Podcast #12: From Fish Tanks to AI to Securing the IoT

Hello everybody, and welcome back to the Momenta Edge Podcast. This is Ed Maguire, Insights Partner at Momenta, and today we’ve got a special guest, its Brian Gilmore who’s responsible for IoT Advocacy at Splunk. Hi Brian, good to talk to you today. 

Thanks Ed, great to be here. Hello everybody, it’s good to be on. 

Brian and I have known each other for a couple of years at least, having met up at many-many IoT conferences in the past. I was an analyst covering the company Splunk for a while and you were right in the thick of it. 

Yes, I’ve been here for about four and a half years now. I started very early-on in terms of the IoT Initiative at Splunk, and it’s been amazing to watch the team and the business grow, it’s really exciting. 

It’s been I think an exciting opportunity to connect a lot of industries for the first time, and we’ll get into that a little bit later in the conversation. The first question I’d like to ask is, what has shaped your view of IoT, Industrial IoT? And if you could provide a bit of your background, and what was the journey that brought you to where you are today? 

My background is like a lot of folks in the tech industry. I was really fortunate to be exposed to technology very early-on, my dad was an elementary school teacher, and so he was always bringing home computers in the early eighties, especially to keep them safe and secure over the summers and things like that. My mom was actually a computer programmer, she got involved with computer science very early-on, she worked for an organization called QDATA Corporation in the late sixties, which is one of the very first computer timesharing companies. So, I was always in the midst of computers and technology which was really exciting. Most of my life I thought I wanted to be a professional musician, but very quickly discovered my computer skills would probably pay the bills a little bit better.  

I went to work for the Dana–Farber Cancer Institute in Boston, working for one of their practices on the IT side of things, and helped them to develop some database driven applications to help them manage their physician practice, and built some reasonable skills there in terms of data access and ingestion, and analytics. I then decided to do a complete pivot and go work for public aquariums; along with my hobbies and music, I was also a big salt-water fish-tank hobbyist, and so went to work for public aquariums a while, and whilst I was in public aquariums I got a lot of exposure to industrial automation systems and discovered those were also great sources of data, and I could do my job a lot better and a lot faster, and a lot more easily if I brought in some of what I’d done in the hospital and connected those systems up to really basic analytics tools, I was using Microsoft Excel at the time. 

Then I pivoted again to go work for a mechanical contracting firm that built one of the last aquariums I worked at, they asked me to come in and build some of the products I had built in a prototype phase at the aquarium for their commercial automation customers, the datacenter customers and things like that. It was really all about connecting sensor data, connecting application data, infrastructure data, getting it in front of the right stakeholders and giving them easy access. The funny part is that was IoT, sort of what we talk about IoT is now, we just never called it that. It was more about just improving operations or making the workforce more efficient. 

That’s a really unique career trajectory coming from aquatic life and music, to look really at Connected Industry. So, as you started working with mechanical systems, how would you characterize some of the unique challenges that were involved with being able to apply the similar sorts of data analytics that people take for granted with IT systems, to these complex and for a public aquarium I would assume the systems were quite specialized as well? 

Yes, the operation of the systems were very specialised, the componentry was off the shelf, we used regular sort of PLC and other type automation equipment, regular off the shelf sensors and things like that. We were really fortunate at the most automated aquarium I worked at, to work with an automation and controls vendor who was really innovative and really interested in exploring new ideas with their customers. We spent a lot of time with them building the hooks and the connections to get the data from these systems instrumented, in a way that we could bring value in the lab, or value in the life-support system office, or value in the husbandry office. 

It was more about systems integration and operational change I think, really than it was technical. It was in the end easy to get the data from the equipment into a database, and then to export from that database into something we could analyse, like Microsoft Excel at the time. Everything we built there was built on pivot tables, and hL cups and dL cups for asset information and things like that, but it was really how we presented it to the end user, and how we trained the end user on it etc., that made it a success. 

