: better understand the human centered technology they've developed. This has 26 00:01:41,959 --> 00:01:45,600 Speaker 1: been such an interesting season, filled with not only amazing guests, 27 00:01:46,040 --> 00:01:49,280 Speaker 1: but also mind blowing technologies that have so much impact 28 00:01:49,360 --> 00:01:53,240 Speaker 1: on the way we live. In episode two, developers Reshakish 29 00:01:53,280 --> 00:01:57,600 Speaker 1: and Nihadka protected farm life from pests using Intel's open 30 00:01:57,720 --> 00:02:02,120 Speaker 1: Vino technology. In episode five, Joe and Juan showcase the 31 00:02:02,200 --> 00:02:05,919 Speaker 1: autonomous public transport and its positive impact on the local 32 00:02:05,920 --> 00:02:09,080 Speaker 1: community of Lake Dona. It's been a real honor to 33 00:02:09,120 --> 00:02:12,000 Speaker 1: talk and learn from all of our guests. As we've 34 00:02:12,040 --> 00:02:18,320 Speaker 1: shown throughout this season, AI exists at the intersections of culture, commerce, philosophy, 35 00:02:18,560 --> 00:02:21,760 Speaker 1: and education, and with that it is subject to a 36 00:02:21,960 --> 00:02:25,320 Speaker 1: very critical lens. In this episode, I want to dig 37 00:02:25,360 --> 00:02:28,720 Speaker 1: a little deeper into the relationship between AI itself and 38 00:02:28,760 --> 00:02:33,519 Speaker 1: the developers guiding these projects from development to distribution. Before 39 00:02:33,520 --> 00:02:35,960 Speaker 1: we get into the virtual nuts and bolts of things. 40 00:02:36,280 --> 00:02:41,120 Speaker 1: I want to introduce a very special guest joining me 41 00:02:41,200 --> 00:02:44,800 Speaker 1: now is Intel's Vice President of its Data Center AI 42 00:02:44,880 --> 00:02:48,440 Speaker 1: Solutions Group and General Manager of Data Center AI Solutions 43 00:02:48,440 --> 00:02:52,480 Speaker 1: Strategy and Product Management, Jenny Brovian. She leads a team 44 00:02:52,480 --> 00:02:56,840 Speaker 1: responsible for product strategy, product planning, management, and execution of 45 00:02:56,960 --> 00:03:01,160 Speaker 1: Intel's data center Silicon Software and systems Worry Matt. Jenny 46 00:03:01,160 --> 00:03:04,639 Speaker 1: also serves as an ambassador for Responsible AI, champining the 47 00:03:04,720 --> 00:03:09,200 Speaker 1: ethical development of AI technology and ensuring responsible practices in 48 00:03:09,200 --> 00:03:13,040 Speaker 1: Intel's product development and deployment. I've been really looking forward 49 00:03:13,080 --> 00:03:14,800 Speaker 1: to this discussion. Welcome to the. 50 00:03:14,800 --> 00:03:17,240 Speaker 2: Show, Jenny, Thanks Graham, it's great to be here. 51 00:03:23,040 --> 00:03:26,400 Speaker 1: I will just start really about you, and this is 52 00:03:26,440 --> 00:03:28,760 Speaker 1: a new role for you. Can you tell us a 53 00:03:28,840 --> 00:03:33,120 Speaker 1: little bit more about the responsibilities you've inherited in this 54 00:03:33,160 --> 00:03:35,960 Speaker 1: new role. I mean, perhaps start with what does it 55 00:03:36,000 --> 00:03:38,680 Speaker 1: mean to be the VP of Data Center AI Solutions Group. 56 00:03:39,280 --> 00:03:43,160 Speaker 3: Yeah, So, simply put, I lead the strategy and the 57 00:03:43,200 --> 00:03:48,120 Speaker 3: product portfolio for Intel's data Center AI Accelerator business. If 58 00:03:48,200 --> 00:03:52,000 Speaker 3: we look at the compute power in GPUs and accelerators, 59 00:03:52,040 --> 00:03:55,240 Speaker 3: this is what has really enabled AI to skyrocket and 60 00:03:55,400 --> 00:03:58,520 Speaker 3: these products are a really critical growth area in the 61 00:03:58,560 --> 00:04:02,720 Speaker 3: industry and also for Intel specifically the silicon and systems 62 00:04:02,720 --> 00:04:05,760 Speaker 3: that we build. They operate on a multi year development cycle, 63 00:04:06,120 --> 00:04:08,760 Speaker 3: so at any point I could be working on selling 64 00:04:08,760 --> 00:04:11,920 Speaker 3: products that are already in the market, or path finding 65 00:04:12,200 --> 00:04:14,920 Speaker 3: and defining products that are as far as five years out. 66 00:04:15,520 --> 00:04:17,560 Speaker 3: And my day to day kind of includes what you 67 00:04:17,640 --> 00:04:21,000 Speaker 3: might expect about running a business, managing investments, working with 68 00:04:21,040 --> 00:04:24,960 Speaker 3: our development teams to build our products, developing teams themselves, 69 00:04:25,040 --> 00:04:27,440 Speaker 3: and employees. But by far the most important thing that 70 00:04:27,480 --> 00:04:31,400 Speaker 3: I do with our customers is understanding their needs and 71 00:04:31,480 --> 00:04:35,160 Speaker 3: their goals to build out AI infrastructure that will power 72 00:04:35,320 --> 00:04:35,719 Speaker 3: the world. 73 00:04:36,279 --> 00:04:39,120 Speaker 1: So as part of that data center AI, I have 74 00:04:39,120 --> 00:04:42,040 Speaker 1: a background of micro electronics, so I'm quite interested in 75 00:04:42,080 --> 00:04:45,719 Speaker 1: some of the future in terms of the Intel's AI 76 00:04:45,960 --> 00:04:48,680 Speaker 1: hardware and silicon. Is anything you could tell us a 77 00:04:48,720 --> 00:04:49,440 Speaker 1: little bit about that. 78 00:04:50,000 --> 00:04:53,000 Speaker 3: We definitely are innovating on a broad scale and a 79 00:04:53,080 --> 00:04:55,960 Speaker 3: long life cycle. As I mentioned, we're working many years out, 80 00:04:56,000 --> 00:04:59,479 Speaker 3: and so it's all about learning alongside our customers and 81 00:04:59,520 --> 00:05:02,400 Speaker 3: our partner in what they're sensing in the market and 82 00:05:02,520 --> 00:05:05,240 Speaker 3: what they believe needs to be deployed in the future, 83 00:05:05,279 --> 00:05:08,599 Speaker 3: and this is moving faster than anything I've seen in 84 00:05:09,160 --> 00:05:11,600 Speaker 3: I've been an Intel for over twenty four years, and 85 00:05:11,720 --> 00:05:15,960 Speaker 3: the pace of innovation in AI is tremendous. Most of 86 00:05:16,000 --> 00:05:18,799 Speaker 3: my time at Intel previously was actually in our networking business, 87 00:05:18,960 --> 00:05:21,240 Speaker 3: and that's kind of one of the superpowers that I 88 00:05:21,279 --> 00:05:23,720 Speaker 3: think I bring to this new role. If you look 89 00:05:23,760 --> 00:05:27,520 Speaker 3: at the needs of deploying at scale AI systems, it 90 00:05:27,600 --> 00:05:33,000 Speaker 3: really is about massively interconnected processors and accelerators. So the 91 00:05:33,040 --> 00:05:36,040 Speaker 3: networking is actually just as important as the compute itself. 92 00:05:36,440 --> 00:05:39,560 Speaker 3: So that's definitely an area of innovation that I'm particularly 93 00:05:39,560 --> 00:05:41,080 Speaker 3: interested in driving. 94 00:05:41,640 --> 00:05:44,760 Speaker 1: Okay, and are there any sort of projects you could 95 00:05:45,640 --> 00:05:48,479 Speaker 1: give us examples of some of the innovation at Intel. 96 00:05:48,760 --> 00:05:51,760 Speaker 3: What I'm really excited about are the products that we're 97 00:05:51,800 --> 00:05:56,560 Speaker 3: delivering in our accelerator and our GPU max portfolio, our 98 00:05:56,600 --> 00:06:01,560 Speaker 3: Gouty accelerators. These are powering next generation AI infrastructure, and 99 00:06:02,080 --> 00:06:04,640 Speaker 3: these are products that are available today, but we're working 100 00:06:04,839 --> 00:06:07,400 Speaker 3: on that innovation for the next generation as well. 101 00:06:07,839 --> 00:06:10,960 Speaker 1: That's great. And if you could explain really the critical 102 00:06:11,600 --> 00:06:16,880 Speaker 1: aspect of this GPU hardware graphics processing unit and why 103 00:06:17,000 --> 00:06:21,760 Speaker 1: is it so important for the AI technology they use 104 00:06:21,800 --> 00:06:26,200 Speaker 1: like chatbots and AI assistance. Why is it so critical. 105 00:06:26,400 --> 00:06:30,520 Speaker 3: When you look at the unique needs of AI computing 106 00:06:30,560 --> 00:06:34,680 Speaker 3: by comparison to more traditional general purpose computing. It really 107 00:06:34,800 --> 00:06:37,760 Speaker 3: is about massive scale and the types of processing that 108 00:06:37,800 --> 00:06:41,640 Speaker 3: we're talking about is massively parallel processing, and so that's 109 00:06:41,680 --> 00:06:46,400 Speaker 3: really what the GPUs bring by comparison to general purpose processors. 110 00:06:46,520 --> 00:06:49,800 Speaker 3: General purpose processors also deliver a tremendous amount of value 111 00:06:49,920 --> 00:06:53,080 Speaker 3: in AI computing because very often what's happening is you're 112 00:06:53,120 --> 00:06:56,200 Speaker 3: serving multiple workloads at the same time. And so what's 113 00:06:56,240 --> 00:06:58,800 Speaker 3: great about working at a company like Intel is that 114 00:06:58,839 --> 00:07:02,000 Speaker 3: there's a variety ofsets that we bring for all different 115 00:07:02,000 --> 00:07:05,240 Speaker 3: types of AI infrastructure that customers are seeking to build out, 116 00:07:05,440 --> 00:07:08,960 Speaker 3: whether we're talking about large scale deployments and centralized data 117 00:07:09,000 --> 00:07:11,720 Speaker 3: centers as well as infrastructure that extends all the way 118 00:07:11,760 --> 00:07:13,880 Speaker 3: out to the edge as well as to the client. 