Enterprise AI: Data Democratization and Cloud Computing - CXOTalk #804

Enterprise AI: Data Democratization and Cloud Computing - CXOTalk #804

Show Video

Today on Episode #804 of CXOTalk, we're talking  about the democratization of technology and the   impact on AI. Our guest is Bob Muglia.  He is the former CEO of Snowflake and   has had a legendary career. I'm here with  my amazing guest co-host QuHarrison Terry.  Bob, as we introduce you, I will  hold up your book. It is called The   Datapreneurs. You have this extraordinary  background in the computer industry.  I joined Microsoft and spent 23 years  there. The first technical guy on  

SQL Server. I helped to build that business. I spent all my time in the product groups. Ran,   in some senses, almost everything at Microsoft for  a little while, except for games. Never did that.  I spent time in Widows Server. I ran Visual  Studio. Helped to put that together. I spent  

time in the Office group and MSN. Then, for the last seven years, was   running the server and tools group. Was president  there and grew that business to about $17 billion.  Since then, by the way, it's gotten a lot bigger.  That business has grown a lot since then. I helped   work with Scott Guthrie at Microsoft now, and it's  many times that size now. It's pretty remarkable.  After that, I spent a couple of  years at Juniper down in the Bay   Area. Then I joined Snowflake in the middle  of 2014 and ran the company for five years. 

I took it from zero revenue. That's an  easy number to remember. It's not hard   to remember zero. And took it to just about  $200 million in revenue before I moved on.  What would you classify as a datapreneur? I'm working with entrepreneurs now. What   I've been since I left Snowflake is  really helping small companies to   grow and investing in small companies. I'm on  about five boards of private, small companies  

all involved in data in one way or another. I realized that when I was at Microsoft, even   though I was working for a really large company,  I was working with entrepreneurs the whole time,   data entrepreneurs or datapreneurs, really. My  role has always been pretty consistently to take,   help, and work with these brilliant people  that have built something truly amazing   and help them to turn it into a viable  product that sells in the marketplace.  That's kind of my sweet spot is helping to  define the technology so that it's something   people want to buy, and then helping to bring it  to market, price it, package it, and sell it in   the marketplace in a way that makes it successful,  which is why I've been focusing on really   early-stage companies because that's what they do.  They're trying to build things from the ground up.  I realized, when I was at Juniper, I tried  to fix a bunch of things. It's really great,   and I really respect the people that go into  broken areas and like to fix them. I decided  

that I prefer building things instead of  fixing somebody else's broken something,   and so that's what I've been working on. Do you want to give us just a very brief   overview of the book and why you wrote it? I felt like I had something to say. I wanted   to put it in a way that was reasonably easy for  people to read, so I worked with a co-author,   Steve Hamm, who was extremely helpful. Frankly,  I could never have gotten the book done without   Steve's help. We decided to focus on the  people aspects of this, the datapreneurs that  

were actually building this incredible software. One of the key elements of this book is something   called "The Arc of Data Innovation." This is  this idea that although we look around us and   we see AI as this big, new thing, the reality  is that while it's new to us, the technology   has been built and created over a period of time. It's all built on incredible work that's been done   over decades and decades, a period. And so, I told  that story in The Datapreneurs, and I describe it   in this context of this arc of innovation, which  is key technologies that have been invented over   the last 40 or 50 years that have led us to  where we are today. We're now in a period where  

this work that's been done over these decades is  really paying off in some very significant ways,   as we have this new AI technology that  we're all looking at and thinking about,   "How is that going to affect our lives?" How should I organize data in my organization   if I'm starting at zero versus if I'm already  an incumbent with a lot of legacy technical   debt? What are the best practices? Well, it's easier when you're starting   from scratch because, if you start in 2023, you  have the incredible, modern data stack to build   on top of. The modern data stack are really a  set of services that work together to provide   Internet scale, incredibly large-scale working  with data, allowing people (companies of really   any size) to work with all of their data. Now it's really all of the different types   of data. Classically, it was structured  data coming out of business systems,  

as well as potentially semi-structured  log data that people were analyzing to   understand the behavior of applications. Now we're in a world where there are new   data sources that are super-interesting like  video and documents and speech. All of these are   elements of data that are now being collected  as a part of companies and can be analyzed. 