That’s an ongoing theme which we find is, this idea of pulling the data in itself from systems is not necessarily that difficult, the challenge is to put it in context which makes a lot of sense.  

You’ve been at Splunk now for around four years. Splunk of course has a far broader scope in terms of the types of businesses, and the types of industries you’ve addressed. Could you talk about some of the unique data analytics challenges you’ve faced as this IoT market has started to evolve? 

The thing you hear people talk most about in terms of challenges I think are the volume, velocity, variety, the big data challenges. I’m not necessarily convinced that’s anything unique to the IoT. Even in IT we have customers indexing and analyzing hundreds of gigabytes, or multiple terabytes, or even a couple of petabytes of data per day, there’s multi-source, multi-format, all of that. It’s a very similar, at least from analytics perspective challenge in the IoT. I think the architectures of course are a big concern in IoT, you hear a lot of questions about, ‘What do we do at the Edge?’, ‘What do we do in the Fog?’ ‘What do we do in the Cloud?’ Again, I’m not sure that’s necessarily a new thing, but in the end, I think the biggest challenge I’ve seen, at least in working with our customers, and then interfacing with the Industrial IoT community is, it’s really about again that acceptability and usability challenge. 

So, if I think about a common industrial analytics application, a services provider or a systems integrator, even a software vendor who is providing a solution for something like that has to build something that’s going to be of value from both the boiler room to the boardroom. So, you have guys who are trying to get insights and value from a system who do half their job with the wrench, and then you have this other sphere of stakeholders who are really looking at the effects on the bottom line, and things like that. I think to be able to put into an upper system which can handle that diversity of used cases, something which can satisfy both the plant floor, the CSOs office, the CIOs office, the COOs office, is really the challenge. We’ve seen a lot of really good success, and of course you also hear about on the news some of the, I would say, less than successful projects that have tried to accomplish some of those things.  

I think again it’s like anything, there’s challenges in people process and in technology, and like I’ve said, I’m still not 100 percent convinced that these are all specialized to IoT; maybe there’s some specific stuff in IoT with like sensors, like volume, variety, and veracity there, but I think generally a lot of these issues have been worked through in other enterprise or IT-type analytic solutions. 

But you’ve hit on a really interesting point there, which is there are different constituencies that have very different types of needs which need to be addressed with data analytics, like the guys who are working in the boiler room, or on the ground with machinery, versus the senior management, or financial management-types.  

Let’s talk about this industry, this idea that we’re first starting to connect machines that have not had data being collected and analysed before. Typically, where do you start, and what’s been your experience in terms of organizationally who would be the champion of implementing an analytic solution for some industrial machinery? 

To go back to something you’ve just said there, I think one of the things that’s easy to assume is the data has not been collected and it’s not already being used. An approach I’ve taken throughout my career because I want ease and fast-time to value and low cost, or is it just because I always look for the simplest path, I guess? There’s a ton of existing infrastructure in applications that really are already gathering/collecting this data for all sorts of other used cases, like automation and control, or something in the manufacturing and execution space. I think the process of connecting to and mining those systems for data that already exists, like the legacy systems, like the process historians, the ICS and SCADA systems is really the place that we’ve found a lot of success. 

Now, of course we have customers who do instrument data directly from all sorts of different industrial assets directly to Splunk, but a lot of times they’re like, ‘Look, we’ve already made this big investment over 20 or 25 years, these five vendors, and across these 50 facilities’ or whatever, and we have data to this layer. But what we really need to do then is to expose that layer to other stakeholders, and then do the enrichment, and like you said put the data in context, so we had to go out and build an eco-system to do that. One of the best things we did very early-on was, to reach out to partners like Kepware Technologies, whose now been acquired by PTC, they’re a great partner of ours and because they already have the technology to communicate with all of that legacy stuff, they can provide a really easy pathway to get data from the legacy system to the Splunk environment.  

