1 00:00:03,389 --> 00:00:06,269 Narrator: You're listening to the humans of DevOps podcast, a 2 00:00:06,269 --> 00:00:09,929 podcast focused on advancing the humans of DevOps through skills, 3 00:00:09,959 --> 00:00:14,309 knowledge, ideas and learning, or the SK il framework. 4 00:00:17,220 --> 00:00:19,710 Richard Whitehead: Right, so the good news is people's jobs like 5 00:00:19,710 --> 00:00:24,450 most definitely not at risk. The problem we're trying to solve or 6 00:00:24,480 --> 00:00:29,520 solve, I think is it's increasing at a greater rate 7 00:00:29,550 --> 00:00:36,300 than I think we can. We can sort of solve the problem. So the 8 00:00:36,300 --> 00:00:39,330 expanding nature of the problem, I think, secures people's 9 00:00:39,330 --> 00:00:43,920 employment for a very, very long time. I think in every aspect 10 00:00:44,460 --> 00:00:48,000 of, of any form of digital transformation, when you look at 11 00:00:48,000 --> 00:00:52,620 any aspect of the business, it doesn't matter how much effort 12 00:00:52,620 --> 00:00:56,010 how much code rewrites how much automation we do. The 13 00:00:56,010 --> 00:00:59,850 opportunity to refer to this as an opportunity, not a problem. 14 00:01:00,180 --> 00:01:04,320 The opportunity is increasing. So, so fast, that I don't think 15 00:01:04,320 --> 00:01:06,240 anybody's going to be able to jump anytime soon. 16 00:01:08,580 --> 00:01:11,070 Jason Baum: Hey, everyone, welcome back. It's Jason Baum, 17 00:01:11,070 --> 00:01:14,490 Director of Member experience at DevOps Institute. And this is 18 00:01:14,490 --> 00:01:19,200 the humans of DevOps podcast. I hope you had a great week. We're 19 00:01:19,200 --> 00:01:23,970 glad you came back to join us. So let me take you back to the 20 00:01:23,970 --> 00:01:29,250 80s The machines are coming to get us scenes in lines from the 21 00:01:29,250 --> 00:01:32,880 Terminator and it's many sequels are forever etched in our 22 00:01:32,880 --> 00:01:37,530 brains. It hauntingly depicts a world hell bent on technological 23 00:01:37,530 --> 00:01:40,380 growth that led to the rise in advanced machine learning 24 00:01:40,380 --> 00:01:43,290 techniques and artificial intelligence that would 25 00:01:43,410 --> 00:01:47,700 ultimately lead to the world's demise. Well, we're on our way 26 00:01:47,700 --> 00:01:52,890 there, folks. But instead of doomsday scenarios, we're all 27 00:01:52,890 --> 00:01:57,570 picturing. Ai ops has emerged as an essential step forward for 28 00:01:57,570 --> 00:02:01,020 enterprises in a variety of different industries. Gartner 29 00:02:01,020 --> 00:02:05,010 defines AI ops as artificial intelligence for IT operations, 30 00:02:05,010 --> 00:02:08,760 which combines big data and machine learning to automate it 31 00:02:08,760 --> 00:02:13,500 operations processes. At its core, it's all about it, teams 32 00:02:13,500 --> 00:02:17,010 and organizations can use AI to manage data in their 33 00:02:17,010 --> 00:02:20,520 environments. Through this approach, teams can employ large 34 00:02:20,520 --> 00:02:24,870 scale datasets, machine learning and automation to MAKE IT ops 35 00:02:24,870 --> 00:02:29,280 faster, simpler and more efficient. Many believe AI will 36 00:02:29,280 --> 00:02:33,600 not only impact organizations, but will become a major facet in 37 00:02:33,600 --> 00:02:37,470 our everyday lives through the emergence of new applications. I 38 00:02:37,470 --> 00:02:41,550 mean, we already see ML and AI every day with things like face 39 00:02:41,550 --> 00:02:45,330 ID smart replies, product recommendations, chat bots, you 40 00:02:45,330 --> 00:02:48,930 name it. Today, we're going to talk to Richard Whitehead 41 00:02:48,960 --> 00:02:55,050 evangelist and chief CTO of Moog soft about what aiops really 42 00:02:55,050 --> 00:02:58,500 means for humans, as opposed to being stuck in the 80s version 43 00:02:58,500 --> 00:03:03,210 of this story. Sorry about that. Richard, are he ready to get 44 00:03:03,210 --> 00:03:03,720 human? 45 00:03:04,320 --> 00:03:05,220 Richard Whitehead: How suddenly? 46 00:03:06,330 --> 00:03:08,760 Jason Baum: First of all, thanks so much for coming on. 47 00:03:09,600 --> 00:03:14,550 appreciate having you here on the podcast. And my apologies 48 00:03:14,550 --> 00:03:18,480 for taking it back to the 80s. But every time I talk about AI, 49 00:03:18,870 --> 00:03:24,270 my mind immediately goes to the machines are coming for us. 50 00:03:25,170 --> 00:03:28,050 Richard Whitehead: And certainly, when I started 51 00:03:28,050 --> 00:03:32,970 working in this space, I think every single PowerPoint 52 00:03:32,970 --> 00:03:36,750 presentation I saw both internally and externally had at 53 00:03:36,780 --> 00:03:42,690 least some reference to Skynet and a terminator. So yes. It's 54 00:03:42,690 --> 00:03:44,490 not that not that far back. 55 00:03:45,150 --> 00:03:48,660 Jason Baum: Yeah, no, it's not that far back. And I think 56 00:03:48,660 --> 00:03:54,180 people are still still scared by it. I have a funny story that 57 00:03:54,180 --> 00:03:58,530 I've told. Perhaps in the past, I'm trying to remember maybe 58 00:03:58,530 --> 00:04:07,170 maybe not, of my mother actually keeps her Amazon name cannot be 59 00:04:07,170 --> 00:04:10,290 said because it's actually in the room with me and we'll start 60 00:04:10,290 --> 00:04:16,380 speaking. But we know what we're talking about. She puts it away. 61 00:04:16,410 --> 