119 00:07:13,960 --> 00:07:16,720 Speaker 3: And so it really is about that continuum of AI 120 00:07:16,760 --> 00:07:20,600 Speaker 3: compute that really needs to be powered by both this 121 00:07:20,800 --> 00:07:25,280 Speaker 3: massively parallel graphics processor units, but also by different types 122 00:07:25,320 --> 00:07:28,440 Speaker 3: of compute depending upon the deployment size and the use case. 123 00:07:29,040 --> 00:07:32,760 Speaker 1: Okay, and we've already discussed in previous episodes on this 124 00:07:32,880 --> 00:07:38,920 Speaker 1: show some pretty excellent and really interesting AI deployments using 125 00:07:39,040 --> 00:07:43,160 Speaker 1: Inteller's partners. We also talked a little bit about the ethical, 126 00:07:43,280 --> 00:07:49,200 Speaker 1: responsible nature of developers when they're creating these sorts of systems. 127 00:07:50,160 --> 00:07:53,480 Speaker 1: What steps are Intel taking to keep these sorts of 128 00:07:53,520 --> 00:07:57,280 Speaker 1: AI implementations responsible and secure. 129 00:07:57,880 --> 00:08:02,880 Speaker 3: Yeah, it's really important as weight and deploy AI technology 130 00:08:03,280 --> 00:08:05,640 Speaker 3: that we look at the positive outcomes that we're trying 131 00:08:05,640 --> 00:08:09,800 Speaker 3: to create. So this really is about creating positive global change, 132 00:08:10,280 --> 00:08:14,480 Speaker 3: about empowering people with all kinds of new tools, and 133 00:08:14,520 --> 00:08:18,560 Speaker 3: really improving everyone's lives around the planet. So those are 134 00:08:18,560 --> 00:08:20,520 Speaker 3: the positive outcomes that we can create, but we also 135 00:08:20,560 --> 00:08:23,240 Speaker 3: need to make sure that we balance that with a 136 00:08:23,320 --> 00:08:27,280 Speaker 3: full approach to lower the risks and also optimize the benefit. 137 00:08:27,320 --> 00:08:29,800 Speaker 3: So it's about managing both sides of the equation. So 138 00:08:30,160 --> 00:08:34,080 Speaker 3: Intel has a tremendous amount of work underway in responsible AI. 139 00:08:34,280 --> 00:08:37,160 Speaker 3: We've been working on this for several years, and this 140 00:08:37,240 --> 00:08:41,720 Speaker 3: initiative that we have has really evolved to include very 141 00:08:41,760 --> 00:08:47,640 Speaker 3: structured and rigorous and multidisciplinary processes to advance AI technology 142 00:08:47,679 --> 00:08:51,240 Speaker 3: responsibly all the way from that entire product development life 143 00:08:51,240 --> 00:08:54,680 Speaker 3: cycle that I was talking about, from development through deployment, 144 00:08:55,160 --> 00:08:58,200 Speaker 3: and we have a whole frameworket INTEL focused on global 145 00:08:58,280 --> 00:09:02,120 Speaker 3: human rights principles and SOPs very well to that we're 146 00:09:02,120 --> 00:09:04,920 Speaker 3: looking at this really from a multidisciplinary approach. We have 147 00:09:04,960 --> 00:09:09,880 Speaker 3: a responsible AI Advisory Council that reviews these goals throughout 148 00:09:09,880 --> 00:09:12,440 Speaker 3: the life cycle of an AI project, and we look 149 00:09:12,480 --> 00:09:16,160 Speaker 3: at potential ethical risks within these projects and actively mitigate 150 00:09:16,200 --> 00:09:19,120 Speaker 3: those risks as early as possible. And really our goal 151 00:09:19,360 --> 00:09:22,199 Speaker 3: is to actively manage it, but also be transparent about 152 00:09:22,240 --> 00:09:26,080 Speaker 3: our position and our practices so that we can really 153 00:09:26,160 --> 00:09:30,439 Speaker 3: address and advance solutions for shared challenges across the industry, 154 00:09:30,840 --> 00:09:33,679 Speaker 3: so that we're not only just improving our own products, 155 00:09:33,760 --> 00:09:37,079 Speaker 3: but we're improving what's happening across the industry as well. 156 00:09:37,200 --> 00:09:41,560 Speaker 3: And so we also contribute to various standards and methods 157 00:09:41,880 --> 00:09:45,480 Speaker 3: at the national level, the international level. There are organizations 158 00:09:45,640 --> 00:09:49,280 Speaker 3: like the Business Roundtable on Human Rights and AI, Global 159 00:09:49,360 --> 00:09:52,760 Speaker 3: Business Initiative on Human Rights, the Partnership on AI, and 160 00:09:52,840 --> 00:09:55,280 Speaker 3: so this is really an opportunity for us to come 161 00:09:55,320 --> 00:10:01,240 Speaker 3: together with our peers to establish these parameters across ethical 162 00:10:01,240 --> 00:10:05,040 Speaker 3: and moral and privacy standards so that we can build 163 00:10:05,120 --> 00:10:09,200 Speaker 3: thriving businesses and innovate really exciting technology, but do so 164 00:10:09,240 --> 00:10:10,240 Speaker 3: in a responsible way. 165 00:10:10,640 --> 00:10:14,360 Speaker 1: Our audience may have heard about the various open source 166 00:10:14,440 --> 00:10:17,719 Speaker 1: or closed source AI models. I'd like to get your 167 00:10:17,800 --> 00:10:23,240 Speaker 1: thoughts about Intel's approach to this perennial open versus closed 168 00:10:23,360 --> 00:10:25,920 Speaker 1: source approach to development. 169 00:10:26,640 --> 00:10:30,319 Speaker 3: Yeah, this is a mission that's really essential to Intel. 170 00:10:30,600 --> 00:10:35,240 Speaker 3: We look at this open approach and open ecosystem priority 171 00:10:35,320 --> 00:10:40,360 Speaker 3: as essential to lowering barriers to entry and unlocking AI 172 00:10:40,440 --> 00:10:45,840 Speaker 3: innovation for developers and for customers, and we're really focused 173 00:10:45,880 --> 00:10:50,200 Speaker 3: on accelerating and open AI software ecosystem specifically that is 174 00:10:50,280 --> 00:10:54,760 Speaker 3: needed to break down proprietary walled gardens and closed approaches. 175 00:10:55,200 --> 00:10:58,160 Speaker 3: We really believe that open ecosystems are more powerful than 176 00:10:58,240 --> 00:11:02,280 Speaker 3: closed ecosystems. They drive a higher level of innovation, and 177 00:11:02,600 --> 00:11:05,920 Speaker 3: they focus on democratizing computing. And we certainly have a 178 00:11:05,960 --> 00:11:08,360 Speaker 3: long history as a company in this area and it's 179 00:11:08,400 --> 00:11:10,640 Speaker 3: been true for many years and other areas of computing, 180 00:11:10,679 --> 00:11:14,160 Speaker 3: and it's definitely true for AI as well. We saw 181 00:11:14,160 --> 00:11:17,960 Speaker 3: the industry move from centralized computing to decentralized computing and 182 00:11:18,080 --> 00:11:21,400 Speaker 3: back to centralized computing with the cloud, and we believe 183 00:11:21,440 --> 00:11:24,880 Speaker 3: that the pendulum of computing's next swing to the edge 184 00:11:24,960 --> 00:11:27,560 Speaker 3: is now also underway. I talked about that end to 185 00:11:27,720 --> 00:11:30,800 Speaker 3: end AI deployment across the entire spectrum, and so as 186 00:11:30,800 --> 00:11:33,040 Speaker 3: we look at all these different areas, we have a 187 00:11:33,160 --> 00:11:38,160 Speaker 3: proven track record of working with open ecosystems, building open ecosystems, 188 00:11:38,480 --> 00:11:43,200 Speaker 3: partnering with software vendors and developers to scale technologies, and 189 00:11:43,240 --> 00:11:46,680 Speaker 3: we're definitely committed to continuing this legacy with AI, and 190 00:11:46,720 --> 00:11:49,160 Speaker 3: we really believe that to realize the full benefits that 191 00:11:49,200 --> 00:11:52,839 Speaker 3: we've been talking about and focusing on delivering AI everywhere, 192 00:11:53,200 --> 00:11:56,920 Speaker 3: it truly means AI everywhere to everyone, to developers, to 193 00:11:57,000 --> 00:12:00,000 Speaker 3: all your devices, to all types of compute, to all 194 00:12:00,040 --> 00:12:02,480 Speaker 3: all of your use cases in business models, and it's 195 00:12:02,520 --> 00:12:06,560 Speaker 3: an open ecosystem and open source software that are foundational 196 00:12:06,600 --> 00:12:08,960 Speaker 3: to ensuring that this can be fully accelerated. 197 00:12:09,640 --> 00:12:11,720 Speaker 1: You talked a little bit as well about some of 198 00:12:11,760 --> 00:12:15,719 Speaker 1: the tools that Intel are supporting. We've had earlier episodes 199 00:12:16,000 --> 00:12:19,760 Speaker 1: leveraging the open Veno platform, and there's also like some 200 00:12:19,840 --> 00:12:23,640 Speaker 1: other AI software tools OneAPI, and also support for the 201 00:12:23,679 --> 00:12:28,240 Speaker 1: open source projects like PyTorch and hiking Phase. Perhaps you 202 00:12:28,240 --> 00:12:31,560 Speaker 1: could explain to us a little bit about these tools 203 00:12:31,760 --> 00:12:33,439 Speaker 1: while they're so useful for developers. 204 00:12:34,080 --> 00:12:36,200 Speaker 3: Yeah, I'll talk about a few of these because it 205 00:12:36,280 --> 00:12:41,040 Speaker 3: really is about unleashing the power of developers to realize 206 00:12:41,160 --> 00:12:44,920 Speaker 3: the power of AI technology. We're creating this foundation in 207 00:12:45,000 --> 00:12:47,559 Speaker 3: our silicon and our systems and our software, but it's 208 00:12:47,600 --> 00:12:50,439 Speaker 3: really about empowering developers. So if we look at a 209 00:12:50,440 --> 00:12:51,800 Speaker 3: few of these, I'm going to start a little bit 210 00:12:51,800 --> 00:12:54,480 Speaker 3: actually with one API and looking close to the hardware. 