What I would do, I mean if I'm starting  from scratch, it's pretty simple. You   subscribe to a bunch of SaaS services. Today, businesses are run on SaaS services,   especially small businesses where they  run very little in-house and in-house   data centers. Maybe nothing, probably,  in fact. Everything is up in the cloud. 

You work with Salesforce, and you work with some  product support company, and you work with Datadog   or something for your operations. You have all  these different SaaS companies you're working   with that are running your business,  and all of those are generating data.  Typically, you have some business  system that is a core part of what   you've created. That can generate a lot of data. Now it's pretty straightforward to take products  

like Fivetran, which allows you to take that  data off of those SaaS systems and collect it   into a centralized data warehouse. Products  like Snowflake, or many other products in   the industry because there are really five  vendors that provide the modern data stack.  Snowflake and Databricks plus the three cloud  vendors (Amazon, Microsoft, and Google) all   have their own offerings in this space. You  choose a platform, you choose a data pipeline   vendor, and you start working with data. You choose a BI tool (Tableau, Power BI,  

something like that), and you can start working  with data. You can begin to do machine learning.  Everything is available for you. Everything  can be in one place. It's much easier.  In 2014, it was so much harder. It was  so much harder before that existed.   Now it's all coming together for people. Please subscribe to our newsletter. Hit   the subscribe button on our website. Subscribe  to our YouTube channel. Check out CXOTalk.com.   We have incredible shows coming up. Bob, this concept of democratizing  

data is so important. What does that have to do  with creating the AI growth that we have today?  This idea of making technology available to  everyone has been core to what I've certainly   been focused on in my entire career. When I look back at what we did (and   my teams did) at Microsoft in some of the  early days (the 1990s and into the 2000s),   what I'm very proud of – and maybe one of the  things I'm the most proud of – is that when you   looked at technology before that timeframe, it was  really only available to the largest companies.  People would buy mainframes, or they'd  buy these big, expensive mini-computers.   Hundreds of thousands or millions of  dollars of outlay in order to buy the   technology required to run a business. Well, Microsoft changed all of that in   the 1990s with products like Windows Server and  SQL Server, both of which are products I spent   a lot of time on. That allowed companies  of essentially any size to run a business. 

Go back to 1990. Your dentist was totally paper  and pencil. The small business, the dry cleaner   down at the corner, was probably also paper and  pencil. That's how business ran in that timeframe.   It was in the 1990s when technology like Small  Business Server, Windows Server, SQL Server,   all of those technologies packaged together that  allowed companies to build applications that ran,   were used, and were available to small businesses. Now, of course, you look everywhere and everything  

is computerized. Now it's services. But back  then, it was on-premises systems running on   a little server in the closet that was in  those rooms. People could do that for tens of   thousands of dollars; much, much less costly  than the alternatives that were available.  Fast-forward to today. We're in a services world  where everything is a service and everything can  

be purchased through some sort of subscription.  You just apply to it and you pay for what you   use. Now it's very cost-effective, or at least  reasonably cost-effective, for organizations   of any size to work with information and to treat  information and data as one of their core assets,   which I think everybody should do. The data you have and that you collect   is a critical element of everybody's business.  Now it is reasonably straightforward to make use  

of that with these tools like the modern data  stack and the products that are part of that.  Now we have this whole set of artificial  intelligence models that can be built on   top of this data foundation that can be set up in  organizations. What that does and what's changed   in 2023, which really didn't exist 24 months  ago – this is pretty new – is that we now have,   with these artificial intelligence,  this idea that there is literally   intelligence inside a computer that you can  take and teach to do things on your behalf.  The way I often describe this is that people have  skills that they've learned in a given domain,   and they understand the attributes of that domain:  how things work, how this talks to that. That's  