00:04:18,750 She actually keeps it in a cabinet and when she wants to 62 00:04:18,750 --> 00:04:22,740 use it, takes it out and plugs it in and then says the magic 63 00:04:22,740 --> 00:04:25,680 words to turn it on because she's afraid of it listening to 64 00:04:25,680 --> 00:04:26,070 her. 65 00:04:27,480 --> 00:04:29,610 Richard Whitehead: That might be a legitimate fear. It's always 66 00:04:29,610 --> 00:04:36,000 listening that it has to but yeah, I think I you know, I'm a 67 00:04:36,000 --> 00:04:40,230 pragmatist. I don't fear artificial intelligence, but I 68 00:04:40,230 --> 00:04:44,190 do occasionally have a sense of disappointment. You know, when 69 00:04:44,190 --> 00:04:48,120 my door camera notifies me that there's a person outside and 70 00:04:48,120 --> 00:04:53,520 it's it's clearly a dog. And, and also some, some shopping 71 00:04:53,520 --> 00:04:57,690 recommendations. I get, you know, I I, if I were to buy an 72 00:04:57,690 --> 00:05:02,580 item such as a dishwasher, I don't feel the need to be to 73 00:05:02,580 --> 00:05:05,040 have somebody suggest that I buy a dishwasher for the next six 74 00:05:05,040 --> 00:05:11,070 months. So I think people, you know, it's less fear, it's more 75 00:05:11,070 --> 00:05:15,180 disappointment in what, what AI can bring to you based on some 76 00:05:15,180 --> 00:05:19,560 of the some of the more obvious and commercial barons that are 77 00:05:19,560 --> 00:05:19,920 out there. 78 00:05:21,270 --> 00:05:25,290 Jason Baum: Yeah, so let's talk about what AI is. So I feel like 79 00:05:25,290 --> 00:05:30,510 that's a good place to start. You heard the Gartner 80 00:05:30,540 --> 00:05:33,930 definition, want to include that because I always feel like 81 00:05:33,930 --> 00:05:38,130 having an academic of sorts definition is important to hear. 82 00:05:39,210 --> 00:05:43,620 I've also heard someone very succinctly put it that AI, 83 00:05:43,800 --> 00:05:47,190 artificial intelligence is simply the problem. And then 84 00:05:47,190 --> 00:05:51,780 once it's solved, it's no longer AI, which I think is a 85 00:05:51,780 --> 00:05:55,710 fascinating way to look at it. I'm curious how you would define 86 00:05:55,710 --> 00:05:55,950 it. 87 00:05:57,270 --> 00:06:01,140 Richard Whitehead: So much. So I definitions a little broader. To 88 00:06:01,140 --> 00:06:04,800 go back to the specific definition, the garden 89 00:06:04,800 --> 00:06:07,680 definition, which is actually evolved into that definition. 90 00:06:08,790 --> 00:06:11,040 Because when you put the letters AI together, people 91 00:06:11,070 --> 00:06:14,730 automatically assume it means artificial intelligence. So AI 92 00:06:14,730 --> 00:06:17,310 Ops is the application of artificial intelligence 93 00:06:17,310 --> 00:06:23,670 techniques to IT operations. So that's really it. And, of 94 00:06:23,670 --> 00:06:26,700 course, means that it's a very broad definition, which means 95 00:06:26,700 --> 00:06:30,030 there are a lot of technologies and techniques and solutions out 96 00:06:30,030 --> 00:06:33,960 there that all fit into this umbrella definition. So when 97 00:06:33,960 --> 00:06:37,320 people talk about artificial intelligence, there's a general 98 00:06:37,320 --> 00:06:42,000 sense that what you're talking about is technology or 99 00:06:42,000 --> 00:06:46,020 computers, that are in some way attempting to replicate humans, 100 00:06:46,650 --> 00:06:49,680 you know, and that's, that's where the inevitable screenshot 101 00:06:49,680 --> 00:06:53,850 of the Terminator, robot appears, and so forth. And I 102 00:06:53,850 --> 00:06:56,190 think that's, that's generally true. So, you know, when you 103 00:06:56,190 --> 00:06:59,490 think of chess playing, computers and things like that, 104 00:07:00,180 --> 00:07:04,530 and that sense has been largely reinforced by some of the early 105 00:07:04,560 --> 00:07:10,530 adoptions of AI in things like service desks. So you know, when 106 00:07:10,530 --> 00:07:13,800 you call into a service desk, your first interaction is likely 107 00:07:13,800 --> 00:07:17,760 to be with some form of AI capability where it attempts to, 108 00:07:18,210 --> 00:07:21,450 to give you an answer very rapidly without any form of real 109 00:07:21,450 --> 00:07:25,020 human interaction. And, you know, they, in most cases, 110 00:07:25,020 --> 00:07:29,790 they're attempting to sort of pass the Turing test. In other 111 00:07:29,790 --> 00:07:32,040 words, you're talking to a computer and is trying to make 112 00:07:32,040 --> 00:07:34,740 it as human like that interaction to be as human like 113 00:07:34,740 --> 00:07:38,790 as possible. So while it's true for things like service desks, 114 00:07:38,820 --> 00:07:41,850 when you when you dig deeper into some technology, and start 115 00:07:41,850 --> 00:07:44,700 talking about, you know, the concept of sort of monitoring 116 00:07:44,700 --> 00:07:48,900 and observability remediation, and things like that, it becomes 117 00:07:48,930 --> 00:07:53,040 less attempting to replicate a human but more attempting to 118 00:07:53,040 --> 00:07:58,590 replicate what a human would try and do if they were involved. So 119 00:07:58,590 --> 00:08:02,220 it's not a human interaction, it's an application of human 120 00:08:02,220 --> 00:08:06,420 based intelligence, but in an automated fashion. So with that 121 00:08:06,420 --> 00:08:09,450 broad definition, you're incorporating not just sort of 122 