That did two things for us, 1) It didn’t require that we duplicate any type of effort or investment that had been made in the past, but 2) It put a layer of industrial expertise between our platform and the really hyper-critical systems. Coming from the West Coast, Silicon Valley, whatever you want to call it, there’s still a lot of skepticism in the heavy industrial world about the software vendors. These guys are really picky, they’ve worked with some of these vendors for 30-years, they have very specific system integrators that they want to work with, and these are the only people they really trust to dig into their gear to get the data out, and to send it somewhere. 

So, it’s about building those relationships with both those technologies, as well as those service providers that really makes that possible. After that of course there’s tons of opportunity once you take advantage of the lower-hanging fruit of the existing data, to re-instrument or to add additional censoring and things like that. 

It’s an interesting point you’ve brought up, this characteristic of… I wouldn’t say its lack of trust, but the industrial customers tend to be obviously very careful about their technology choices, because of the stakes involved with these production applications, they’re built to be resilient.  

I wanted to move to talk about your experience working as a platform versus an application, and for listeners who many not be as familiar with Splunk, Splunk is a data analysis platform, rather than an application. I used to have these conversations with Godfrey Sullivan who was the CEO when the company became public, for several years, and we would talk about this idea of being able to turn solutions into applications. Of course, as a platform you’re spoiled for choice because you can go down a lot of different directions, but of course when you’re creating a new market, and you’re trying to identify problems to be solved, it’s really helpful to have certain patterns or business cases that you address. That’s why application vendors will be highly focused in a specific industry. 

I’d love to get your perspective now you’ve been working with the platform, and the IoT part of Splunk was started as a relatively small but still very much focused on solving problems on, we’ll say, ad-hoc basis, but you’re building solutions from the ground-up. How have you seen that evolve, and are there any use cases which stand out to you where a custom-built solution, or an idea somebody had was able to be replicated by say some of your other customers, or people internally? 

That’s a great question. We started with a platform, I think the Splunk Enterprise platform is multi-purpose, for sure. What’s interesting is, one of the things we’ve had to do a lot is to differentiate ourselves against IoT platforms, we’re not an IoT platform we’re a data analytics platform. IoT platforms like the commercial or the opensource ones we all know so well, they’re another data source for us. I guess we were pulled into the IoT by our customers, and by our partners in a way. I think we had very early success with a project I worked on before I worked for Splunk at Eglin Airforce base pulling in data from the smart buildings and the sensors, to help them optimize their energy consumption and utilization of their facilities. I think we had very early public success with New York Air Brake and their use of Splunk, to analyze data from the locomotive and the breaking control systems.  

These were people at these companies who had either had some exposure to Splunk before, like I knew the guy from New York Air Brake had worked with Splunk at another customer before he took the job in New York Air Brake, or they ran into it at a trade show where somebody else in their organisation just passed it onto them. They just connected with it, they just understood, ‘Wait a second, this is what I’ve been trying to do for so long. It’s easy to connect the data up and then once I’m connected I can analyse that data, explore it, build dashboards on it, everything’s in real-time, it feels really good. It gets you off the ground to that level one of maybe searching and exploration, and things like that, really quickly. 

Then what we’ve discovered, whether it’s New York Air Brake, or whether its DV Cargo, and now we have DV Cargo who is using us for a locomotive maintenance, and other use cases in Germany we have Real Flying Doctor Service doing airplanes, we had all these different customers who were all customizing and building their own custom applications on top of Splunk. Part of what we do in being a platform is that customers can build everything from dashboards, all the way up to full-fledged standalone applications where somebody goes to a webpage and log in, they have access to asset analytics, or ICOs and SCADA Cyber Security solutions. We spent a lot of time with the customers over I would say the first 2 to 3 years that I worked here, understanding what were the customizations they were doing? What were they struggling with? What did the feel was much harder than it needed to be? Or it was slower than it needed to be, or it was more expensive from an hour’s perspective than it needed to be? 