knowledge that everyone has and it's up in their  head. What's now possible for the first time is   to kind of bottle that knowledge, to take that  knowledge that you have and stick it inside one   of these AI models and allow that model to  do much of that work for you. That provides   a whole new set of opportunities and makes  things possible that just were not possible,   like I say, 24 months ago. It's pretty exciting. On the one hand, you've got this ease of access   because of all of the SaaS tools and the lower  costs, lower barriers to entry that you were   describing earlier. However, AI is so different  because of the open-ended nature of the result.   And so, where does the cultural element come into  play that, "Okay, we have all of these tools,   but our organization is hierarchical, we're  structured, and we have silos. We want predictable   results, and AI is not doing that"? There is  this cultural element. How does that fit in? 

One of the most important things I've said is  that when you were building these services that   you purchase or you create for your internal  users, those services (in one way or another)   imbue the values of your organization. You  can see that in decisions that get made,   in the way those products are offered, and the  kinds of things that they present to end users.  The culture comes through.  The values come through.  This has been true for a long time. I could  see it in products that I knew well. I could   understand how decisions that were made by  people and the values of those people were   reflected in the products that they built. You can actually look at that and understand  

that if you can trace it back to the people that  built it. Now multiply that by 100 because the   values of the organization are going to get  programmed into these models that get created.  It's important to really understand your  values. That's something that I think is   incredibly important for every company to do. I  encourage every small company I work with to build   their values and really live by their values. That was something that was very important to   me when I built Snowflake was that it was going to  be a values-based company. Fortunately, the team,  

they wanted it, and we created a company  that people liked to work with. It wasn't   just a good product but it was a good company  as well because it was a values-based company.  Well, those values are now going to get imbued  in all of the AI that gets created by every   company. I think this is just an opportunity  for people to really understand what they're   all about and then take the core of that and  implement it in these models. Now it starts   to become reproducible because what we're  going to start to see is a lot of questions   are going to get answered by these models. One of the areas of initial progress that   the technology is really ready for right now is  to solve problems like helping people to answer   product support questions without having to  go to a product support specialist. Models  

are really good at answering questions like that.  And if we can augment the models with knowledge,   we can make sure the models answer the  questions correctly for the customer.  But again, the values will begin to show  through in this, so it's really important   that we think about that and build it that way. When you go back just ten years ago (in 2013)   and look at it, you were ruminating on  the concept of starting Snowflake and   getting going in that engine. You looked at the  industry, and you said, "This is where we're  

going. We're going to be more data-centered,  and these are the things that are missing."  In your book, you talk about the next 20  years. There's a point where most futurists,   they show the exponential chart, and  they say, "This is the singularity." 

However, in your book, you added a few  steps in between there. You said, "Yes,   this is where we're at with ChatGPT and large  language models. But before we get to singularity,   there are all these fundamental steps that  need to be filled in." Where is the future  

thinker in Bob at today and what do you think  a technologist needs to solve for the next ten?  Well, in the short run, I think what everybody  is really focusing on in the next 12 months or   so is to take this technology that exists in  foundation models (primarily large language   models) and apply it in applications. We're  mostly waiting for the apps right now.  We've had this incredible hype cycle in the first  six months of this year. I've never seen hype   bigger than the AI hype, which is good, I think. It deserved it. I actually think it deserved  

it. I'm pleased it happened, but it did. Now I think we're kind of at the peak of   the hype. In Gartner terms, we're sort of starting  to enter the trough of disillusionment right now   as people wait for the applications. But in this case, I think, again,   following Gartner, I think that's going to be a  short trough. Pretty soon, next year, we'll be   onto the slope of enlightenment where people begin  to understand how the technology will really be   applied in their business. I think that will start  to happen in the next year as we see more products   become finalized and people can begin working with  them; the Adobe products, the Microsoft products,   plus the incorporation of AI in virtually  every other product that people are using. 