00:08:09,450 --> 00:08:13,260 that chess playing type thing, that's actually the least of the 123 00:08:13,260 --> 00:08:16,020 component. But you are talking about things like machine 124 00:08:16,020 --> 00:08:21,360 learning, you're talking about sort of some sophisticated 125 00:08:21,360 --> 00:08:24,900 algorithms that maybe do linear regression, and you're talking 126 00:08:24,900 --> 00:08:28,470 about some techniques that are in the periphery of artificial 127 00:08:28,470 --> 00:08:31,770 intelligence, such as natural language processing, you know, 128 00:08:31,770 --> 00:08:35,280 I've mentioned already mentioned two that I'm personally 129 00:08:35,340 --> 00:08:38,550 relatively familiar with, which is machine learning, and natural 130 00:08:38,550 --> 00:08:42,570 language processing. And these are things that you don't 131 00:08:42,570 --> 00:08:46,920 necessarily think of when you're talking about AI, but that 132 00:08:46,920 --> 00:08:51,210 they're absolutely relevant and very pertinent to solving 133 00:08:51,210 --> 00:08:52,260 specific problems. 134 00:08:53,010 --> 00:08:56,910 Jason Baum: And with that, that's a good lead into this 135 00:08:56,940 --> 00:09:04,170 next question is, what are some examples of, of problems that 136 00:09:04,170 --> 00:09:07,680 we're looking to solve with AI ops and machine learning 137 00:09:07,680 --> 00:09:10,890 specifically? And And also, when you hear about, okay, we're 138 00:09:10,890 --> 00:09:15,060 looking to solve problems, and automate and speed up some 139 00:09:15,060 --> 00:09:19,650 processes. I think a lot of the some of the misconception then 140 00:09:19,650 --> 00:09:23,760 is, well, now I'm going to lose my job. AI is going to replace 141 00:09:23,760 --> 00:09:28,650 me. So perhaps you could address both in this in in your answer, 142 00:09:28,920 --> 00:09:32,250 what's the problem? And then as we solve it with AI ops, how are 143 00:09:32,250 --> 00:09:34,920 we like, are people's jobs at risk? 144 00:09:36,000 --> 00:09:38,460 Richard Whitehead: Right, so the good news is people's jobs are 145 00:09:38,460 --> 00:09:43,230 most definitely not at risk. The problem we're trying to solve 146 00:09:43,260 --> 00:09:48,300 solve, I think, is it's increasing at a greater rate 147 00:09:48,360 --> 00:09:55,080 than I think we can. We can sort of solve the problem. So the 148 00:09:55,080 --> 00:09:58,110 expanding nature of the problem, I think, secures people's 149 00:09:58,110 --> 00:10:02,820 employment for a very, very long time. I think in every aspect 150 00:10:03,240 --> 00:10:06,780 of, of any form of digital transformation, when you look at 151 00:10:06,780 --> 00:10:11,430 any aspect of the business, it doesn't matter how much effort 152 00:10:11,430 --> 00:10:14,790 how much code rewrite how much automation, we do the 153 00:10:14,790 --> 00:10:18,240 opportunity, I want to refer to it as an opportunity, not a 154 00:10:18,240 --> 00:10:22,890 problem, the opportunity is increasing so fast, that I don't 155 00:10:22,890 --> 00:10:25,470 think anybody's going to be out of a job anytime soon. In fact, 156 00:10:25,470 --> 00:10:30,630 I think if you look at some of the roles, the newer roles that 157 00:10:30,630 --> 00:10:33,750 are emerging as a result of digital transformation, such as 158 00:10:34,110 --> 00:10:39,660 site reliability engineers, sort of developers in general and 159 00:10:40,080 --> 00:10:43,860 sort of operations, folks who are emerging in sort of a DevOps 160 00:10:43,860 --> 00:10:47,880 type, type capacity, you know, that's an expanding market 161 00:10:47,880 --> 00:10:52,830 opportunity, not a shrinking one. So individual teams might 162 00:10:52,830 --> 00:10:57,900 be smaller as a result of, of this technology. But the market 163 00:10:57,900 --> 00:11:01,950 opportunity in general is such that I think it's going to give 164 00:11:01,950 --> 00:11:06,270 me a long time before the demand for people in these roles cools 165 00:11:06,270 --> 00:11:10,200 off. So yeah, not not a problem there. We're dealing with a an 166 00:11:10,200 --> 00:11:16,500 exploding market opportunity. So basically, it sort of comes down 167 00:11:16,500 --> 00:11:19,590 to, I think I mentioned it earlier, the notion of 168 00:11:19,590 --> 00:11:23,940 automation. So when we talk about AI replicating human 169 00:11:23,970 --> 00:11:28,830 activity, I tend to think of an in this this sense, when you're 170 00:11:28,830 --> 00:11:31,620 looking at something that a human would do on their date, 171 00:11:31,650 --> 00:11:35,250 day to day, sort of line of work, when you're solving a 172 00:11:35,250 --> 00:11:39,840 problem addressing an incident, debugging something, the 173 00:11:39,840 --> 00:11:43,290 question we always ask ourselves is, what's the most common task, 174 00:11:43,290 --> 00:11:46,140 what's the most common and repetitive tasks that a human 175 00:11:46,260 --> 00:11:50,640 performance, and those are the things that I think are easy 176 00:11:50,640 --> 00:11:54,420 targets for AI, to replicate, because they tend to be the 177 00:11:54,420 --> 00:11:59,310 mundane tasks, you know, we tend tend to refer to them a lot of 178 00:11:59,310 --> 00:12:03,030 toil. The stuff that you do every single time, that doesn't 179 00:12:03,030 --> 00:12:08,640 necessarily add value, but it's just a task that has to be 180 00:12:08,640 --> 00:12:12,420 performed in order to move on to the the next job of of