211 00:12:54,920 --> 00:12:58,559 Speaker 3: We have a OneAPI programming model that's really the software 212 00:12:58,559 --> 00:13:02,280 Speaker 3: foundation of our overall AI strategy. So this is really 213 00:13:02,440 --> 00:13:06,840 Speaker 3: about providing a standard programming model that allows these developers 214 00:13:06,880 --> 00:13:10,680 Speaker 3: to get value across multiple different hardware architectures. I talked 215 00:13:10,720 --> 00:13:14,600 Speaker 3: before about the fact that you've got heterogeneous hardware architectures 216 00:13:14,600 --> 00:13:18,040 Speaker 3: depending upon the type of AI solution that you're developing 217 00:13:18,200 --> 00:13:21,760 Speaker 3: and deploying, and so being able to access the value 218 00:13:21,800 --> 00:13:24,920 Speaker 3: of these different hardware types from one set of source 219 00:13:24,960 --> 00:13:28,680 Speaker 3: code is really crucial. It's a really critical aspect of 220 00:13:28,720 --> 00:13:33,080 Speaker 3: our goal to democratize AI and allow developers to access 221 00:13:33,120 --> 00:13:36,280 Speaker 3: all of this to build and deploy AI solutions everywhere. 222 00:13:36,679 --> 00:13:39,240 Speaker 3: So if we look at the different stages of AI 223 00:13:39,480 --> 00:13:42,880 Speaker 3: development in terms of both development of the model and 224 00:13:43,000 --> 00:13:46,120 Speaker 3: treatment of the data throughout the entire life cycle. One 225 00:13:46,160 --> 00:13:49,839 Speaker 3: of the really fastest growing areas of AI development is 226 00:13:49,880 --> 00:13:52,480 Speaker 3: in fine tuning. So there's so many models out there. 227 00:13:52,480 --> 00:13:55,920 Speaker 3: You mentioned hugging face, right, that's a massive community for models, 228 00:13:56,200 --> 00:13:58,320 Speaker 3: and it's just again the pace of innovation there is 229 00:13:58,360 --> 00:14:01,840 Speaker 3: absolutely huge. So if you choose a pre train model, 230 00:14:02,200 --> 00:14:03,760 Speaker 3: that's a great way to get ahead start on a 231 00:14:03,760 --> 00:14:06,920 Speaker 3: solution that you're trying to build and ultimately means faster 232 00:14:07,040 --> 00:14:09,560 Speaker 3: timed insights for your enterprise. And so you can take 233 00:14:09,559 --> 00:14:11,960 Speaker 3: a pre train model and then it's really okay, what 234 00:14:12,000 --> 00:14:14,199 Speaker 3: do you do to get ready for what you're trying 235 00:14:14,240 --> 00:14:17,560 Speaker 3: to do. So this aspect of fine tuning it, bringing 236 00:14:17,600 --> 00:14:20,280 Speaker 3: in your own data and making sure that it's ready 237 00:14:20,320 --> 00:14:24,040 Speaker 3: for your application, it's a really set of crucial steps 238 00:14:24,080 --> 00:14:27,720 Speaker 3: to accelerate that development and deployment life cycle. Then once 239 00:14:27,760 --> 00:14:29,840 Speaker 3: you do that fine tuning, right, you've got your data, 240 00:14:29,960 --> 00:14:32,400 Speaker 3: You've fine tuned your model. Now it's about okay, how 241 00:14:32,440 --> 00:14:35,000 Speaker 3: do I deploy that and get that ready for inferencing. 242 00:14:35,400 --> 00:14:37,080 Speaker 3: So obviously you've got to worry not only about the 243 00:14:37,120 --> 00:14:40,800 Speaker 3: deployment but updates across the various locations where you've deployed 244 00:14:40,800 --> 00:14:43,320 Speaker 3: the model. And this is really where open Veno comes in, right. 245 00:14:43,400 --> 00:14:45,720 Speaker 3: The open Vino tool cloud can really be a good 246 00:14:45,760 --> 00:14:49,840 Speaker 3: asset in deploying all sorts of AI technology, but specifically 247 00:14:50,040 --> 00:14:53,960 Speaker 3: vision and natural language processing models across a variety of 248 00:14:53,960 --> 00:14:56,600 Speaker 3: different types of deployment targets. And if we look maybe 249 00:14:56,600 --> 00:14:59,680 Speaker 3: a little bit more broadly across other tools, we've got 250 00:14:59,680 --> 00:15:03,360 Speaker 3: libr that enable the AI ecosystem across a variety of 251 00:15:03,400 --> 00:15:08,480 Speaker 3: toolkits and optimized software and libraries and frameworks, particularly ensuring 252 00:15:08,520 --> 00:15:12,480 Speaker 3: that the popular frameworks are optimized for what developers are seeking. 253 00:15:12,600 --> 00:15:15,360 Speaker 3: So what this is important for is enabling orders of 254 00:15:15,400 --> 00:15:21,480 Speaker 3: magnitude performance improvements and increasing productivity for developers across AI workloads. 255 00:15:21,480 --> 00:15:23,920 Speaker 3: So if we look across all of these different tools, 256 00:15:24,400 --> 00:15:27,280 Speaker 3: whether it's you know, the libraries, one API, open Veno 257 00:15:27,640 --> 00:15:30,760 Speaker 3: optimizations for these frameworks, our goal is really to give 258 00:15:31,040 --> 00:15:34,320 Speaker 3: developers the openness and the choice that they need with 259 00:15:34,560 --> 00:15:37,960 Speaker 3: all of the hardware architectures that they're using, and making 260 00:15:37,960 --> 00:15:40,680 Speaker 3: them easy to program and easy to access the power 261 00:15:40,800 --> 00:15:44,040 Speaker 3: of that underlying hardware and really ideally with one code 262 00:15:44,080 --> 00:15:45,920 Speaker 3: base across many different architectures. 263 00:15:46,280 --> 00:15:51,280 Speaker 1: Yeah, our guess. Previously there weren't traditionally AI engineers or 264 00:15:51,320 --> 00:15:53,440 Speaker 1: AI developers, And I think one of the powers of 265 00:15:53,520 --> 00:15:55,560 Speaker 1: the tools that you were just talking about is that 266 00:15:55,880 --> 00:15:58,520 Speaker 1: you can be a developer, but not specifically an AI developer, 267 00:15:58,560 --> 00:16:00,360 Speaker 1: and this will give you a leg up to actually 268 00:16:00,400 --> 00:16:03,520 Speaker 1: start innovating and start experimenting with all of these models. 269 00:16:04,040 --> 00:16:04,240 Speaker 2: Yeah. 270 00:16:04,240 --> 00:16:05,840 Speaker 3: I think that's a really good point. And really the 271 00:16:05,880 --> 00:16:08,960 Speaker 3: developer landscape is changing. As you said, when you look 272 00:16:09,040 --> 00:16:12,880 Speaker 3: at the entire population of developers that we are seeking 273 00:16:12,920 --> 00:16:16,320 Speaker 3: to reach, something like eighty percent of developers operate at 274 00:16:16,320 --> 00:16:20,400 Speaker 3: the framework level and above. So meeting developers where they 275 00:16:20,440 --> 00:16:24,120 Speaker 3: are is absolutely crucial to our strategy. And so there 276 00:16:24,120 --> 00:16:26,520 Speaker 3: are those who will operate at that level and never 277 00:16:26,680 --> 00:16:29,520 Speaker 3: directly access the hardware, but they need to be able 278 00:16:29,560 --> 00:16:33,000 Speaker 3: to unleash the power of that hardware for the applications 279 00:16:33,000 --> 00:16:36,560 Speaker 3: that they're building, and so ensuring that no matter where 280 00:16:36,600 --> 00:16:39,520 Speaker 3: they are, we have the tools in order to empower 281 00:16:39,560 --> 00:16:42,320 Speaker 3: them to achieve those goals, that's really our objective. 282 00:16:42,840 --> 00:16:46,680 Speaker 1: Yeah, and this leads onto the Intel Developer Cloud initiative 283 00:16:46,880 --> 00:16:50,520 Speaker 1: that has Intel's latest hardware and software for testing and building. 284 00:16:50,640 --> 00:16:53,680 Speaker 1: Can you explain a little bit about that and the 285 00:16:53,720 --> 00:16:55,520 Speaker 1: importance for the developer community. 286 00:16:56,200 --> 00:16:59,800 Speaker 3: Yeah, I mentioned getting access to all of that great hardware, right, 287 00:17:00,080 --> 00:17:03,120 Speaker 3: Developers really they want the most performance solutions, and for 288 00:17:03,160 --> 00:17:05,439 Speaker 3: that they need to have access to the latest and 289 00:17:05,480 --> 00:17:08,840 Speaker 3: greatest AI platforms. And so that's really what Intel Developer 290 00:17:08,840 --> 00:17:12,880 Speaker 3: Cloud provides. It's access to current platforms, and it's also 291 00:17:12,920 --> 00:17:16,639 Speaker 3: access to new and future platforms that may be available 292 00:17:16,920 --> 00:17:21,120 Speaker 3: prior to their launch for initial development and testing. And 293 00:17:21,200 --> 00:17:23,960 Speaker 3: so this is something that we've been building over time, 294 00:17:24,000 --> 00:17:28,040 Speaker 3: and we've expanded it to provide developers and partners early 295 00:17:28,320 --> 00:17:32,399 Speaker 3: and efficient access to a variety of Intel products and technologies, 296 00:17:32,800 --> 00:17:35,000 Speaker 3: in many cases from a few months all the way 297 00:17:35,080 --> 00:17:38,040 Speaker 3: up to a year in advance of full product availability. 