I think the technology that sits behind the AI is  improving dramatically so that, over the next 12   months or so, people will be able to leverage  this AI as a part of their modern data stack   solution using Databricks, Snowflake, Fabric,  or BigQuery – whatever they want to use – and   to work inside that environment to actually build  AI applications for themselves. They'll understand   better how to do it. We're at that stage right now  where AI is being established into the industry.  Between now and then, I see advancements in  databases. A lot of my focus has been on how  

databases will change in the next ten years. I think relational technology is ready for a   breakthrough in the sense that it's ready to  begin to leave SQL behind. SQL and relational   have been copesetic and tied together since  IBM invented both in the 1970s. It's great,   and SQL is fantastic. It's very appropriate  for working with structured data.  But now there are all these different  kinds of data. We have semi-structured   data. We have complex data in the form  of videos and documents and things. 

Relational technology can apply to that but  we're being held back in some senses by SQL.   While I see SQL continuing to be incredibly  important; very, very critical; still the   standard for working with data and slicing and  dicing data; I think we'll begin to see products   that provide much more sophisticated database  technologies that break free of the box of SQL.  To the point where we'll have  just unstructured databases   that still can give us the same outputs as SQL? There's no such thing as an unstructured database.   There's no such thing truly as an unstructured  file. What would an unstructured file do?  People call this unstructured. It's not  unstructured. It's complex structure.  A video is complex structure. A  document is a very complex structure. 

These are not unstructured documents.  These are not unstructured things.   We've just talked about them that way  because they've been opaque – not to us.  You can take a picture, an unstructured  picture of a horse eating hay in a barn,   and a four-year-old can identify that. But until  a few years ago, a computer just thought it was  

a bunch of bits. Now, with machine learning, a  computer will identify that as a horse eating hay.  It's not unstructured. It's now data. All of  these formats are rich opportunities with data.  Relational is a very broad set of mathematics.  It can be applied to data of any shape.  Where we've worked with data in the form  of tables, we can now start to work with   it relationally in the form of semi-structured  documents as well as any shape. That's where   this idea of a knowledge graph comes in  that you can take and create a shape of   objects that can describe almost anything. Most importantly, we will begin to see knowledge  

graphs model business process. The world is  moving to modeling, in general, whether they   are explicit models done through a relational  technology or whether they are statistical   models based on these new neural network-based  machine learning artificial intelligence things.  There are still models and, more and more, we'll  move to modeling our world both explicitly and   statistically. That will really help people  to understand what their business is all   about and make better business decisions and  drive things forward; figure out what to do. 

Let's shift gears slightly. This discussion  of models is a very important one,   but another important aspect of all of this is  the cloud. You have been involved with the cloud,   helping shape what we think of as  the modern cloud computing. And so,  

where does cloud fit into again the possibility  of AI and what we're currently experiencing?  The future is cloud. I'll just say that. I think  the future is that more and more companies will   adopt the cloud in one sense or another to  run their business. Even companies that think   of themselves as running things on-premises will  use a variety of cloud services in a lot of ways.  The cloud continues to expand. One of the  most interesting places where the cloud   is moving to is the edge; devices that get  closer to you inside cities and things. I   think that's going to become progressively more  important as we move into a world of robotics. 

One of the key things that I think is going  to happen over the next 10 to 15 years is we   will enter what I think of as the era of  robotics where robots of various shapes   will be flying around as drones, walking around  in our houses, eventually. They'll be running   down the sidewalk doing all sorts of things. These robots will be working and interacting   with us in our lives, and I think the cloud  will control all of that. It will start in  

these centralized, massive data centers, but  it will extend out to edge data centers that   are in every city around the world to help  control these things effectively in real time.  If you have two drones flying around,  they better know where each other is   because you don't want these things to crash  into each other. Traffic control is going to   be a whole new set of things. You really need a  cloud-like system to do that. It's the only way  