actually 181 00:12:12,420 --> 00:12:16,710 resolving the issue or, or finding the error. And that's 182 00:12:16,710 --> 00:12:20,700 something that I think, is sort of overlooked, people tend to 183 00:12:20,700 --> 00:12:23,760 think of AI as being an end goal, we're going to completely 184 00:12:23,760 --> 00:12:27,330 replace a human, and you throw data at it and you get a 185 00:12:27,330 --> 00:12:31,530 solution at the other end, I tend to think of AI certainly, 186 00:12:31,830 --> 00:12:36,810 we sometimes refer to it at moogsoft as applied AI, it's 187 00:12:36,810 --> 00:12:40,170 basically a very small tool, you can take a very small tool to 188 00:12:40,170 --> 00:12:43,470 achieve a very specific task. And it could be something as 189 00:12:43,470 --> 00:12:51,090 simple as doing a bit of triage, augmenting some information. So 190 00:12:51,780 --> 00:12:54,780 that's one less task you have to do. That's one less system, you 191 00:12:54,780 --> 00:12:57,570 have to log into to get some additional information. If that 192 00:12:57,570 --> 00:13:00,840 information can be gathered for you, and presented to you. 193 00:13:01,950 --> 00:13:05,550 That's one less mundane task you have to perform in order to get 194 00:13:05,550 --> 00:13:09,120 to the really important stuff, which is using your your human 195 00:13:09,120 --> 00:13:15,240 brain to resolve the issue. And so yeah, so applied AI is a good 196 00:13:15,240 --> 00:13:16,110 way of looking at it. 197 00:13:16,499 --> 00:13:18,059 Jason Baum: That's what it's all about. Right? Getting rid of 198 00:13:18,059 --> 00:13:21,689 that mundane. I think that's the goal. Right? Right. 199 00:13:21,750 --> 00:13:25,170 Richard Whitehead: So certainly a DevOps ideal, right? Yeah. The 200 00:13:25,170 --> 00:13:25,860 daily work. 201 00:13:25,979 --> 00:13:32,009 Jason Baum: Yeah. Efficiency. So um, so it's is it is it once 202 00:13:32,009 --> 00:13:36,809 we're, it's set up, right, we've got it, the mundane is gone. 203 00:13:37,079 --> 00:13:40,529 It's working, you know, everything is being automated, 204 00:13:40,709 --> 00:13:45,659 is simply plug and play. And now we just let it go. Or, you know, 205 00:13:45,689 --> 00:13:48,599 can we trust the machines to continue it? And just in 206 00:13:48,599 --> 00:13:52,049 perpetuity, I guess forever? We're learning the machine is 207 00:13:52,049 --> 00:13:54,689 learning and, and everything is all set? 208 00:13:55,830 --> 00:13:58,770 Richard Whitehead: Well, there's a couple of angles to that. The 209 00:13:58,770 --> 00:14:03,000 first one is, can you just unleash the power of AI and have 210 00:14:03,000 --> 00:14:06,840 it do its job? And that the second aspect of that is, you 211 00:14:06,840 --> 00:14:10,590 know, is it a one one time deal? Do you just use it up once and 212 00:14:10,710 --> 00:14:21,060 let it run? So to address the can you unleash it, maybe maybe 213 00:14:22,710 --> 00:14:26,520 one of the talks or one of the challenges that I I've had 214 00:14:26,550 --> 00:14:30,360 dealing with sort of sort of very conservative minded IT 215 00:14:30,360 --> 00:14:35,250 operations folks, when trying to bring in something as fuzzy as 216 00:14:35,280 --> 00:14:39,510 AI, to a previously incredibly deterministic world where 217 00:14:39,510 --> 00:14:44,010 everything is well understood, and every action has a very well 218 00:14:44,010 --> 00:14:48,870 understood reaction. What are the challenges? Well, how can 219 00:14:48,900 --> 00:14:53,700 how trustworthy is it? Is it going to get the same results 220 00:14:53,700 --> 00:14:56,640 each time? And the answer is well, not always, because if the 221 00:14:56,670 --> 00:15:00,450 input data is different, then it might respond differently. Um, 222 00:15:00,780 --> 00:15:03,840 so from a trustworthy standpoint, you have to sort of 223 00:15:03,840 --> 00:15:06,900 take a step back and think, well, there are many different 224 00:15:06,900 --> 00:15:12,780 types of AI technology, even down to something like machine 225 00:15:12,780 --> 00:15:15,450 learning, there's the concept of supervised and unsupervised 226 00:15:15,450 --> 00:15:19,860 machine learning. And so if you're gonna just want to throw 227 00:15:19,860 --> 00:15:24,690 some data at the proper system, and have it do its thing, you're 228 00:15:24,690 --> 00:15:28,050 probably describing unsupervised machine learning. There are 229 00:15:28,050 --> 00:15:30,570 certain techniques or certain areas where that's very 230 00:15:30,660 --> 00:15:36,180 applicable. That's particularly trustworthy. In areas where 231 00:15:36,180 --> 00:15:39,900 there's no real learning that needs to be done. I think a lot 232 00:15:39,900 --> 00:15:42,720 of the concern that people have over sort of machine learning 233 00:15:42,720 --> 00:15:47,970 and AI is, is where training has to occur, and how accurate is 234 00:15:47,970 --> 00:15:50,760 the training, but there are certain techniques that just 235 00:15:50,760 --> 00:15:54,960 work. So you don't need to build a model, you just react to the 236 00:15:54,960 --> 00:16:00,090 data that's coming in. And so an example would be, you have a 237 00:16:00,090 --> 00:16:03,540 flow of data coming into a system, and you're looking at 238 00:16:03,540 --> 00:16:07,110 that data in real time and trying to identify patterns. So 239 00:16:07,110 --> 00:16:10,860 you're not necessarily comparing it to a historical model, you're 240 