298 00:17:38,320 --> 00:17:41,640 Speaker 3: And what's important about cloud based development is it's the 299 00:17:41,680 --> 00:17:44,800 Speaker 3: best way to get access to clusters of systems for 300 00:17:44,880 --> 00:17:47,320 Speaker 3: at scale development and testing. I talked a little bit 301 00:17:47,440 --> 00:17:50,359 Speaker 3: before about the fact that when you look at large 302 00:17:50,359 --> 00:17:52,960 Speaker 3: scale AI deployments, it's not just about a single node, 303 00:17:52,960 --> 00:17:55,199 Speaker 3: it's not just about a single processor. It really is 304 00:17:55,320 --> 00:17:59,160 Speaker 3: about clusters and large scale development, and so being able 305 00:17:59,200 --> 00:18:02,920 Speaker 3: to access clusters in a cloud based environment is really 306 00:18:02,920 --> 00:18:05,560 Speaker 3: important for developers to be able to not just do 307 00:18:05,600 --> 00:18:09,320 Speaker 3: that initial development and testing, but actually ultimately be able 308 00:18:09,320 --> 00:18:11,840 Speaker 3: to test it at scale. And so if we look 309 00:18:11,880 --> 00:18:14,560 Speaker 3: specifically at some of the products that are available in 310 00:18:14,560 --> 00:18:18,600 Speaker 3: Intel Developer Cloud, it includes our Xeon processors, so fourth 311 00:18:18,640 --> 00:18:22,720 Speaker 3: gen and fifth gen Intel zon scalable processors. It includes 312 00:18:22,840 --> 00:18:26,600 Speaker 3: our Xeon Mac series, also our GPU Mac series processors, 313 00:18:26,640 --> 00:18:30,399 Speaker 3: our Zon D processors, our Gouty accelerators, and our GPU 314 00:18:30,480 --> 00:18:33,439 Speaker 3: Flex series. So really what we're trying to do is 315 00:18:33,520 --> 00:18:37,720 Speaker 3: provide developers with that full breadth of solutions that meet 316 00:18:37,760 --> 00:18:40,600 Speaker 3: the needs of AI solutions that they're seeking to develop 317 00:18:40,640 --> 00:18:43,359 Speaker 3: across that entire spectrum. There was an actually really great 318 00:18:43,359 --> 00:18:46,680 Speaker 3: example that I heard from our peers in Intel Developer 319 00:18:46,680 --> 00:18:52,280 Speaker 3: Cloud about a professor at Kansas State University who was 320 00:18:52,320 --> 00:18:56,600 Speaker 3: specifically working on COVID research and so he was using 321 00:18:56,880 --> 00:19:01,760 Speaker 3: machine learning techniques enabled by one API on Intel Developer Cloud, 322 00:19:01,840 --> 00:19:04,800 Speaker 3: and his goal was to focus initially on drug candidates 323 00:19:05,200 --> 00:19:10,000 Speaker 3: for clinical studies and experimental drugs, and he was able to, 324 00:19:10,480 --> 00:19:14,280 Speaker 3: through access to Intel Developer Cloud, use that compute power 325 00:19:14,320 --> 00:19:17,200 Speaker 3: that was available in that deployment in order to drive 326 00:19:17,240 --> 00:19:19,720 Speaker 3: breakthroughs in his medical research. And so, you know, It 327 00:19:19,760 --> 00:19:22,520 Speaker 3: definitely is about our direct customers to the technology and 328 00:19:22,560 --> 00:19:25,479 Speaker 3: giving them access to our latest generation products, but it's 329 00:19:25,520 --> 00:19:29,000 Speaker 3: also about users of technology across the entire spectrum, including 330 00:19:29,359 --> 00:19:31,640 Speaker 3: key partners and research driving innovation and AI. 331 00:19:34,560 --> 00:19:48,120 Speaker 1: We'll be right back after a quick break. Welcome back 332 00:19:48,200 --> 00:19:55,800 Speaker 1: to Technically Speaking and Intel podcast. You have a really 333 00:19:55,800 --> 00:19:59,199 Speaker 1: interesting project with the Agon National Laboratory. Can you tell 334 00:19:59,280 --> 00:19:59,880 Speaker 1: us a little. 335 00:19:59,640 --> 00:20:04,320 Speaker 3: Bit about Yeah, this is a multi year collaboration and 336 00:20:04,400 --> 00:20:07,280 Speaker 3: I'm really excited actually about the progress we've made and 337 00:20:07,320 --> 00:20:09,440 Speaker 3: I'd love to talk about some of our innovation there. 338 00:20:09,560 --> 00:20:13,880 Speaker 3: So Argon National Lab is a leading research center in 339 00:20:14,000 --> 00:20:17,240 Speaker 3: the US, and it's really focused on being on the 340 00:20:17,280 --> 00:20:22,560 Speaker 3: forefront of our nation's efforts to deliver excescale computing capabilities 341 00:20:22,600 --> 00:20:27,040 Speaker 3: to advance science. And they've got a specific project around GENAI, 342 00:20:27,400 --> 00:20:30,919 Speaker 3: which is a collaboration between Argon and Intel and a 343 00:20:31,000 --> 00:20:34,399 Speaker 3: number of other partners to enable the power of GENAI 344 00:20:34,560 --> 00:20:38,520 Speaker 3: to create state of the art AI models specifically targeted 345 00:20:38,600 --> 00:20:41,320 Speaker 3: at science. So these are models that are being trained 346 00:20:41,320 --> 00:20:45,240 Speaker 3: on scientific techts and code and science data sets from 347 00:20:45,400 --> 00:20:49,040 Speaker 3: a whole bunch of different diverse scientific domains. And so 348 00:20:49,359 --> 00:20:53,680 Speaker 3: if we look at using these foundational technologies, and we're 349 00:20:53,800 --> 00:20:57,160 Speaker 3: unleashing the power of the hardware but focused on technologies 350 00:20:57,280 --> 00:21:00,399 Speaker 3: called Megatron, and we're using the deep speed frame works. 351 00:21:00,760 --> 00:21:03,639 Speaker 3: This Genai product is going to serve us a number 352 00:21:03,640 --> 00:21:07,040 Speaker 3: of different scientific disciplines. If you look across biology and 353 00:21:07,240 --> 00:21:10,840 Speaker 3: cancer research, and climate science, cosmology, material science, the list 354 00:21:10,960 --> 00:21:12,959 Speaker 3: really goes on and on when you talk about a 355 00:21:12,960 --> 00:21:16,400 Speaker 3: machine of this size. So we really have an opportunity 356 00:21:16,440 --> 00:21:20,720 Speaker 3: to work on really hard, large scale problems. And if 357 00:21:20,760 --> 00:21:24,120 Speaker 3: we look at the types of problems that are slated 358 00:21:24,200 --> 00:21:27,639 Speaker 3: for initial deployment on the Aurora supercomputer, it's areas like 359 00:21:28,080 --> 00:21:33,480 Speaker 3: developing safe and clean fusion reactors, neuroscience research, understanding the 360 00:21:33,560 --> 00:21:38,000 Speaker 3: dark universe, designing more fuel efficient aircraft. These are problems 361 00:21:38,000 --> 00:21:40,680 Speaker 3: that require an intensive amount of compute and an intensive 362 00:21:40,680 --> 00:21:43,919 Speaker 3: amount of data, and so it really is about a 363 00:21:44,040 --> 00:21:46,639 Speaker 3: level of scale of the number of parameters that are 364 00:21:46,640 --> 00:21:50,120 Speaker 3: supported by the model, far in excess of any computing 365 00:21:50,119 --> 00:21:52,920 Speaker 3: that's deployed today, and so it's really important for us 366 00:21:53,040 --> 00:21:55,600 Speaker 3: to partner with Aurora to prove out the ability of 367 00:21:55,640 --> 00:21:58,439 Speaker 3: this technology to be deployed at scale. And so what 368 00:21:58,480 --> 00:22:01,040 Speaker 3: we announced last month that the super computing twenty three 369 00:22:01,080 --> 00:22:05,440 Speaker 3: conference in partnership with Ragon was starting with a one 370 00:22:05,520 --> 00:22:10,480 Speaker 3: trillion parameter GBT three large language model on the Aurora supercomputer. 371 00:22:10,960 --> 00:22:13,200 Speaker 3: And the way that we were able to support that 372 00:22:13,400 --> 00:22:16,439 Speaker 3: very large number of parameters was to be able to 373 00:22:16,520 --> 00:22:21,000 Speaker 3: leverage the underlying capabilities of our Intel Mac series GPUs. 374 00:22:21,240 --> 00:22:24,880 Speaker 3: As I mentioned before, those GPUs are essential to driving 375 00:22:25,080 --> 00:22:28,480 Speaker 3: massively parallel compute to be able to compute that tremendously 376 00:22:28,600 --> 00:22:30,720 Speaker 3: large model and large number of parameters. 377 00:22:30,920 --> 00:22:31,920 Speaker 2: But we didn't stop there. 378 00:22:32,200 --> 00:22:34,919 Speaker 3: It was crucial to prove out the ability to handle 379 00:22:34,920 --> 00:22:37,639 Speaker 3: one trillion parameters, which we did on sixty four nodes, 380 00:22:37,680 --> 00:22:40,359 Speaker 3: which was far fewer than typically would be required, but 381 00:22:40,480 --> 00:22:43,480 Speaker 3: extend beyond that scale. Beyond that, so we ran four 382 00:22:43,520 --> 00:22:46,760 Speaker 3: instances on two hundred and fifty six nodes which demonstrated 383 00:22:46,800 --> 00:22:50,440 Speaker 3: the ability to pave the path to scale to training 384 00:22:50,560 --> 00:22:54,480 Speaker 3: trillions of parameter models with trillions of tokens on more 385 00:22:54,520 --> 00:22:55,639 Speaker 3: than ten thousand nodes. 386 00:22:55,680 --> 00:22:58,320 Speaker 2: That's really our ultimate goal. So when you think. 387 00:22:58,160 --> 00:23:01,720 Speaker 3: About workloads that fall in these different domains that I 388 00:23:01,760 --> 00:23:05,480 Speaker 3: was talking about before, so things like modeling complicated chemical 389 00:23:05,520 --> 00:23:10,880 Speaker 3: processes in drug design, or enabling efficient screening of massive 390 00:23:10,960 --> 00:23:14,920 Speaker 3: chemical data sets to focus on innovation and molecular science. 