you can possibly solve these problems. While on-premises remains important,   customers have security concerns. Some companies  feel that's very crucial to them. More and more,   these things are going to be cloud-based services. In general, data collection in the cloud and   analysis in the cloud is so much more appropriate  because it's very bursty in its nature. And so,  

instead of having to have dedicated  computer resources to do things, you   just have compute resources when you need them. We have some members in the audience that have   a ton of questions. I'm going to take one of  the questions right now. It's from a frequent   CXOTalk listener and viewer of the show,  Arsalan Khan. He's got a really good question.  What he's asking is, if we all become data  collectors and we start to just unearth and   just say, "This data is valuable. This is  that. This is this," who is going to be   the enforcer and really make sure we need to  collect this data, how do we store this data,   what are the privacy laws pertaining to the data?  What do you feel is the right outlook for that?  The issues of regulation is always a really  interesting conversation. There are clearly  

steps of things that are very important. Different parts of the world have different   views on privacy. Europe's view on privacy is  different than the view in the United States, and   certainly different than in some Asian countries. Those sorts of compliance requirements have   to be taken into account in everything. I  think that the tools are going to continue  

to improve to make this easier and easier. GDPR was actually, in my opinion, a big step   forward. The alternatives were much worse before  then. The alternative was confusion before then.   GDPR at least makes things clear (or generally,  reasonably clear) as to what you need to do. 

Remember there's no such thing as unstructured  data, so there's no such thing as confusion in   your world. [Laughter]  Well, there can be confusion. There are a lot  of things that are confusing in this world.  I was just going to say that. [Laughter] There are a lot of confusing things in this world.  Yeah. Frankly,  

a lot of regulations are pretty confusing.  You look at some of these regulations,   and they're almost impossible to understand. But compliance will always be an important   part of this and, in particular, access control.  All of these things are very important. Frankly,   there are some problems that still need  to be addressed in the modern data stack.  It's not hard to put access control on your  data in the modern data stack. What's a little  

hard is to manage that and to understand  exactly what access people have. That's   still pretty hard. Those problems will get  solved over the next few years, though.  You actually need knowledge graphs. The thing  that's interesting is that all these problems   are appearing where it's becoming clear you  need a semantic model for something. You need  

to describe what you're trying to accomplish. Semantic models for businesses, this is really   important. Where are the business rules? Where do  the business rules exist inside your organization?  An interesting question for any CXO to  ask: Where are those rules? Do they exist?  Well, I'm going to tell you where they  mostly exist. They exist inside applications,   often very opaque. You don't know what the  hell they're doing, but they're in there.   These are applications you're running. They exist in Slack messages. They exist  

on whiteboards. They exist in people's heads.  They are almost never declared explicitly and   never are they declared explicitly in a  way where they can be operationalized.  That's all going to change with knowledge  graphs. That's what knowledge graphs are all   about is to take what's today implicit about the  business and make it explicit, and to define the   semantics associated with the business. Now here's the interesting learning,   and I've heard this now from at least eight  different sources that are people trying to   use these large language models to do things. One of the areas of advancement that people  

are trying to solve is to use English as  a BI language, as a business intelligence   language. Instead of having to write a SQL query  directly, you could just ask a model to find your   data for you. Well, what people are finding  is they try and build these systems is that,   to solve that problem, they need to have some  kind of semantic model sitting behind the large   language model to explain how the data works. There are these very simple demos. You have   orders, and you've got customers. You do a  very simple demo, and you run a query. Boy,  

isn't it cool? You can get an answer.  Just ask a question in English.  Try that with your real data and the  complexity of the data environment that you   have. It won't work the way you would expect. To make it work, we have to help the language   models understand our business. How can they  possibly...? If we don't understand our business,   how can they understand our business? The  only way you can do it is being explicit.  One thing I want to ask you here, and  then I'm going to pass it back to Mike,   is that you have very clear, explicit expectations  of the technology and where we're going. However,  

one of the things that I see when I look  at just the current landscape is the way   we structure our teams if we get these  technologies to work the way they should.  This is past the hallucinations. This is the  past the semantic models actually working.   Once that works, why do I need the current  team structures that I have? How do you think   of reskilling and upskilling your organization? Well, I think that team structures will change.   They always change as technology advances.  Technology advances change the way teams work. 