00:16:10,860 --> 00:16:14,100 just looking at the data as it is, in real time trying to 241 00:16:14,130 --> 00:16:17,340 determine patterns. So that's a good example of unsupervised 242 00:16:17,460 --> 00:16:20,220 data, because there's no training model, you're looking 243 00:16:20,220 --> 00:16:24,090 at the data in real time, and coming up with an answer. So 244 00:16:24,090 --> 00:16:27,450 that's a good example of something where you can just 245 00:16:27,720 --> 00:16:32,040 turn it on and let it do its magic. There are other areas 246 00:16:32,040 --> 00:16:37,770 where, you know, training becomes more of a more of an 247 00:16:37,770 --> 00:16:42,690 important component. And I think, from our standpoint, when 248 00:16:42,690 --> 00:16:45,870 applying those techniques to an operations type environment, 249 00:16:46,140 --> 00:16:49,170 that's where the human becomes important. Because the 250 00:16:49,170 --> 00:16:52,950 supervised model at that point, the training is done by a human. 251 00:16:53,850 --> 00:16:58,410 So the system would say to you, this is something that I 252 00:16:58,500 --> 00:17:03,150 determined from the input data, what do you think, and the human 253 00:17:03,180 --> 00:17:07,800 has the opportunity to train it. So you know, practical turn, 254 00:17:07,800 --> 00:17:12,360 that might be the ability to tag data, or press a button to give 255 00:17:12,360 --> 00:17:19,830 it a thumbs up or a thumbs down. And that sort of human guided 256 00:17:19,830 --> 00:17:24,450 supervised learning. Again, it becomes trustworthy, because the 257 00:17:24,450 --> 00:17:27,390 human has provided the input. It's not something that the 258 00:17:27,390 --> 00:17:30,450 system has determined on its own, that you're actually giving 259 00:17:30,450 --> 00:17:34,560 it some sort of positive affirmation. So if the model is 260 00:17:34,710 --> 00:17:38,940 good, it's because a human has trained it to be good, based on 261 00:17:38,940 --> 00:17:39,930 their current knowledge. 262 00:17:41,970 --> 00:17:44,880 Unknown: The tools we use as a team have a direct influence on 263 00:17:44,880 --> 00:17:47,970 how we work together, and the success we create. We built 264 00:17:47,970 --> 00:17:51,420 range with that in mind, by balancing asynchronous check ins 265 00:17:51,420 --> 00:17:54,510 and real time collaboration, branch helps remote and hybrid 266 00:17:54,510 --> 00:17:57,210 dev teams build alignment and baton back on the calendar 267 00:17:57,930 --> 00:18:01,350 branch connects dozens of apps like JIRA and GitHub, in one 268 00:18:01,350 --> 00:18:04,290 place. So everyone can share progress and updates on work, 269 00:18:04,530 --> 00:18:08,610 making standups more focused and engaging for everyone. Visit us 270 00:18:08,610 --> 00:18:12,150 range.com/devops To learn more, and try arrange for it. 271 00:18:15,599 --> 00:18:21,059 Jason Baum: Interesting, so as a follow up to that does risk of 272 00:18:21,059 --> 00:18:25,439 getting it wrong, play into the decision of whether the machine 273 00:18:25,589 --> 00:18:31,619 is let to let it go type, like what you're saying just unleash 274 00:18:31,619 --> 00:18:35,609 it, as opposed to a human being being kind of on the other end 275 00:18:35,609 --> 00:18:39,629 sort of helping it does risk play into that of getting it 276 00:18:39,629 --> 00:18:40,049 wrong? 277 00:18:41,040 --> 00:18:42,960 Richard Whitehead: Well, the good news is, in most IT 278 00:18:42,960 --> 00:18:46,380 operations environment, the relative level of the risk is 279 00:18:46,380 --> 00:18:52,770 fairly low. But not in every case, obviously. And that's 280 00:18:52,770 --> 00:18:58,440 where a lot of concern, I think comes I have no idea who coined 281 00:18:58,440 --> 00:19:02,550 the phrase, but I like it, which is new to err is human. To 282 00:19:02,550 --> 00:19:06,600 really mess it up you need a computer. And that's one of the 283 00:19:07,680 --> 00:19:12,150 that's one of the the challenges with with automation, is that 284 00:19:12,570 --> 00:19:16,050 you can really make a problem worse by fully automating some 285 00:19:16,050 --> 00:19:22,050 kind of reaction to it. Risk is certainly an issue. When you 286 00:19:22,290 --> 00:19:25,140 look at some of the stories in the press about artificial 287 00:19:25,140 --> 00:19:30,420 intelligence. Nobody ever really publishes the good stories, 288 00:19:30,480 --> 00:19:33,030 that's just that just happened. That's life. That's we're all 289 00:19:33,030 --> 00:19:36,510 used to that. We take that for granted. It's the negative sides 290 00:19:36,540 --> 00:19:40,350 of AI that get a lot of the publicity. And, you know, 291 00:19:40,350 --> 00:19:45,810 there's a lot of concern about bias in learning models, and, 292 00:19:46,140 --> 00:19:49,710 and some of those sort of issues. And that's really sort 293 00:19:49,710 --> 00:19:54,720 of a big data problem where you're dealing with large 294 00:19:54,720 --> 00:19:57,540 amounts of data from questionable sources that have 295 00:19:57,540 --> 00:20:03,450 been used to train models. And from my standpoint, the way you 296 00:20:03,450 --> 00:20:07,650 mitigate that risk is you move away from third party data. And 297 00:20:07,650 --> 00:20:11,910 you try and focus solely on your environment. So don't use 298 00:20:12,120 --> 00:20:17,100 external training data. And you can do that in an IT operations 299 00:20:17,100 --> 00:20:20,100 environment, it's