391 00:23:15,640 --> 00:23:19,840 Speaker 3: This is the performance necessary to tackle these massive problems 392 00:23:19,840 --> 00:23:23,480 Speaker 3: and some of the most important questions that are facing science, 393 00:23:23,520 --> 00:23:27,639 Speaker 3: but ultimately really humanity today. And so it's really this 394 00:23:27,720 --> 00:23:31,560 Speaker 3: work that we're driving between Intel and Argon National Laboratory, 395 00:23:31,920 --> 00:23:34,680 Speaker 3: in which we've already started to prove that we've got 396 00:23:34,720 --> 00:23:39,440 Speaker 3: these tangible examples of how people can really expect unprecedented 397 00:23:39,520 --> 00:23:43,960 Speaker 3: levels of innovation and groundbreaking research focused on this foundation 398 00:23:44,240 --> 00:23:45,560 Speaker 3: of exoscale computing. 399 00:23:46,440 --> 00:23:49,680 Speaker 1: So we talked about the large models, maybe talk a 400 00:23:49,720 --> 00:23:52,639 Speaker 1: little bit about some of the smaller AI models. What 401 00:23:52,680 --> 00:23:55,240 Speaker 1: are your thoughts on that and how it's currently being used. 402 00:23:55,960 --> 00:23:59,000 Speaker 3: The airtime really is on a lot of the large models, 403 00:23:59,040 --> 00:24:01,080 Speaker 3: isn't it. And certainly there's a tremendous amount of innovation 404 00:24:01,160 --> 00:24:04,280 Speaker 3: that's happening there. But they're very costly to deploy, train, 405 00:24:04,560 --> 00:24:08,960 Speaker 3: and operate, and it's got tremendous potential and promise from 406 00:24:09,080 --> 00:24:12,399 Speaker 3: everyone who is developing them and ultimately using them, but 407 00:24:12,600 --> 00:24:14,399 Speaker 3: we need to make sure we solve the problems of 408 00:24:14,520 --> 00:24:18,560 Speaker 3: access and cost as well. And so really what we 409 00:24:18,720 --> 00:24:21,040 Speaker 3: want to look at also optimizing is not just those 410 00:24:21,119 --> 00:24:22,919 Speaker 3: large models, but also the other end of the spectrum, 411 00:24:22,960 --> 00:24:25,520 Speaker 3: as you said, and this is where we see even 412 00:24:25,560 --> 00:24:29,160 Speaker 3: more action, more new developments, more new use cases. That's 413 00:24:29,160 --> 00:24:32,720 Speaker 3: why these communities that are driving model innovation. Hugging faces 414 00:24:32,800 --> 00:24:35,960 Speaker 3: is a great example, is really accelerating this end of 415 00:24:36,000 --> 00:24:40,200 Speaker 3: the spectrum. And really smaller models are what enable a 416 00:24:40,280 --> 00:24:43,280 Speaker 3: much larger number of developers to make AI come to 417 00:24:43,359 --> 00:24:47,280 Speaker 3: life where they're focused on developing new applications, where companies 418 00:24:47,320 --> 00:24:51,840 Speaker 3: are extracting value by building new business models around AI deployments. 419 00:24:52,400 --> 00:24:55,080 Speaker 3: And one of the fastest growing portions of just the 420 00:24:55,119 --> 00:24:59,919 Speaker 3: overall AI development and data flow is looking at deployment specifically. 421 00:25:00,080 --> 00:25:02,679 Speaker 3: And one of the most interesting challenges in driving this 422 00:25:02,760 --> 00:25:06,040 Speaker 3: wider range of deployments is making these models smaller but 423 00:25:06,200 --> 00:25:09,719 Speaker 3: still delivering the same level of accuracy. And so you 424 00:25:09,800 --> 00:25:13,919 Speaker 3: start to look at deployment areas like smartphones and smart 425 00:25:13,920 --> 00:25:22,840 Speaker 3: home devices, to surveillance and industrial, IoT sports, and entertainment domains, manufacturing, transportation. 426 00:25:22,960 --> 00:25:25,640 Speaker 3: There's so many different domains and this is an area 427 00:25:25,640 --> 00:25:27,960 Speaker 3: that I'm particularly really excited about. I talked about you 428 00:25:28,000 --> 00:25:30,679 Speaker 3: know how, I came from the networking business at Intel 429 00:25:31,040 --> 00:25:33,919 Speaker 3: and also focused a great deal on edge computing, and 430 00:25:34,040 --> 00:25:37,720 Speaker 3: these are areas where we're not talking about at scale 431 00:25:38,119 --> 00:25:40,960 Speaker 3: AI computing, but we are absolutely talking about needing to 432 00:25:41,040 --> 00:25:43,560 Speaker 3: unleash the power of AI, and so these are areas 433 00:25:43,600 --> 00:25:46,639 Speaker 3: where those smaller AI models are essential. And if we 434 00:25:46,680 --> 00:25:49,800 Speaker 3: just look at one area in transportation, airlines are exploring 435 00:25:50,200 --> 00:25:53,960 Speaker 3: the use of AI copilots to help do things like 436 00:25:54,000 --> 00:25:58,960 Speaker 3: suggest a better altitude to prevent creating contrails, to meet 437 00:25:59,320 --> 00:26:03,200 Speaker 3: climate and sstainability objectives. If we look in the entertainment industry, 438 00:26:03,520 --> 00:26:07,359 Speaker 3: game developers are using smaller AI models that are based 439 00:26:07,400 --> 00:26:11,280 Speaker 3: on gestures and conversation data, you know, as gamers interact 440 00:26:11,640 --> 00:26:15,359 Speaker 3: to build more lifelike game characters. So the opportunities are 441 00:26:15,400 --> 00:26:18,520 Speaker 3: really endless as we look across both business use cases 442 00:26:18,560 --> 00:26:21,160 Speaker 3: as well as consumer use cases, and you can really 443 00:26:21,200 --> 00:26:24,680 Speaker 3: see in these examples where the needs of those small 444 00:26:24,760 --> 00:26:28,120 Speaker 3: models need to continue to grow and where we really 445 00:26:28,119 --> 00:26:30,280 Speaker 3: need to focus on driving innovation at that end of 446 00:26:30,280 --> 00:26:31,120 Speaker 3: the spectrum as well. 447 00:26:31,600 --> 00:26:33,880 Speaker 1: Do you think in the future, you know, you might 448 00:26:33,920 --> 00:26:37,400 Speaker 1: have the AI small low models like on everyone's desktop 449 00:26:37,440 --> 00:26:37,880 Speaker 1: sort of thing. 450 00:26:38,480 --> 00:26:40,680 Speaker 3: For sure, we're already there. 451 00:26:40,720 --> 00:26:42,480 Speaker 2: So you know, we talk a lot about chat GPT. 452 00:26:42,600 --> 00:26:45,280 Speaker 3: Everyone likes to go into their browser and innovate and 453 00:26:45,480 --> 00:26:48,760 Speaker 3: leverage the power of that technology by using that very 454 00:26:48,840 --> 00:26:51,160 Speaker 3: large model that exists in the cloud. But we also 455 00:26:51,200 --> 00:26:54,800 Speaker 3: have local GPT applications where you can use the power 456 00:26:54,840 --> 00:26:57,920 Speaker 3: of your local compute, even your PC. And of course 457 00:26:57,920 --> 00:27:00,879 Speaker 3: AI's driving a number of innovations around on the AIPC 458 00:27:01,119 --> 00:27:03,880 Speaker 3: and ensuring that you can unleash the power of AI 459 00:27:04,160 --> 00:27:07,879 Speaker 3: locally within your office, at your location, and as I mentioned, 460 00:27:07,880 --> 00:27:11,200 Speaker 3: with edge computing, even devices that have far less horsepower 461 00:27:11,240 --> 00:27:12,119 Speaker 3: than your local PC. 462 00:27:13,000 --> 00:27:16,600 Speaker 1: I'm looking forward to it. Also, there's another partnership that 463 00:27:16,640 --> 00:27:21,000 Speaker 1: you have with MILA, which is the Quebec Artificial Intelligence Institute. Yes, 464 00:27:21,440 --> 00:27:23,680 Speaker 1: and it has the largest concentration of world class deep 465 00:27:23,720 --> 00:27:27,720 Speaker 1: learning academic researchers and they have a mission of inspiring 466 00:27:27,760 --> 00:27:31,440 Speaker 1: innovation and progress and AI for the benefit of humanity. 467 00:27:32,240 --> 00:27:35,120 Speaker 1: Can you talk a little bit about the work Intel's 468 00:27:35,160 --> 00:27:36,480 Speaker 1: done with them. Yeah. 469 00:27:36,520 --> 00:27:37,119 Speaker 2: Absolutely. 470 00:27:37,320 --> 00:27:40,320 Speaker 3: This is really under that mission that we have of 471 00:27:40,400 --> 00:27:45,200 Speaker 3: advancing AI everywhere, and we're really grateful for the collaborations 472 00:27:45,280 --> 00:27:48,639 Speaker 3: that we have to advance this across many different areas 473 00:27:48,680 --> 00:27:51,679 Speaker 3: and it really requires partners across the value chain. So 474 00:27:52,040 --> 00:27:54,800 Speaker 3: we talked about hugging face a couple of times. Hugging Face, 475 00:27:54,960 --> 00:27:58,720 Speaker 3: for those who don't know, develop tools for building applications 476 00:27:58,800 --> 00:28:02,000 Speaker 3: using machine learning, and it really exists as the single 477 00:28:02,359 --> 00:28:06,520 Speaker 3: largest model hub for transformers and large language models and 478 00:28:06,560 --> 00:28:10,320 Speaker 3: all types of models. And together we're partnering with Hugging 479 00:28:10,320 --> 00:28:13,760 Speaker 3: Face here at Intel to build optimization tools to accelerate 480 00:28:14,160 --> 00:28:18,120 Speaker 3: training and inference with transformers. And so what's exciting about 481 00:28:18,160 --> 00:28:20,480 Speaker 3: these partnerships is when we really see them in action, 482 00:28:20,720 --> 00:28:25,800 Speaker 3: and we see customers and partners leveraging these Intel optimizations 483 00:28:25,800 --> 00:28:28,960 Speaker 3: and tools to really advance AI and deliver on this 484 00:28:29,080 --> 00:28:33,040 Speaker 3: goal of AI everywhere. And our strategic research and collaboration 485 00:28:33,280 --> 00:28:36,480 Speaker 3: with MELA is really going to speed the research and 486 00:28:36,680 --> 00:28:39,880 Speaker 3: development of advanced AI to solve some of the world's 487 00:28:39,880 --> 00:28:43,520 Speaker 3: most critical and challenging issues. As part of the specific commitment, 488 00:28:43,560 --> 00:28:47,960 Speaker 3: more than twenty researchers between Intel and MELA are planning 489 00:28:48,000 --> 00:28:52,400 Speaker 3: to focus on developing advanced AI techniques to tackle areas 490 00:28:52,440 --> 00:28:57,680 Speaker 3: like climate change and material discovery, and also molecular science, 491 00:28:57,720 --> 00:29:02,680 Speaker 3: so molecular drivers of disease and drug recovery, and if 492 00:29:02,720 --> 00:29:06,040 Speaker 3: we look specifically in that area, biology in particular is 493 00:29:06,200 --> 00:29:10,320 Speaker 3: just a tremendously exciting frontier in natural sciences, and so 494 00:29:10,680 --> 00:29:13,400 Speaker 3: you can imagine the results that could be created. You've 495 00:29:13,400 --> 00:29:15,880 Speaker 3: got the opportunity to truly usher in the era of 496 00:29:15,960 --> 00:29:19,600 Speaker 3: precision medicine, where we have an opportunity to learn from 497 00:29:19,680 --> 00:29:22,360 Speaker 3: all of the massive data that exists to bring benefit 498 00:29:22,480 --> 00:29:26,280 Speaker 3: to certainly to humanity but specifically to individuals. And so 499 00:29:26,640 --> 00:29:29,440 Speaker 3: it's just one area of focus, but the potential, as 500 00:29:29,480 --> 00:29:31,920 Speaker 3: you might imagine, is huge, and this is where Intel 501 00:29:31,960 --> 00:29:35,880 Speaker 3: and Mila are really leaning into evolving that relationship and 502 00:29:35,960 --> 00:29:39,960 Speaker 3: really focusing on this opportunity for technology to really benefit humanity. 503 00:29:40,720 --> 00:29:43,720 Speaker 1: We've seen that AI has been used already for customer 504 00:29:43,720 --> 00:29:47,400 Speaker 1: service and communication, and we've also talked about the ethical 505 00:29:47,400 --> 00:29:50,640 Speaker 1: and responsible use of AI. What are some of the 506 00:29:50,720 --> 00:29:53,680 Speaker 1: challenges that business face when trying to implement an AI 507 00:29:53,800 --> 00:29:57,160 Speaker 1: solution in respect to trying to keep it responsible. 508 00:29:57,760 --> 00:30:00,200 Speaker 3: Yeah, we've already talked about the fact that as we 509 00:30:00,200 --> 00:30:03,239 Speaker 3: look at the promise of AI, it definitely is not 510 00:30:03,480 --> 00:30:06,600 Speaker 3: one size fits all, and there's a tremendous amount of 511 00:30:06,600 --> 00:30:09,880 Speaker 3: complexity and it doesn't just come from how do you 512 00:30:09,920 --> 00:30:12,600 Speaker 3: manage the data? How do you drive data science innovation? 513 00:30:13,040 --> 00:30:16,040 Speaker 3: There's also technology complexity that we have to work through 514 00:30:16,040 --> 00:30:18,400 Speaker 3: and help our customers work through in order to really 515 00:30:18,520 --> 00:30:22,320 Speaker 3: derive that business benefit. And so for us, it's really 516 00:30:22,320 --> 00:30:25,440 Speaker 3: about the innovation that we drive, but also the ecosystem, 517 00:30:25,600 --> 00:30:28,640 Speaker 3: so the data and AI software ecosystems and the hardware 518 00:30:28,640 --> 00:30:33,160 Speaker 3: ecosystems that are available across that entire spectrum of AI 519 00:30:33,280 --> 00:30:36,480 Speaker 3: solutions that I was talking about, from cloud and large 520 00:30:36,480 --> 00:30:39,920 Speaker 3: scale centralized deployments to edge to client. There's so many 521 00:30:39,960 --> 00:30:42,480 Speaker 3: different ways that this needs to be implemented, and the 522 00:30:42,480 --> 00:30:46,560 Speaker 3: complexity of that technology needs to be managed. As AI 523 00:30:46,960 --> 00:30:51,520 Speaker 3: gets widely adopted and deployed and consumption increases, we also 524 00:30:51,600 --> 00:30:54,080 Speaker 3: need to ensure that it's accessible and cost effective. I 525 00:30:54,120 --> 00:30:58,360 Speaker 3: talked earlier about the massive cost associated with training models 526 00:30:58,440 --> 00:31:01,520 Speaker 3: and with driving inference on large scale models, and so 527 00:31:01,600 --> 00:31:03,520 Speaker 3: this definitely is a top goal for us as we 528 00:31:03,560 --> 00:31:06,600 Speaker 3: look at helping our customers derive as much business value 529 00:31:06,640 --> 00:31:09,040 Speaker 3: as they can. So you've got to look at the 530 00:31:09,160 --> 00:31:11,880 Speaker 3: entire spectrum. How do you bring data into the model, 531 00:31:12,160 --> 00:31:15,160 Speaker 3: how do you keep track of your models and model versions, 532 00:31:15,200 --> 00:31:17,320 Speaker 3: how do you drive updates, how do you do training, 533 00:31:17,760 --> 00:31:19,960 Speaker 3: and of course fine tuning like we talked about before, 534 00:31:20,000 --> 00:31:22,600 Speaker 3: but also retraining, and then how do you look at 535 00:31:22,600 --> 00:31:25,760 Speaker 3: the different deployment options and then ultimately integrate them into 536 00:31:25,880 --> 00:31:29,240 Speaker 3: business applications. And really, you know a lot of existing 537 00:31:29,240 --> 00:31:32,520 Speaker 3: business applications, there's a legacy code base there, there is 538 00:31:32,640 --> 00:31:36,120 Speaker 3: legacy infrastructure that isn't going to go away, and so 539 00:31:36,240 --> 00:31:38,520 Speaker 3: you have to manage all of that together. And so 540 00:31:38,960 --> 00:31:41,960 Speaker 3: our goal is really to offer a complete portfolio of 541 00:31:42,160 --> 00:31:46,560 Speaker 3: AI hardware and software and tools optimize for any needs 542 00:31:47,000 --> 00:31:49,120 Speaker 3: and then help them choose the best solution that meets 543 00:31:49,200 --> 00:31:49,760 Speaker 3: their needs. 544 00:31:50,360 --> 00:31:53,720 Speaker 1: Yeah, and kind of leads me to talking about business 545 00:31:53,760 --> 00:31:56,360 Speaker 1: in general. My family has a history in small businesses 546 00:31:57,040 --> 00:32:00,120 Speaker 1: and I'm particularly interested to know your thoughts on how 547 00:32:00,000 --> 00:32:03,080 Speaker 1: so maybe the smaller end of town can leverage some 548 00:32:03,160 --> 00:32:04,920 Speaker 1: of this really cool stuff that's going on. 549 00:32:05,440 --> 00:32:07,720 Speaker 3: Yeah, what's exciting about it is that even a small 550 00:32:07,760 --> 00:32:11,040 Speaker 3: business can take advantage of AI technology deployed locally, but 551 00:32:11,160 --> 00:32:14,120 Speaker 3: also take advantage of AI technology that exists in large 552 00:32:14,120 --> 00:32:17,520 Speaker 3: scale deployments as well that are interconnected through cloud based deployment. 553 00:32:17,640 --> 00:32:18,920 Speaker 2: So how do we do this? 554 00:32:19,520 --> 00:32:22,240 Speaker 3: It really is about taking these hardware and these software 555 00:32:22,280 --> 00:32:26,280 Speaker 3: tools provided by Intel, provided by our ecosystem to enable 556 00:32:26,440 --> 00:32:31,440 Speaker 3: organizations to accelerate performance and really expedite the results for 557 00:32:31,800 --> 00:32:34,040 Speaker 3: the goals and the KPIs that they're driving and really 558 00:32:34,120 --> 00:32:37,920 Speaker 3: ultimately improve their return on investment, and then just turn 559 00:32:37,960 --> 00:32:41,000 Speaker 3: it into a learning cycle, drive that AI development and 560 00:32:41,080 --> 00:32:44,360 Speaker 3: workflow process, but really learn from that and ultimately drive 561 00:32:44,400 --> 00:32:48,080 Speaker 3: improvements and streamline development for expanded use cases from there. 562 00:32:48,120 --> 00:32:50,280 Speaker 3: So it's kind of starting in one domain but then 563 00:32:50,320 --> 00:32:54,320 Speaker 3: having potential to expand to other areas as well. If 564 00:32:54,320 --> 00:32:56,400 Speaker 3: you look at an area like healthcare, this can mean 565 00:32:56,440 --> 00:33:01,520 Speaker 3: accelerating research and patient outcomes with more accurate analysis of 566 00:33:01,640 --> 00:33:04,920 Speaker 3: areas like medical imaging. If you look at manufacturing, there's 567 00:33:05,040 --> 00:33:08,200 Speaker 3: large scale manufacturing and very small scale manufacturing, right, but 568 00:33:08,320 --> 00:33:11,760 Speaker 3: all of them have potential to transform data into insights 569 00:33:12,120 --> 00:33:16,440 Speaker 3: that can improve performance and minimize downtime and improve safety. 570 00:33:16,440 --> 00:33:20,080 Speaker 3: Those are of concern of a manufacturing site of any size. 571 00:33:20,240 --> 00:33:23,360 Speaker 3: Retail's a huge area of focus for small business no 572 00:33:23,480 --> 00:33:26,600 Speaker 3: matter the size of the retail institution. They're all very 573 00:33:26,600 --> 00:33:30,280 Speaker 3: interested in wanting to understand their customers better. So you 574 00:33:30,320 --> 00:33:34,040 Speaker 3: look at areas like inventory management and loss management and 575 00:33:34,120 --> 00:33:37,480 Speaker 3: loss control and other key metrics, and so taking the 576 00:33:37,560 --> 00:33:40,520 Speaker 3: data that you have gathering even more data and then 577 00:33:40,640 --> 00:33:43,840 Speaker 3: using the power of AI to drive those business outcomes 578 00:33:44,000 --> 00:33:46,640 Speaker 3: is essential. Really, it just starts with understanding what the 579 00:33:46,640 --> 00:33:49,080 Speaker 3: business needs and challenges are before you ever even talk 580 00:33:49,080 --> 00:33:52,480 Speaker 3: about the technology. What are the business needs and the challenges, 581 00:33:52,840 --> 00:33:55,600 Speaker 3: and then partnering with Intel, partnering with our customers and 582 00:33:55,640 --> 00:34:00,360 Speaker 3: the value chain in identifying the best AI solutions and outcome. Really, 583 00:34:00,400 --> 00:34:02,600 Speaker 3: you know, for us, we're just trying to accelerate the 584 00:34:02,640 --> 00:34:05,680 Speaker 3: time to market for those AI enabled offerings across every 585 00:34:05,680 --> 00:34:09,520 Speaker 3: industry to maximize business value for both the large enterprises 586 00:34:09,560 --> 00:34:11,960 Speaker 3: as well as the small businesses. 