If you just look at data teams, in general,  if you look at a modern data team today,   it looks a lot different than a modern data team  did ten years ago. A data team ten years ago, if   they were really sophisticated, they were working  with Tableau, and they had one data warehouse.  Yeah. But the other team  

over here was working with Excel, and this team  over here was doing this other different thing.  Today, they have these things all  centralized. That required different   ways of structuring and thinking about teams. One of the areas of conversation, there's  

been an organizational structure described  called data mesh. People have heard of that.  Data mesh is essentially a way of organizing  teams and thinking of data as a product within   that team and then providing that to other groups  within the organization. Totally appropriate. It's   exactly the way to think about things for large  organizations that have multiple data teams. 

My point is that as technology changes,  the organizational structure that you put   in place to support that has to evolve. It  will be different in different companies.  Having talked to many companies, some are  very centralized with very centralized IT   and everything goes through that. Some are super  decentralized and they have to think differently.  I would never encourage anyone to make a  giant change in their culture unless they   feel there's a strong business reason to do  it. But you need to have adjustments in the   way you work with your teams within your existing  culture (whether it's centralized, decentralized,   whatever its element is) to allow you to  run your business as the technology changes.  Large language models will have an impact.  For sure, they will have an impact on that.  Exactly how, I don't know. Honestly, I'm not sure. It'll change. It'll be very dependent on  

how the technology evolves. I'm sure we have not seen the   coolest applications of this. There will be new  things coming that people will be wowed by that   will change the way business works that we've  not even seen yet, and so we have to see how the   technology gets applied. But it will change. Bob, what you're describing, of course,   makes perfect sense. But in a way, it's a kind  of science fiction for many organizations today.  

And so, going back to a question that Qu  asked earlier, how can organizations adopt   this? Then maybe that's a great lead into  Isaac Azimov and literally science fiction.  Everybody should adopt the modern data stack,  some incarnation of the modern data stack. You   should be looking at your data sources,  whether they're internal applications or   whether they're third-party SaaS applications,  or they're the apps you purchased and you run. 

Whatever you're doing (on-premises or cloud),  take that data. Put in place one of the modern   data stack providers like Snowflake. Adopt  that as a centralized repository for your data.   Buy tools that will allow you to move the data out  of your different applications, your operational   apps, into the centralized repository.  And begin to work with it. Get the data   together. Begin to allow people to analyze it. Begin to become data-driven. This is the first   major element for a company is to adopt a  data-driven mentality where they begin to   use their data sources as the way to answer  questions. This is the biggest cultural  

transformation that I think many companies have  to go through is to begin to trust the data.  What I have found is that valid data that people  believe is the fastest way to close a discussion   and to make a decision. In an organization,  decision-making, you're trying to make the   best decision you can in a timely manner is  critical for every organization. By far the   best way to do that is to do it data-driven. What I have found is that I think my intuition   is pretty good and it is often very  wrong. What I intuitively believe is   often disproven by the data. And I have learned  to follow the data and not follow my intuition. 

I think, if you look at really successful  companies, if you look at a company like Google   or Amazon, these are data-driven companies. Build off the data, literally, right? I look   at your resume, your track record. You've been a  part of some of the most influential and impactful   technologies to touch base in the technological  landscape the last three, four decades.  The question that I have is, in the last decade,  you did the unthinkable. You could have just rode   off into the sunset and chilled, been a VC,  and just literally been the iconic force that   you are. But instead, you hopped into the saddle  and became an entrepreneur. There are a lot of  

learnings there that I'm sure you've had. You weren't just any entrepreneur. You   actually took your company public.  And you did it in less than ten years,   which is already pretty astronomical in itself. The question I have for you is, how do you think   about the team and the structures within the  org? Obviously, the thoughts that you have here   (and if it's the data you're following), I want  to know more about just how you led meetings,   how you organized teams, things of that nature.  What are the quick takeaways in that department?  I actually have a very, very short toolbox of  management techniques. It's not enormous. I   don't have 100 different things that I pull  out of my toolbox. There are only a few. 