much easier to do that if you're, you're not 300 00:20:20,100 --> 00:20:24,300 dealing with sort of medical data from the last 10 years, 301 00:20:24,810 --> 00:20:29,730 that may or may not be tainted by some some, some poor poor 302 00:20:29,730 --> 00:20:32,460 quality data that was introduced that you have no control over. 303 00:20:32,820 --> 00:20:34,950 You're dealing with an IT operations environment you're 304 00:20:34,950 --> 00:20:38,070 dealing with, with infrastructure and technology 305 00:20:38,070 --> 00:20:42,390 that's in your control that you have. So you can you can build 306 00:20:42,390 --> 00:20:48,480 models and do training, from data that that high quality data 307 00:20:48,480 --> 00:20:50,490 that you that has good provenance, you know where it 308 00:20:50,490 --> 00:20:54,420 came from. So a lot of those concerns, like I say, that are 309 00:20:54,420 --> 00:20:57,990 based on poor quality models and poor quality data from 310 00:20:57,990 --> 00:21:01,950 questionable sources. The good news is it operations has less 311 00:21:01,950 --> 00:21:05,520 of a concern with that data, because we know where it comes 312 00:21:05,520 --> 00:21:05,760 from. 313 00:21:06,329 --> 00:21:10,109 Jason Baum: So with all of that, and it sounds like there's a lot 314 00:21:10,109 --> 00:21:13,409 of management that has to go on behind the scenes, who's doing 315 00:21:13,409 --> 00:21:17,999 that, who's who's going to manage the solution? How has 316 00:21:18,029 --> 00:21:22,379 like the tech team changed? How has it been the work being 317 00:21:22,379 --> 00:21:27,299 distributed? Where does AI ops play into this now? Do you need 318 00:21:27,299 --> 00:21:32,369 a data scientist? Do existing team members take on new roles? 319 00:21:32,939 --> 00:21:34,949 How are you structuring it? 320 00:21:36,660 --> 00:21:38,520 Richard Whitehead: Right? So yes, we obviously have first 321 00:21:38,520 --> 00:21:41,130 hand experience with that as a technology provider in that 322 00:21:41,130 --> 00:21:46,740 space. And the answer to the Do we need a data scientist, is if 323 00:21:46,740 --> 00:21:49,830 you're going to build a solution yourself, if you're going to 324 00:21:49,830 --> 00:21:53,040 roll your own as it were, then yes, you're going to need a data 325 00:21:53,040 --> 00:21:57,120 scientist we have, we have data scientists on board as part of 326 00:21:57,120 --> 00:22:00,660 our team. They're slightly outside the engineering team. 327 00:22:02,310 --> 00:22:04,830 And just like every other organization, they have 328 00:22:04,830 --> 00:22:10,560 different skills. They come from different backgrounds. The war 329 00:22:10,560 --> 00:22:14,760 science scientists, and they are engineers, the the sort of 330 00:22:14,760 --> 00:22:20,280 programming languages tend to be more Python are focused and so 331 00:22:20,280 --> 00:22:23,550 forth. So different people. Absolutely. If, however, you're 332 00:22:23,550 --> 00:22:29,100 in IT operations, you probably shouldn't necessarily be looking 333 00:22:29,130 --> 00:22:31,920 at getting a data scientist on board. Because there are 334 00:22:31,920 --> 00:22:35,280 technologies out there commercial technologies, open 335 00:22:35,280 --> 00:22:38,760 source technologies, where that work has been done for you. And 336 00:22:38,760 --> 00:22:41,280 I think when people ask me, you know, am I going to have to 337 00:22:41,280 --> 00:22:48,510 retrain my staff? I chuckle and say, No, the impact of AI on 338 00:22:48,540 --> 00:22:54,000 operations is minor, it's almost trivial compared to some of the 339 00:22:54,000 --> 00:22:56,940 seismic shifts we've already seen in the last five to 10 340 00:22:56,940 --> 00:23:03,060 years. As operations people, and we shift from this everything 341 00:23:03,180 --> 00:23:07,650 from to everything is code type environment, we now have 342 00:23:07,650 --> 00:23:11,700 operators who are themselves, they look just like software 343 00:23:11,700 --> 00:23:15,510 engineers, they're conversant in one, two, maybe even three 344 00:23:15,540 --> 00:23:19,740 programming languages. They're fully conversant with code 345 00:23:19,740 --> 00:23:24,900 repositories. And that that shift is far bigger than 346 00:23:24,900 --> 00:23:27,480 anything that the introduction of AI is ever going to change. 347 00:23:28,140 --> 00:23:31,710 So no, you're not going to have to become a data scientist just 348 00:23:31,710 --> 00:23:38,940 to operate this. The technology is going to be in a form that's 349 00:23:38,940 --> 00:23:42,060 easily consumable. It's going to look like software, it's going 350 00:23:42,060 --> 00:23:44,820 to like software, you'll treat it like software, you're not 351 00:23:44,820 --> 00:23:47,430 going to be building models yourself, the technology is 352 00:23:47,430 --> 00:23:50,580 going to be doing that for you. So no, don't think you'll need a 353 00:23:50,580 --> 00:23:55,980 data scientist. But absolutely, you're going to need to have 354 00:23:55,980 --> 00:23:59,040 people who are very consistently conversant with with software 355 00:23:59,040 --> 00:24:03,480 and infrastructure as code and that sort of thing. 356 00:24:04,349 --> 00:24:08,039 Jason Baum: So where does AI ops ml ops? Where does that fit 357 00:24:08,039 --> 00:24:09,989 within DevOps culture? 