587 00:34:12,080 --> 00:34:16,080 Speaker 1: In terms of improving business productivity. Are there any specific 588 00:34:16,160 --> 00:34:21,320 Speaker 1: examples of AI technologies at Intel actually achieve some really 589 00:34:21,360 --> 00:34:22,400 Speaker 1: significant gains. 590 00:34:23,120 --> 00:34:26,480 Speaker 3: Yeah, so for sure, I'd love to share a couple examples. 591 00:34:26,640 --> 00:34:29,160 Speaker 3: So we did talk a little bit before about healthcare 592 00:34:29,160 --> 00:34:31,240 Speaker 3: and sciences, So I want to kind of shift gears 593 00:34:31,239 --> 00:34:32,760 Speaker 3: into a little area. 594 00:34:33,080 --> 00:34:33,680 Speaker 2: Oh, shift gears. 595 00:34:33,680 --> 00:34:35,160 Speaker 3: I was going to talk about manufacturing, So I guess 596 00:34:35,160 --> 00:34:38,880 Speaker 3: it's a little bit of a pun. So a major 597 00:34:39,200 --> 00:34:43,400 Speaker 3: beverage bottler in Asia turned to Intel ANDAO two in 598 00:34:43,640 --> 00:34:47,280 Speaker 3: twenty twenty two to build a framework that could transform 599 00:34:47,640 --> 00:34:52,040 Speaker 3: a variety of manufacturing and safety inspection processors at a 600 00:34:52,160 --> 00:34:55,200 Speaker 3: number of their regional factories, and what they really hoped 601 00:34:55,200 --> 00:34:59,200 Speaker 3: to do was to take the safety monitoring process into 602 00:34:59,360 --> 00:35:03,520 Speaker 3: the digital AI age through machine vision and AI, but 603 00:35:03,719 --> 00:35:07,399 Speaker 3: still integrate that into their core IT systems. I talked 604 00:35:07,400 --> 00:35:09,680 Speaker 3: a little bit before about the fact that you need 605 00:35:09,719 --> 00:35:13,879 Speaker 3: to integrate into legacy IT systems legacy software. So within 606 00:35:14,000 --> 00:35:17,400 Speaker 3: tell NAO two on board, the company embarked on a 607 00:35:17,440 --> 00:35:21,680 Speaker 3: whole safety transformation project and this was focused on revamping 608 00:35:21,680 --> 00:35:26,160 Speaker 3: security processes at these factories really all the way across China. 609 00:35:26,719 --> 00:35:29,480 Speaker 3: So this company, they were really focused on taking humans 610 00:35:29,520 --> 00:35:33,040 Speaker 3: out of the safety monitoring loop wherever possible to improve results, 611 00:35:33,160 --> 00:35:36,560 Speaker 3: and they had some tremendous results. They were able to 612 00:35:36,600 --> 00:35:40,239 Speaker 3: reduce manual workloads by eighty percent, they were able to 613 00:35:40,480 --> 00:35:43,319 Speaker 3: trim the costs that were related to their health and 614 00:35:43,360 --> 00:35:47,840 Speaker 3: safety and environmental compliance by sixty percent. They were able 615 00:35:47,880 --> 00:35:52,040 Speaker 3: to boost the violation detection rate from less than twenty 616 00:35:52,040 --> 00:35:55,160 Speaker 3: percent to ninety percent, which varied from a four to 617 00:35:55,239 --> 00:35:59,120 Speaker 3: five x improvement in time saved as well, and they 618 00:35:59,160 --> 00:36:02,000 Speaker 3: also reduce safety violations by thirty five percent. So this 619 00:36:02,040 --> 00:36:05,080 Speaker 3: is through the use of AI technologies, its computer vision, 620 00:36:05,320 --> 00:36:09,440 Speaker 3: and so it's taking inputs from multiple different cameras on 621 00:36:09,520 --> 00:36:12,520 Speaker 3: the factory floor and being able to detect different patterns 622 00:36:12,560 --> 00:36:16,080 Speaker 3: like worker movement or worker machine interaction, or being able 623 00:36:16,080 --> 00:36:17,439 Speaker 3: to enforce safety. 624 00:36:17,080 --> 00:36:19,120 Speaker 2: Boundaries around equipment. 625 00:36:19,440 --> 00:36:22,799 Speaker 3: You achieve tremendously better results for the business's bottom line, 626 00:36:22,800 --> 00:36:25,120 Speaker 3: but also safety for the people who were involved in 627 00:36:25,120 --> 00:36:29,240 Speaker 3: the factory. And so really almost overnight they witnessed enormous 628 00:36:29,239 --> 00:36:33,600 Speaker 3: benefits that totally altered how they approach monitoring and inspection, 629 00:36:33,920 --> 00:36:36,879 Speaker 3: and they put this advanced solution in place, and it 630 00:36:36,960 --> 00:36:41,520 Speaker 3: really accommodated the very complex and very intense automation requirements 631 00:36:41,520 --> 00:36:44,280 Speaker 3: that they were seeking to build. Let me shift gears 632 00:36:44,320 --> 00:36:47,080 Speaker 3: a little bit again and talk about the entertainment industry. 633 00:36:47,160 --> 00:36:50,680 Speaker 3: So if we look at the broadcasting industry, it's really 634 00:36:50,719 --> 00:36:56,000 Speaker 3: benefited already enormously from digital transformation, but at the same time, 635 00:36:56,239 --> 00:36:59,399 Speaker 3: the hardware and the networking infrastructure sometimes really hasn't kept 636 00:36:59,400 --> 00:37:02,800 Speaker 3: pace with that level of change, and so many businesses 637 00:37:02,800 --> 00:37:06,719 Speaker 3: still rely on cumbersome and sometimes dedicated hardware that can 638 00:37:06,760 --> 00:37:09,800 Speaker 3: be very expensive, and very often they're actually locked into 639 00:37:09,840 --> 00:37:13,600 Speaker 3: specific brands and equipment. And so if you think about 640 00:37:13,640 --> 00:37:18,759 Speaker 3: the massive opportunity associated with very large entertainment events, so 641 00:37:19,120 --> 00:37:22,239 Speaker 3: eyes around the world tuning into Formula on racing or 642 00:37:22,400 --> 00:37:26,200 Speaker 3: World Cup or next year Summer Olympics in Paris. This 643 00:37:26,280 --> 00:37:28,840 Speaker 3: can translate into a lot of great business outcomes in 644 00:37:28,880 --> 00:37:32,439 Speaker 3: advertising and other revenue dollars for broadcasters, but viewers really 645 00:37:32,440 --> 00:37:36,520 Speaker 3: want a seamless experience as well, and so these broadcasters 646 00:37:36,520 --> 00:37:40,240 Speaker 3: depend upon a number of different vendors across so many 647 00:37:40,239 --> 00:37:44,239 Speaker 3: different events and locations, and they have to deliver on 648 00:37:44,360 --> 00:37:47,200 Speaker 3: those performance needs. They have to deliver on performance and 649 00:37:47,520 --> 00:37:51,760 Speaker 3: low latency and real time visibility and monitoring and really 650 00:37:51,840 --> 00:37:54,399 Speaker 3: rich viewing experience. It's not just about viewing the event, 651 00:37:54,480 --> 00:37:57,200 Speaker 3: it's about viewing additional data associated with the event, and 652 00:37:57,239 --> 00:38:00,640 Speaker 3: real time analytics and real time commentary. And they really 653 00:38:00,680 --> 00:38:04,600 Speaker 3: want a solution that is powerful and agile and easy 654 00:38:04,640 --> 00:38:08,759 Speaker 3: to deploy, that doesn't sacrifice performance, doesn't sacrifice quality. And 655 00:38:08,840 --> 00:38:11,440 Speaker 3: also they need to be able to from a technology standpoint, 656 00:38:11,560 --> 00:38:14,960 Speaker 3: leverage on site infrastructure as well as the cloud and 657 00:38:15,040 --> 00:38:18,560 Speaker 3: do that seamlessly between the different locations. So this is 658 00:38:18,560 --> 00:38:21,440 Speaker 3: where we've been able to innovate alongside our partners, and 659 00:38:21,480 --> 00:38:23,080 Speaker 3: in this case the example, I want to talk about 660 00:38:23,120 --> 00:38:23,960 Speaker 3: is Fox Sports. 661 00:38:24,400 --> 00:38:26,440 Speaker 2: Fox Sports used TAG Video. 662 00:38:26,320 --> 00:38:32,319 Speaker 3: Systems monitoring and visualization platform which ran on Xon CPUs. 663 00:38:32,760 --> 00:38:35,320 Speaker 3: They use this in all of their control rooms and 664 00:38:35,360 --> 00:38:39,880 Speaker 3: their operations to show sixty four of the soccer matches 665 00:38:40,040 --> 00:38:42,760 Speaker 3: in the twenty twenty two FIFA World Cup in catter 666 00:38:43,000 --> 00:38:46,520 Speaker 3: across the US and also on Fox and FS one channels, 667 00:38:46,960 --> 00:38:49,920 Speaker 3: and these matches were being live streamed also on the 668 00:38:49,960 --> 00:38:52,120 Speaker 3: Fox Sports app, and they also had to be of 669 00:38:52,160 --> 00:38:56,000 Speaker 3: course available on demand for replays later. So the system 670 00:38:56,040 --> 00:38:58,680 Speaker 3: had to be able to monitor the integrity of over 671 00:38:58,800 --> 00:39:03,000 Speaker 3: twelve hundred sources and drive over one hundred and fifty displays. 