My absolute favorite technique is the regular  meeting that you have with an organization to   drive an outcome. The cadence of that, if  you were the leader of the organization,   the cadence of that is often a week, that you do a  weekly meeting with your team to drive an outcome.  Let me give you an example of one of the first  times I think we did this incredibly successfully,   and that was at Microsoft. It was not the  first time, but it was a really exemplary time. 

When I was running the Windows Server team and  VMware was exploding in the marketplace, this   was back in the early 2000s. This was the time  when virtualization was just taking over in the   IT industry. It was a huge threat to us in some  senses, VMware was, and we had our own product,   HyperV, which we were competing with them on. I ran a process because this was a massive  

change to the Windows Server business. We  were literally moving from physical licensing   to virtual licensing. It's a big change. It was a huge business. It was a $5 billion   business even back then, so you don't take these  things lightly. Mistakes can be very, very costly.  We ran a process over many weeks where  I had about a dozen people across the   company meeting. Strategists, financial  people, product people, marketing people,   sales people, people from every part of the  organization thinking about the problem.  We talked about how we would restructure things.  Over time, we came to a new licensing model that  

ultimately turned out to be a real win-win. I look back on that. While we were very   competitive with VMware back then, you look  back, and what happened? Both companies won.  What a great outcome. VMware won  and Microsoft won. Actually, all   three won because the customer won, too. It was a  win-win-win, and that's always what you look for,   especially when you think about partners. I always think about partnerships as tactical.  

Every partnership is tactical. There's  no such thing as a strategic partnership.   There are just long-term tactical partnerships. To me, you're always seeking that win-win with   your partner. If you can't find that, the that  partnership is going to dissolve. I look back,   and the process of doing that is always  a process I find where it requires the   thoughts of multiple people working together. I am so much smarter when my brain is combined   with the thoughts and ideas of so many  other people. And those all come together  

to create a stew of the best possible ideas. Mike Prest, who is CIO of a private equity   investment group, asks on LinkedIn. He says,  "Customer-centric transparent data collection   policies can help industries self-regulate. Your  thoughts on the issue of companies having opaque   data collection policies and the fact that  consumers are less likely to trust and use   these companies?" The issue of data and trust. Well, it's a huge issue. Right? Companies will  

have their own policies associated with it, and  different businesses require different things.  Companies that are advertising-based, for example,  as their primary revenue model. Their primary   customers are advertisers, not their consumers. What I think of is I believe in transparency on   these things as much as possible. When we created  Snowflake, we did this in a very transparent way,   the way we worked with data, because  that's what customers were doing;   they were trusting us with their data. I am a big fan of transparency. I know   how important it is. Yet I recognize there  are businesses where there probably won't  

be transparency. Mostly those are  businesses where the interest of the   business is not aligned with myself as a consumer. Arsalan Khan comes back, and he says, "An image is   essentially data. If we create an image through  text-to-image, basically the bottom line is   who owns the work product: the AI or  the artist? How do we manage that mess?"  These are derivative works, and  so you're creating new things.  I think what's going to happen is creators  (authors, publishers, people who create something,   artists, whatever), they're going to decide  what they want AI to do with their work,   whether they want AI to be able to work on it. If  they don't want to, I think they'll be, "No. Don't   tread on me," and the AI will avoid those people. My guess is going to be that that is a temporary  

thing and that everybody is going to want  the AI to understand your work product.  If OpenAI, Google, or anyone wants to read The  Datapreneurs, please read The Datapreneurs with   your next large language model and make sure it  knows. I want it to be in there, and that's a   decision that I, as an author, makes. I think it  will be a decision made by the creator, basically.  You refer to Isaac Asimov as a prophet. Why? Over 60 years ago, he foresaw all of this in   amazing ways, really; in truly amazing ways. He  wrote his first robot novels and defined what's  

known as the Asimov's 3 Laws of Robotics  in 1942. Okay? That was before the digital   computer was invented. Just think about that. Asimov was thinking about a world where people   would live with intelligent machines. And the big  thing he did is he didn't treat these robots as   Frankensteinian, creations that shouldn't have  been made. He thought about them as machines   that were created by humans to serve humans. What are we doing now? We're doing exactly that. 