358 00:24:12,569 --> 00:24:14,399 Richard Whitehead: At the end of the day, it's it's just 359 00:24:14,399 --> 00:24:21,479 technology. It's it's a tool. Okay, so it's, it's neither a 360 00:24:21,479 --> 00:24:28,019 good fit nor bad fit. It's just technology. If good AI ops 361 00:24:28,019 --> 00:24:33,209 technology will fit very well. Because it just looks like 362 00:24:33,209 --> 00:24:37,079 software, it reacts like software, you can configure it 363 00:24:37,349 --> 00:24:42,149 as code. The changes you make are going to be very easy to 364 00:24:42,239 --> 00:24:50,399 work with. The technology will offer both a strong UI but also 365 00:24:51,329 --> 00:24:55,889 strong API's so the technology can fit into and be integrated 366 00:24:55,889 --> 00:25:00,299 into a DevOps tool chain. It's just part of it. Part of the 367 00:25:00,299 --> 00:25:06,149 value stream. It shouldn't stick out as necessarily being 368 00:25:06,149 --> 00:25:10,259 something that's, that's a standalone industry or a 369 00:25:10,259 --> 00:25:13,229 standalone job title, you shouldn't have to hire an AI ops 370 00:25:13,229 --> 00:25:16,679 engineer. It's just, it's just technology. 371 00:25:18,510 --> 00:25:21,990 Jason Baum: So what what are you excited about? With the future 372 00:25:21,990 --> 00:25:24,690 of AI ops? What's what's coming down the pipeline that should 373 00:25:24,690 --> 00:25:25,980 get us all excited? 374 00:25:28,020 --> 00:25:31,560 Richard Whitehead: I think, you know, for me, as somebody who is 375 00:25:31,560 --> 00:25:35,610 involved in the very early stages, just one, the first 376 00:25:35,610 --> 00:25:38,070 thing is the adoption of it. It's the fact that we've made 377 00:25:38,070 --> 00:25:42,390 that shift from this is scary, I don't know if I can trust it to, 378 00:25:43,530 --> 00:25:47,250 gosh, I can't imagine life without it. Do you remember what 379 00:25:47,250 --> 00:25:49,890 it was, like 10 years ago, when we have to do this stuff 380 00:25:49,890 --> 00:25:55,800 ourselves? How How dull and boring was that aiops also 381 00:25:55,800 --> 00:25:59,820 brings some stability. And there's a certain irony to that, 382 00:25:59,820 --> 00:26:03,450 because when we talk about things like the fuzzy logic of 383 00:26:03,450 --> 00:26:07,200 AI, people think of that as being kind of non deterministic 384 00:26:07,200 --> 00:26:11,220 and scary. The reality is, it makes systems much more, much 385 00:26:11,220 --> 00:26:15,480 more robust. So the ability for a system to be able to adapt 386 00:26:16,140 --> 00:26:21,510 means that when you get certain changes, AI adapts along with 387 00:26:21,510 --> 00:26:26,280 it, and becomes very flexible, and means that the sort of the 388 00:26:26,280 --> 00:26:30,090 total cost of ownership, the maintenance of an AI system 389 00:26:30,720 --> 00:26:35,550 drops significantly, because it's adaptive. And that's, I 390 00:26:35,550 --> 00:26:39,510 think, really significant, that that's a, that's another thing 391 00:26:39,510 --> 00:26:43,230 that just improves your sort of daily life is knowing that when 392 00:26:43,230 --> 00:26:48,000 you plug something in, yes, you're going to have to maintain 393 00:26:48,000 --> 00:26:50,220 it. But it's not something that's going to be a full time 394 00:26:50,220 --> 00:26:52,770 job. It's not something that every single day, you're going 395 00:26:52,770 --> 00:26:55,950 to have to touch and tweak. And I think, you know, people, 396 00:26:56,910 --> 00:27:01,740 people forget that when they talk about automation. And you 397 00:27:01,740 --> 00:27:05,640 hear that term, sort of no ops, floating around of like, well, 398 00:27:05,640 --> 00:27:07,860 we just fully automate everything. And that's it that 399 00:27:07,860 --> 00:27:11,850 humans can go on vacation and never touch it again. Well, 400 00:27:12,060 --> 00:27:16,740 life's not like that. One of the benefits, one of the goals, even 401 00:27:17,040 --> 00:27:20,490 of digital transformation is the ability for things to change at 402 00:27:20,940 --> 00:27:24,570 a blistering pace, you want things to be incredibly reactive 403 00:27:24,600 --> 00:27:29,340 and very dynamic. And you throw into that the natural entropy of 404 00:27:29,340 --> 00:27:33,570 any, any system, and changes absolutely guaranteed, and the 405 00:27:33,570 --> 00:27:37,530 rate of change is accelerating. So nothing's ever going to be 406 00:27:38,400 --> 00:27:42,510 installed and forgotten about, this isn't a telecommuter. Most 407 00:27:42,510 --> 00:27:44,670 of us are not dealing with a telecommunications environment, 408 00:27:45,060 --> 00:27:48,630 where you install a switch, and then you'd love it and take care 409 00:27:48,630 --> 00:27:56,670 of it for 25 years. Everything changes dramatically. So having 410 00:27:56,670 --> 00:28:00,510 a system that's at least a little bit adaptive, and doesn't 411 00:28:00,510 --> 00:28:03,840 require, you know, constant attention. You know, that's 412 00:28:03,840 --> 00:28:06,540 something that makes people very, very happy. And I think 413 00:28:06,930 --> 00:28:09,930 that's something I'm, I'm looking forward to, people seem 414 00:28:09,930 --> 00:28:14,940 to benefit from. Also the just generally looking at new 415 00:28:14,940 --> 00:28:21,840 opportunities. So as I mentioned, as we start to deploy 416 00:28:21,870 --> 00:28:24,750 AI ops, in production environments, it's the little 417 00:28:24,750 --> 00:28:27,930 things that are the game changes, the little benefits 418 00:28:27,930 --> 00:28:31,110 that are multiplied over, you know, hundreds of times a week 419 00:28:32,070 --> 00:28:34,350 that make everybody go Yeah, okay, this is really cool. I'm 420 00:28:34,350 --> 00:28:37,620 glad we installed that that made that made a big difference. 