672 00:39:03,480 --> 00:39:07,799 Speaker 3: So how did they tackle this massive challenge. They deployed 673 00:39:08,040 --> 00:39:11,040 Speaker 3: this really first of its kind live production system called 674 00:39:11,080 --> 00:39:15,320 Speaker 3: a flypack. This system includes a full control room, forty 675 00:39:15,680 --> 00:39:19,560 Speaker 3: tech core racks, ten venue racks, and these equipment racks. 676 00:39:19,600 --> 00:39:22,279 Speaker 3: They were built around Intel Xeon processors and they can 677 00:39:22,280 --> 00:39:24,440 Speaker 3: actually be flown in a plane. They can be flown 678 00:39:24,480 --> 00:39:27,880 Speaker 3: via seven forty seven from venue to venue instead of 679 00:39:27,920 --> 00:39:31,400 Speaker 3: having to travel aboard container ships. So the flypack arrives, 680 00:39:31,480 --> 00:39:34,359 Speaker 3: it's fully pre wired and it can be powered up 681 00:39:34,400 --> 00:39:37,480 Speaker 3: and ready to go within six hours and they report 682 00:39:37,480 --> 00:39:39,920 Speaker 3: that it's allowed them to shave weeks off of their 683 00:39:39,960 --> 00:39:43,200 Speaker 3: setup times. So it really is about improving interoperability and 684 00:39:43,280 --> 00:39:47,760 Speaker 3: portability and agility, and really it's transforming and revolutionizing how 685 00:39:47,920 --> 00:39:51,960 Speaker 3: live production is done. And so Fox Sports really also 686 00:39:52,000 --> 00:39:54,200 Speaker 3: has been able to change and add to the system 687 00:39:54,320 --> 00:39:57,400 Speaker 3: as their needs evolve. So a lot of really exciting 688 00:39:57,400 --> 00:40:00,640 Speaker 3: innovation in that combination of harnessing the power of compute 689 00:40:00,800 --> 00:40:04,560 Speaker 3: and edge applications and AI and really impacting a number 690 00:40:04,600 --> 00:40:05,440 Speaker 3: of different industries. 691 00:40:06,480 --> 00:40:09,440 Speaker 1: And just to wrap up, I'd really like to get 692 00:40:09,480 --> 00:40:12,960 Speaker 1: your thoughts on the future of the data center and 693 00:40:13,040 --> 00:40:17,480 Speaker 1: AI technologies at Intel and what's your number one area 694 00:40:17,520 --> 00:40:19,400 Speaker 1: of excitement for the future. 695 00:40:20,200 --> 00:40:20,480 Speaker 2: Yeah. 696 00:40:20,840 --> 00:40:24,120 Speaker 3: I'm relatively new to this role, and in the short 697 00:40:24,120 --> 00:40:26,040 Speaker 3: amount of time that I have spent in this role, 698 00:40:26,120 --> 00:40:30,400 Speaker 3: I have seen growth like nothing I've ever experienced. The 699 00:40:30,560 --> 00:40:34,080 Speaker 3: pace of innovation in terms of the solutions that are 700 00:40:34,080 --> 00:40:37,919 Speaker 3: being deployed, but the pace really incumbent upon Intel for 701 00:40:38,000 --> 00:40:42,320 Speaker 3: innovating technology to empower those solutions is massive. It's truly 702 00:40:42,400 --> 00:40:45,560 Speaker 3: undeniable that AI is changing the way that we're living 703 00:40:45,560 --> 00:40:47,440 Speaker 3: our lives, the way that we're working, the way that 704 00:40:47,440 --> 00:40:51,799 Speaker 3: we're driving business transformation, and I'm really an optimist when 705 00:40:51,800 --> 00:40:55,440 Speaker 3: it comes to the power of AI. My goal professionally 706 00:40:55,640 --> 00:40:59,640 Speaker 3: is to innovate and deliver the technology and the products 707 00:41:00,120 --> 00:41:04,600 Speaker 3: that really enable AI to help humanity become the best 708 00:41:04,640 --> 00:41:08,000 Speaker 3: version of itself. There are so many big problems that 709 00:41:08,000 --> 00:41:09,480 Speaker 3: we have to solve. We've talked about a couple of 710 00:41:09,560 --> 00:41:13,160 Speaker 3: them today. When you look at climate to discoveries and 711 00:41:13,239 --> 00:41:18,000 Speaker 3: science to driving social change, AI can be an engine 712 00:41:18,040 --> 00:41:21,040 Speaker 3: to create solutions for all of these and more. 713 00:41:21,080 --> 00:41:21,200 Speaker 1: So. 714 00:41:21,239 --> 00:41:25,839 Speaker 3: It's really exciting to be part of the technology innovator, who, 715 00:41:25,880 --> 00:41:29,080 Speaker 3: of course is one of the most significant in the 716 00:41:29,120 --> 00:41:32,480 Speaker 3: history of computing, but really on this precipice of AI 717 00:41:32,560 --> 00:41:37,400 Speaker 3: innovation and really driving this next phase of at scale 718 00:41:37,920 --> 00:41:41,759 Speaker 3: compute deployments and AI innovation to power those deployments and 719 00:41:41,800 --> 00:41:43,840 Speaker 3: create solutions for the future. 720 00:41:44,280 --> 00:41:47,200 Speaker 1: I definitely share your optimism. Thanks very much, Jennie. That 721 00:41:47,280 --> 00:41:47,720 Speaker 1: was awesome. 722 00:41:47,920 --> 00:41:49,799 Speaker 2: Thanks Graham, it was really great to talk to you. 723 00:41:54,160 --> 00:41:56,520 Speaker 1: Thanks very much to my guest Jenny ber Ivan for 724 00:41:56,680 --> 00:41:59,560 Speaker 1: joining me on the season finale of Technically Speaking and 725 00:41:59,600 --> 00:42:04,759 Speaker 1: Intel podcast. This episode illustrates that the significant advancements and 726 00:42:04,840 --> 00:42:08,320 Speaker 1: contributions made in the field of open source AI worldwide. 727 00:42:08,719 --> 00:42:12,280 Speaker 1: As a developer actively engaged in AI projects, particularly appreciate 728 00:42:12,280 --> 00:42:16,400 Speaker 1: the support provided by corporations like Indel, both financially and 729 00:42:16,440 --> 00:42:21,160 Speaker 1: in terms of fostering open source developer communities. Journey provided 730 00:42:21,200 --> 00:42:25,279 Speaker 1: an extensive discussion on the topic of responsible AI, emphasizing 731 00:42:25,280 --> 00:42:29,280 Speaker 1: the necessity for companies to ensure that AI implementations adhere 732 00:42:29,280 --> 00:42:32,759 Speaker 1: to ethical principles. In my view, it is imperative for 733 00:42:32,840 --> 00:42:36,520 Speaker 1: businesses to explore responsible AI initiatives and establish a set 734 00:42:36,520 --> 00:42:40,239 Speaker 1: of standards that can be clearly understood and implemented by 735 00:42:40,239 --> 00:42:45,279 Speaker 1: their executives, managers, employees, and contractors. It is essential for 736 00:42:45,360 --> 00:42:49,040 Speaker 1: employers to encourage their workforce to voice concerns if they 737 00:42:49,239 --> 00:42:52,640 Speaker 1: perceive any AI projects to be in conflict with their personal, 738 00:42:52,680 --> 00:42:57,280 Speaker 1: moral and ethical standards. Engaging in open and candid discussions 739 00:42:57,280 --> 00:43:00,840 Speaker 1: with our peers is crucial in developing technology that benefit 740 00:43:00,920 --> 00:43:04,120 Speaker 1: humanity as a whole. The development of large scale AI 741 00:43:04,200 --> 00:43:08,400 Speaker 1: models and supercomputers by companies like Intel may appear daunting 742 00:43:08,640 --> 00:43:13,040 Speaker 1: and seemingly unattainable for smaller enterprises. However, it's important to 743 00:43:13,080 --> 00:43:17,120 Speaker 1: remember that technological advancements often start with significant investments by 744 00:43:17,160 --> 00:43:20,759 Speaker 1: pioneers in the field. The first computers, costing millions of 745 00:43:20,800 --> 00:43:23,480 Speaker 1: dollars and possessing only a fraction of the power of 746 00:43:23,480 --> 00:43:27,040 Speaker 1: a modern calculator, were necessary stepping stones that led to 747 00:43:27,040 --> 00:43:30,760 Speaker 1: the explosion of personal computing and the advent of mobile 748 00:43:30,800 --> 00:43:34,279 Speaker 1: devices in the same vein the cost of deploying and 749 00:43:34,360 --> 00:43:38,440 Speaker 1: operating AI technology is expected to decrease over time, enabling 750 00:43:38,480 --> 00:43:41,880 Speaker 1: businesses of all sizes to utilize AI in ways that 751 00:43:41,920 --> 00:43:46,359 Speaker 1: positively impact the employees and customers. I'm excited to see 752 00:43:46,360 --> 00:43:48,560 Speaker 1: all the cool things that people come up with to 753 00:43:48,600 --> 00:43:52,160 Speaker 1: improve our lives. Thanks for following along for the first 754 00:43:52,200 --> 00:43:55,799 Speaker 1: season of Technically Speaking, an Intel podcast. To learn more 755 00:43:55,800 --> 00:43:59,160 Speaker 1: about how Intel is revolutionizing the future of AI, check 756 00:43:59,160 --> 00:44:03,000 Speaker 1: out Intel doc Slash Stories. If you enjoyed this season, 757 00:44:03,239 --> 00:44:07,200 Speaker 1: please stay subscribed to get updates about season two, which 758 00:44:07,200 --> 00:44:09,680 Speaker 1: will be coming out in the spring of twenty twenty four. 759 00:44:10,239 --> 00:44:17,840 Speaker 1: Thanks again for listening. Technically Speaking was produced by Ruby 760 00:44:17,880 --> 00:44:21,359 Speaker 1: Studios from iHeartRadio in partnership with Intel and hosted by 761 00:44:21,400 --> 00:44:25,719 Speaker 1: me Graham Class. Our executive producer is Molly Sosher, our 762 00:44:25,760 --> 00:44:28,920 Speaker 1: EP of Post production is James Foster, and our supervising 763 00:44:28,960 --> 00:44:33,240 Speaker 1: producer is nikiir Swinton. This episode was edited by Sierra 764 00:44:33,320 --> 00:44:39,040 Speaker 1: Spreen and written by Tyree Rush.