So, we started in the IoT space, two weeks ago we launched this Industrial Asset Intelligence announcement where we’re talking about limited availability release that we’re coming out with here soon, and this is that sort of that next iteration of the Splunk platform where it sits on top of Splunk, it runs in Splunk, but it gets those types of users a little bit further towards that complete solution. But again, you’re still going to have all kinds of customization in the end, because one thing I’ve discovered is, every IoT customer, every industrial customer, even when they’re very similar in terms of manufacturing customers, or transportation customer,  or whatever, they all have something very specific they need to do based on the specifics of their operations, or their organization etc. etc. So, you have to leave that little bit of flexibility in there, and remain a platform and help the customer either build turnkey solutions themselves, or through systems integrators and service providers, if that makes sense? 

Yes. The Industrial Asset Intelligence market addresses a business problem that’s near and dear to the team here at Momenta, we work with a number of our clients and people in the eco-system around these problems. I’d be interested to get a sense from you, what has led you to dedicate a lot more focus around Asset Intelligence, and also as you begin to incorporate more and more predictive analytics and AI capabilities, to what extent do these types of approaches, these solutions, to what extent will they always remain a solution that will lead a significant degree of customization versus getting close to an application where say 80 percent of the problem gets solved, and then the last 20 percent can be automated. I would love to just get your perspective on that. 

I think you hit the nail on the head with your last comment there. I think the 80 percent solution is really that sweet-spot, where an application has to pull in a set of capabilities that’s going to apply to a broad number of users. So, with Asset Intelligence we’ve added in some usability features for people who may be interacting with response from the plant floor, rather than the data centre or the data science office. There’s requirements in terms of interface and user experience, and so we’ve added some things in there to add, I would say, expected features for some of these personas.  

Then when you’re talking about things like machine learning, this is clearly something that’s on our road map, and something that’s anybody who’s looking to do analytics, or is doing analytics in the industrial space is focused on. I think we’re at this really interesting point right now where the power and the capability of machine learning, I hesitate to call it AI, but of machine learning is very well understood. But it’s the application of it that gets really difficult, so for example, you can say, ‘I’m going to use such-and-such a library to forecast or cost/detect anomalies’, or whatever it might be. But then, when you really get down to things like feature preparation and extraction, or the enrichment strategy, or even which specific algorithm for forecasting to choose. You really have to have a lot of domain expertise, not only in what it is you’re trying to predict, but then also a lot of data science expertise as well.  

So, we’re in this early phase where the machine learning portion of it is still much more services-heavy, either customer in terms of hours, or service provider in terms of services, than a lot of the rest of it, it’s a little bit more advanced in terms of self-serve and things like that. 

I’d be interested to get your take, as this market has evolved, whether you’ve seen any industries or types of users, or specific use cases that really stand out as being able to apply technology in creative ways, and think of it… I hate to use the term, outside the box, but industries at least that are thinking ahead or outside of say conventional wisdom and are able to really affect meaningful transformation of the business. 

I think there’s a lot of emerging technologies of course that everybody’s paying a lot of attention to. I think the technology is I would say outpacing the application right now. You talk to customers a lot of times, you definitely want to say the really big sophisticated organizations like right after the machine learning, and the Artificial Intelligence, they want to apply augmented reality to improve safety, or they’re looking for ways to apply blockchain to solve something in their supply chain, or whatever it might be. But then again there’s this whole set of customers who are really just looking to do business better, or cheaper, or faster, or more safely, or more secure, and when you talk to them, they’re like, ‘We would just love real-time visibility of the plant floor outside the plant floor’. That’s a very simple used case that you hear from a lot of customers, or, we’re looking at high-level KPIs on a department level, or on an organization level, and we’d love to take into account real-time information from the production environment, or from the vehicle fleet; if we’re looking at a health score for the company and its flipped from 75 to 55 over the past 24-hours, how do I drill down through that and see what line of business is that coming from? Okay, well it’s coming from manufacturing, or its coming from logistics. If you drill down to logistics what is it? Okay, we have all these key performance indicators for vehicle maintenance, or safety procedures and things like that. They want to be able to drill down through all of these different KPIs to basically get to the root cause where at least from the executive office perspective, they at least know who they need to contact to say, ‘Hey, what’s going on? Are you aware that this is affecting the business?’ 