421 00:28:38,490 --> 00:28:42,660 expanding that to do some some other intriguing use cases, 422 00:28:42,840 --> 00:28:46,050 finding new cases, new use cases is something I'm really excited 423 00:28:46,050 --> 00:28:46,410 about. 424 00:28:46,829 --> 00:28:48,959 Jason Baum: It sounds like when this is going to when it's when 425 00:28:48,959 --> 00:28:51,629 it you know what's working is when you kind of forgot about 426 00:28:51,629 --> 00:28:57,629 it. All right. Yeah. Right. That's, that's the end goal. So 427 00:28:57,659 --> 00:29:01,349 I look, we're coming up to the end. This is I could talk about 428 00:29:01,349 --> 00:29:04,649 the subject forever. I think it's fascinating. I love hearing 429 00:29:04,649 --> 00:29:09,449 you speak about it. It's, it's, gosh, I can't believe we're 430 00:29:09,449 --> 00:29:13,289 here, right? This point when some of these, these mundane 431 00:29:13,289 --> 00:29:17,429 tasks are just no longer going to be a thing are already not a 432 00:29:17,429 --> 00:29:23,129 thing. So I do like to ask, kind of like, this isn't like a 433 00:29:23,129 --> 00:29:26,789 gotcha question. But but sometimes it is. Today's is not. 434 00:29:27,779 --> 00:29:32,249 I like to ask a thinker. So what's one question you wish I'd 435 00:29:32,249 --> 00:29:35,039 asked you? And how would you have answered it? 436 00:29:39,150 --> 00:29:43,620 Richard Whitehead: Um, just just from sort of a personal point of 437 00:29:43,620 --> 00:29:46,530 view as a tinkerer and an experiment, you know, I wish we 438 00:29:46,530 --> 00:29:50,580 had more time to talk about natural language processing. You 439 00:29:50,580 --> 00:29:53,610 know, I think I've been doing this for a very long time 440 00:29:53,610 --> 00:29:56,340 somebody asked me, How long have you been writing regex Richard 441 00:29:56,760 --> 00:30:01,350 and I it's, it's measured in decades. Um, I think might be 442 00:30:01,350 --> 00:30:08,040 three decades now. And for me, you know, I, I joke that, you 443 00:30:08,040 --> 00:30:10,800 know, I've only been writing regex for 30 years. So I'm a 444 00:30:10,800 --> 00:30:15,210 relative noob I'm still learning. And then along comes 445 00:30:15,210 --> 00:30:21,630 natural language processing. And by, by using sort of NLP, you 446 00:30:21,630 --> 00:30:28,260 can do things in, in a couple of seconds. That would take maybe, 447 00:30:28,320 --> 00:30:33,630 I don't know, 3030 minutes to express as a regular expression. 448 00:30:34,320 --> 00:30:38,070 And, you know, for me, there are certain things that I enjoy 449 00:30:38,070 --> 00:30:41,730 doing from from years ago, you know, I still, I still write 450 00:30:41,730 --> 00:30:46,290 code in using VI. And, you know, I still spend a lot of time on 451 00:30:46,290 --> 00:30:50,670 the command line on Linux systems. But if I never have to 452 00:30:50,670 --> 00:30:55,320 write another regex, again, I'd be a happy person. So, so that 453 00:30:55,320 --> 00:30:58,080 so the power of things like natural language processing 454 00:30:58,080 --> 00:31:05,910 just, it impresses me, and also improves my daily life. So there 455 00:31:05,910 --> 00:31:07,560 you go. That's, I answered that question. 456 00:31:07,679 --> 00:31:10,709 Jason Baum: Great. Awesome. I love it. You should have been 457 00:31:10,709 --> 00:31:14,819 interested in interviewing yourself. And you would also 458 00:31:14,819 --> 00:31:18,389 have gotten through that line better than I just did. Well, I 459 00:31:18,389 --> 00:31:21,569 really appreciate your time, Richard and educating us on AI 460 00:31:21,569 --> 00:31:26,639 ops, ml ops, and you know, how it fits into into DevOps as a 461 00:31:26,639 --> 00:31:31,349 tool and just in general makes our lives easier and not coming 462 00:31:31,349 --> 00:31:35,669 to cause doomsday. So I really appreciate you coming on. 463 00:31:36,089 --> 00:31:37,619 Richard Whitehead: It's all good. It's not Skynet. 464 00:31:38,250 --> 00:31:43,440 Jason Baum: Thank goodness. If anyone names our company Skynet, 465 00:31:43,470 --> 00:31:47,250 I think question there. There. Well, maybe just funny, I don't 466 00:31:47,250 --> 00:31:50,400 know. Well, thank you so much, Richard, I really appreciate 467 00:31:50,400 --> 00:31:53,610 your time. And thank you for listening to this episode of the 468 00:31:53,610 --> 00:31:56,610 humans of DevOps Podcast. I'm going to end this episode The 469 00:31:56,610 --> 00:32:00,390 way I always do, encouraging you to become a member of DevOps 470 00:32:00,390 --> 00:32:03,540 Institute to get access to even more great resources just like 471 00:32:03,540 --> 00:32:07,230 this one. Until next time, stay safe, stay healthy, and most of 472 00:32:07,230 --> 00:32:09,780 all, state humans live long and prosper. 473 00:32:15,450 --> 00:32:17,580 Narrator: Thanks for listening to this episode of the humans of 474 00:32:17,580 --> 00:32:21,120 DevOps podcast. Don't forget to join our global community to get 475 00:32:21,120 --> 00:32:24,480 access to even more great resources like this. Until next 476 00:32:24,480 --> 00:32:27,780 time, remember, you are part of something bigger than yourself. 477 00:32:28,230 --> 00:32:29,010 You belong