Remarkably, if you look at his books, a lot of  what he was writing about was happening when? The   2030s and 2040s. Guess when it's actually going  to happen. His timing is almost exactly right on.  I believe, by 2040, we will have humanoid  robots that will live amongst us and perform   tasks working for us. Helping to take care of  elderly people. Helping us in our household.   Helping us in a whole bunch of ways. This is coming, and Asimov foresaw it.   And in order for it to work, we need,  effectively, the laws of robotics. 

To give you an idea about how far ahead he was,  later in his career, he augmented the laws of   robotics. The original laws were must not harm a  human or allow a human to come to harm. It must   accept orders, except when those disobey the  first law. Then it can preserve its existence.  Later on, he created the zeroth law, which  is that a robot may not harm humanity   or, through inaction, allow humanity to come  to harm. Now think about how prescient that   is. Thinking about that so far ahead of time  and look at where we are and what we're doing.  He was thinking about that  then. That's a prophet for you. 

What is your least favorite  word in the realm of technology?  I have learned, over time, that technology is  going to make almost everything possible – over   time. It's just a question of getting the  timing right and realizing what's possible.  Impossible. I would say the word I like the  least is "impossible" because things that   we thought were impossible now are possible. I  think that that's going to continue to be true.  Apparently, you would not be a great default  to know CIO. You don't believe in that. 

I believe in yes. I believe in  making things happen. I believe   in solving problems. I'm a big fan of that. This is from Lisbeth Shaw. "How can we avoid   an AI-mediated dystopian future?" Our values. By making sure that  

we think about the values we create. The machines will reflect our values.   Because people will create these machines, we  will have machines that reflect every value,   everyone, that people have. That means the good, the bad,   and the really awful. We're going to see some  really awful things too, but we have to manage  

that just like we manage everything else. I've lived my entire life under the nuclear   umbrella. I grew up in the era where I ducked  and covered in the 1960s when it was very real.  Here we are. We've been able to survive  as humanity. We can survive AI too.  Skynet. There's no... Skynet only happens  if we want it, if we make it happen.  Again from Arsalan Khan. He's really on  a roll. He says, "In an organization,  

which is more important: the data or the people?" Always the people. The people are everything   always. The data is what comes from  people. People create everything.  Are you going to start another company? I help people start companies. I help people  

build companies. That's what I'm doing now. But would you do it again? I feel like   you've still got something in  you (from this conversation).  For a bunch of personal reasons, it's probably  not the right thing for me to do. One of my  

challenges is that I don't know how to do  this anything less than 100 hours a week.   I wish I could be good at that. I'm not. I'm not. I have to recognize my limitations on that. And   so, now, by working through others, I can keep  my time reasonable and help a lot of other   people be successful, which is great. That's  what I love to do: help people be successful.  Okay. With that, I'm afraid this very fast  conversation is out of time. A huge thank   you to Bob Muglia. He is an industry legend  and the author of The Datapreneurs. And to my   excellent co-host, QuHarrison Terry. Qu, it  was a pretty fast and furious conversation. 

It was. Bob put it in fifth gear pretty  fast, and we stayed there the whole time.  Everybody, thank you for watching. Now,  before you go, please subscribe to our   newsletter. Hit the subscribe button on our  website. Subscribe to our YouTube channel.  

Check out CXOTalk.com. We have incredible shows  coming up and special surprises. Check it out.  We'll see you soon. Have a good  one, everybody. Take care. Bye-bye.

2023-09-27 02:54

Show Video

Other news