Those types of use cases they’re vast right now, you talk to a lot of different customers, and that’s where they’re trying to head. From I would say the most stereotypical used cases when you take a look at things like predictive maintenance I think clearly is a huge driver, but predictive maintenance to me is just a piece of reduction of unplanned downtime as well. There’s a number of different strategies to take to get to that new unplanned downtime, predictive maintenance being one of them. So, we see a lot of customers using analytics to build predictive maintenance strategies, but also then to monitor the performance of those predictive maintenance strategies. 

The use case that I’ve found the most compelling, and I think from especially the industrial IoT perspective is really right now the most urgent, is the cyber security perspective. You’ve got the really interesting double-edged sword of connectivity, where it gives you access to a whole bunch of new information, and new insights and all of that, but in a way that connectivity also increases the risk to your business if it’s not done well, or if it’s not monitored well. So, I think people must think of the security as a sort of side-effect of the operations, or the other types of things that people are trying to improve with the IoT, the industrial IoT, but I think it has to be viewed much more as a fundamental component of it. You hear a lot of companies talking about secure by design in IoT and industrial IoT now, and I think all of that is a really good idea because if it’s not secure it is going to inject risk or increase risk, rather than reduce it. 

It’s amazing how important that is. For many companies there might be a trigger that will cause them to end up making security a priority. I saw you at the Treaty in Niagara Conference earlier this week, and just chatting with them it was remarkable too that their CTO Kevin Smith is very focused on cyber security, just because when you have these millions of assets that are connected, every one of them ends up being a vector for vulnerability. Clearly the ability to secure, at least to anticipate any vulnerabilities is really critical. 

Yes, you can only look at them as a vector for vulnerability as well, but they’re also a target. It’s funny, we work quite a bit with Booz Allen Hamilton around ICS and SCADA Cyber Security challenge, and a lot of the research they’ve done I’ve enjoyed reading it, they talk about as they’ve worked with customers, and we’ve seen this as we work with customers as well, that the assumption has always been that IoT is going to be an increase in surface, that is going to be leveraged by threats and will increase access or risk to the enterprise network.  

But you see a lot of trying to penetrate the OT or the IoT network to actually affect the IoT devices; or even coming in the other way, can you get on the enterprise network and somehow traverse to the OT or to the Operations Network, and do all kinds of incredibly damaging things over there in terms of just shutting equipment off, or putting assets or even humans at risk? It’s all really scary to think about people doing that just because they’re malicious, or because they’re trying to sabotage things, or they’re trying ransomwares or something. I look at that whole space and just find it amazing that there’s not tons of ransomware attacks going on in the OT space. When I think about that and all of the things that need to be solved right away, I think that’s one of the biggest ones for sure.  

The next thing is the technology is all there, and the best practices are there, they have to be customized and they have to be catered for sure, but I think generally the IT security best practices would get the OT and the IoT security much further along I think, than people expect.  

It’s reassuring to hear that, I think most people are still struggling with the amazing amount of confusion that faces them when they’re dealing with the IT security landscape, but the reality is, one of our earlier podcasts with David Bower who used to be CISO at Merrill Lynch, and he was in Morgan Stanley before that, there are some best practices that are out there, and if you just follow some basic hygiene and apply a bit of preventive care, you can really avoid a lot of heartache down the road, there’s no doubt about that. 

The news, which is kind of old news, I think it’s from the past summer, is of course the news about the fish tank being used to exfiltrate data from the Casino and the high-roller database and things like that. When you think about that, first of all everybody calls it a temperature sensor, and there was a temperature sensor involved, but my guess is that the sensor itself was not a piece of this, it was probably the unsecured Linux or Windows box that that thing was attached to. I think generally when you have devices that have operating systems, that are exploitable on your network and you’re not aware of them, they’re a risk generally whether or not it’s an IT device, I can bring my Echo in and pop it on my desk and connect it up to the network, that’s something I can do. I can also bring in a little embedded PC and do the same thing. So, I think the practices of understanding are:- 

• What’s on the network? 
• What it is? 

• If its protected? 
• Who owns it? 
• Is it trusted? 

All of these different types of things, that goes across IT, IoT, OT, etc. and again it’s just not something that’s specific to IoT in my mind. 

That’s a great point.  I wanted to ask one question about the serendipity that comes from doing data, analytics, and exploration. We’ve talked about this need to address real pain points, unplanned downtime, but I think some of the interesting case studies that I’ve heard about, people using platforms such as yours, or data mining platforms is what they find out in second order benefits as it were. I think you might have told me about somebody that was analyzing elevator data, a landlord discovering some interesting things. Do you have any thoughts about what you can find with just using a tool to explore data once you’ve got it? 

I think the elevator story you’re talking about it actually pre-dates me, so it may have been Godfrey who told you about it. One of our customers very early-on, they were using Splunk to analyze the operations of the elevators in the buildings that they managed for their customers. What they discovered was, it was a side-effect that said all this information, not only on the performance of the elevators, but also where the elevators were stopping, by being able to trend-line that information over time they could see for a particular floor, the number of stops per day decreased from 600 in one month, to 300 six-months later, and then 150 three-months later. They could go back to their customers and understand what was the cause of that, had they downsized, did they need less floorspace, was their business in trouble, did they want to move to another facility? 

Those types of like you said, serendipitous used cases for me is really what it’s all about. By having new access, and to be able to put people who think about the processes and the operations of these businesses, or of these assets or mechanical systems, whatever they are, by giving them a playground to go in and search, explore, and analyse, it’s a culmination of intuition support. I remember the first time I really had unfettered data access from one of my systems, I could literally go in and say it feels to me like whenever ‘this’ happens, this always happens. Suddenly I had this method to go, ‘Whoa, how could I be so wrong?’ or, ‘Oh my God I’m right. I wonder what’s causing that?’ Then you pivot, you find the cause and then you can actually say, ‘Well okay, if I can monitor for that cause, and I can alert when that thing happened, I can prevent this other much worse thing from happening’. 

That’s what people do when they’re using these systems, is they’re exploring, they’re understanding, they’re learning. I think another really interesting and totally unexpected secondary benefit from systems like this for me, has been the pathway and the new communication in the organisation. One of my old bosses before I worked for Splunk, we used Splunk at the company that I worked at before, he called it a campfire technology of how he’d seen it work at the Airforce Base, which I talked about earlier. When I asked him what he meant by that he said, ‘There’s certain technologies where people are drawn to them, number one. Then when they arrive they sit down, they stop, and they talk and communicate. It was a shock, he was totally right, because even at the Airforce you saw people who were literally cleaning and replacing air filters, having conversations with the energy managers, with the commanders, all the brass, and all of that, about stuff they were all seeing now across all of their facilities.  

It wasn’t just, ‘We’re consuming more power’, it was, ‘Hey, should we think about moving these conferences to this other building, because it’s just much more energy efficient than the building we’ve been doing these big conferences in. Or, ‘Hey, if we need to build another building, let’s see if the data can tell us which buildings are efficient, and if we can understand something about them from a perspective of:- 

• When they were built. 

• How they were built.  
• How they’re roofed.  
• How many square feet they are. 
• How many hours a day they’re occupied. 

All of these different types of things. When you have a system that gives you that access to data, and it lets you be creative and curious, you’re going to solve these cases far beyond what you decided to begin with, and probably what you used to validate the investment and the platform to begin with. It’s those things I feel make a platform super-sticky, and super-long-lived, and really make it the place that people go to get understanding. 

You’ve really highlighted the value of domain knowledge and context. When you combine it with the analytics and the data itself, you really need to have people who were deeply embedded in the processes and the businesses, and the functions of their daily lives in their organizations, to be really able to tease out those nuggets of insight that are buried in the data. 

Absolutely. Being a data-driven organization gets you so far. Being an organization who hires and allows people to become data-driven is really I think the Holy Grail, and that works no matter what the data source is, whether it’s IT, IoT, OT etc. 

I’d like to just turn the conversation forward and get your views on where you think the market is headed, as you’ve seen the evolution, of what we now call Industrial IoT, over the past several years. What do you think are some of the key developments ahead of us, and forces that will shape how we see adoption of advanced technologies and potential transformation playing out over the next several years? 

Honestly, right now it feels to me sort of IoT-driven outcomes, and the use of the word IoT are kind of inversely proportional. I expect we’ll see a decline of the term IoT, and more adoption of what we describe today as IoT. I believe in order for it to become ubiquitous and commonplace, it’s going to have to become transparent. We rarely talk about TCP/IP anymore, but we use it all day every day kind of thing, so I think that’s one thing. There has to be a lot of consolidation in the IoT platform space as well, it causes quite a bit of confusion in terms of you’ve got so many vendors and their products are all so great, but all so similar. Then of course you have a lot of the legacy vendors who are saying, what you already have is basically an IoT platform as well, and you can use it there. 

So, my guess is there’s going to be a lot of consolidation in that space. I would imagine eventually you’re going to get down to five to ten leading vendors in the IoT platforms, and then five to ten leading vendors in the ICS and SCADA space, and then you’re going to have I would say a significant number of open source or new open source solutions that overlap with both the commercial IoT platforms, and then the commercial legacy software, just because it’s one place that open source really hasn’t penetrated yet. But I think as people become more and more comfortable with open source, just as they’ve become more and more comfortable with cloud, you’re going to see a lot more influence of those two technologies, or those two paradigms whatever you want to call them, on IoT and on the OT space. 

That’s great. I like to wrap up my conversations with a request for resources or recommendations, and I have to say it’s been super-informative and illuminating talking to you about your experiences and perspective. I’d love to get a recommendation or two, if there’s any books or other resources that you’d like to share, either with friends or professional colleagues? 

I think one of the most I would say thought-provoking books I’ve read in a long time is a sort of hybrid non-fiction/fiction book called, ‘Life 3.0: Being Human in the Age of Artificial Intelligence’ by Max Tegmark. 

I’m familiar with him, but I haven’t read the book, he’s absolutely quite illuminating in his field. 

He’s drawing a lot from the sort of Elon-type perspective, as well as some of the other futurists in terms of AI. It’s really fascinating to get that vision of where AI could go, in both the short and the long-term, and the impact that it could have on being human. Thinking outside of even the IoT space, if you had told us 15 or 20 years ago, the major impact that even social media would have on our world, I would have said, ‘You’re absolutely nuts, it’s never going to have those types of impacts across so many facets of our lives’, and based on my early read of this book I think AI will be many, many times more impactful. 

The book covers a lot of stuff that is of interest to me, one of the things that I’ve been really fascinated in, I got to do some work with an autonomous vehicle organization not too long ago, and a lot of these technologies I think are starting to force the requirement for a real deep imagining of what the moral implications are of some of these technologies as well. I think he does a really good job through covering the convergence of technology, immorality, and spirituality and all sorts of things. Its areas when you hear about so much about just the bits and the bytes of the technology, to really understand the broader implications is pretty fun for me. 

It’s a terrific recommendation, and I think a lot of times when we’re deeply immersed and ensconced in understanding and implementing technologies, we don’t always think about the downstream implications and the human implications. It sounds like a fascinating book, I’m putting it on my list for up next. 

Again Brian, I want to thank you for taking the time to share your insights with us.  We’ve been listening to Brian Gilmore who is the IoT Evangelist at Splunk, and this is Ed Maguire, Insights Partner at Momenta Partners. Thank you all for joining us for another Edge Podcast, and we hope you’ve enjoyed your experience. Please send us any comments, questions, and follow-up.