Today’s guest speaker solves problems through a research oriented mindset. Join Armen Kerhlopain and I, as we discuss why Virgin Racing believes in a collaborative world, how the data science bowl on Booz Allen Hamilton has promoted an innovation culture and why hackathons enabled business use cases for the modern workplace.
This is HumAIn.
Welcome to HumAIn. My name is David Yakobovitch and I will be your host throughout this series. Together we will explore AI through fireside conversations with industry experts from business executives and AI researchers to leaders who advanced AI for all. HumAIn is the channel to release new AI products, to learn about industry trends and to bridge the gap between humans and machines in the fourth industrial revolution. If you liked this episode, remember to subscribe and leave a review.
Hey humans, I’m pleased today to bring to you to HumAIn podcast Dr. Armen Kerhlopain. He serves today as the chief science officer for Genpact, which is a New York stock exchange listed with over 90,000 employees globally. They are fortune companies that focus on gaining value from data, data science, analytics, machine learning, AI, and of course digital transformation.
Dr. Armen Kerhlopain, Thanks for being with us.
Pleasure to be on the line, David.
Awesome. We had the opportunity to meet earlier in the summer when you were working on a really cool hackathon for AI and Social good. I would love to hear what were some of the results of that hackathon that we met up with in person?
Absolutely, David. It was quite an exciting event, focusing a set of folks around AI for good. So we had over a dozen cities participate. Nearly a thousand participants and the project outputs were just fantastic. Everything from using computer vision, teaching computers to see, to assess urban greenery, the project being to help understand the experience of citizens to that of making governments more transparent using natural language processing, teaching computers to read. So absolutely fantastic, a fantastic meeting there.
That’s super important, especially with making governments transparent. I had the opportunity to sit in on a couple teams and see the ideas they’re working on, particularly in New York. So I was physically present for one of the events and we were talking about recycling with new garbage cans and recycle cans. What’s best is that the raccoon doesn’t flip over a garbage can or what’s the best way we can properly dispose of waste, but we’re also talking about languages.
And languages in the city like New York has hundreds and hundreds of spoken languages, but the government only has about eight of them, where there’s translations. From English, and Spanish, and Russian, and Chinese, and Korean, and Arabic, and Hebrew, and well beyond that, and maybe a couple others, there isn’t a lot of accessibility there. So it was so cool to see all those ideas emerge and think AI for social good is a really emerging space. That’s not the only project that your team works on with an AI for social good, what are some of the other initiatives you guys got going on?
We do work with Sir Richard Branson’s electric race car team, Envision Virgin Racing. And the theme for the AI for good events, and the challenge that Richard put forward was around sustainability. And for sustainability to work it hits on the item that you just highlighted, David, around accessibility. So whether it’s information around languages, and from a technical point of view, it’s absolutely superb that we have all these open source libraries, something we very much focused on Genpact bringing that to bear in addition with our native R and D units. So it’s quite the exciting portfolio things.
That’s super cool. And ,of course, open source is continuing to eat up the world in data science and technology. I think in the last 10 years the software industry has quickly evolved into infrastructure as a service, where you have a lot of these tools and now they can work with Terraform or Ansible or other different infrastructure tools, which have now evolved into replicated other automation techniques.
In data science we’re seeing a lot of that as well. A lot of packages that are open source are growing up in Python, and Julia, and AR, and we’re seeing how TensorFlow 2.0 has recently been released and came out of the beta. What are some of the open source tools or packages that your teams are using?
So a number of the ones you referenced, we do a lot of work in Python and we find a social-psych learn, Numpy and Simpy, the whole suite there. Also importantly data visualization, for example, Bouquet or even the Stalwart in Matplotlib. It’s quite intriguing, David, as you highlighted. The major cloud platform, whether we consider Google cloud platform, Amazon web services, Microsoft Azure, and other players. It’s increasingly clear that the business model is around compute and storage, and what that leaves us is a gap in domain specific applications. So although there’ll be a lot of raw methods that are available, whether it’s around model building, data processing pipelines, to actually make a difference, you need to cross that last mile.
Which is bringing these formidable methods to a specific domain, whether that’s an insurance claim, whether that’s in drug safety or whether that’s in the supply chain. And for me, that’s really the most exciting area to work in taking all these waves of innovation and transforming industries.
So, why do you think that business today is so focused on compute and storage? I wanted to share, and I want to hear your thoughts about that. So I know this year, Amazon web service, one of the big cloud platforms you mentioned, they announced all their revenue numbers, and they said: In 2019 less than 1% of our revenue comes from AI and data science algorithms.
So this software-like recognition for facial performance, and even Amazon transcribe, to go from voice to text and back and forth, is less than 1%. And they said: Actually over 80% of our revenue is compute and storage. So why do you think that is?
I would say the two are actually highly linked and the revenue number I would like to hear is around how much revenue there’s pull through the AI led capabilities and how much corresponding storage and compute is used to support that. Let’s take an example here, by large we think of calling an Uber or a Lyft as a digital process. And in actuality, it’s minimally digital. There’s the matching algorithm, which is key. So specifically, which drivers should go to get which passenger and what’s the route, but 99% of the job is manual. It’s someone driving and someone stepping into a car. So with that, it becomes a key question for AI applications at what points can interventions be so formidable, so transformative, that they, in some ways, become the process?
That’s really fascinating. What we think about with Lyft, with Level 5 Labs, and Uber with their self-driving initiatives, with Carnegie Mellon and Nelsa out in Phoenix, that is a process we’re going to move towards.
But here we are in 2019 and that’s still beginning. Maybe some predictions have said we might be there by 2025, maybe sooner, maybe later. That’s still to be determined. But we even see competitors in that space like Neuro, which is a startup in California that early has self-driving grocery delivery. Where you’ve started to see in the summer in the Brooklyn Navy yard, there’s been a couple of startups having self-driving shuttle buses to bring startup employees between different buildings as well. But those have all been in very controlled environments,very focused lines where there’s not much deviation or risk to have noise.
It was interesting, one of the professors I work with on a daily basis, we had this conversation the other day, we were going through the history of AI and the evolution back from like the 1950s with McCarthy and all of that. And we said: People were really bullish and self-driving back in the 1970s and 1980s. In fact, researchers stick their entire career on it. They were at the Stanford labs and they were out in California and in the middle of nowhere, they were drawing paint on lines on these roads and trying to train these cars to self-drive.
We saw there was a big delay in that. The delays, now we fast forward 35 years and we think we’re on the precipice of self-driving and sometimes it shows that you need modern breakthroughs to help that be possible. And that breakthrough is from a data science perspective, having these more affordable compute and storage options, as well as now these algorithms that have grown up in systems that can, almost instantaneously, produce results.
So, what’s super fascinating, and I’m diving very much into the self-driving phenomenon, because well, that’s what a lot of the news talks about nowadays. But to make that relevant for what you guys are working on, a lot of your product and research is in the medical industry and the bio industry. And you mentioned insurance claims and drug discovery, any new trends or new projects or opportunities you’re seeing that you can talk about in those spaces?
Certainly, David. If we consider autonomous vehicles, as an example, there’s parallels both in terms of how an organization should ratchet up and consider its business model with increasing levels of autonomy and it has implications across a wide number of areas. So, for example, you can map autonomous vehicle progression, the known scale one to five, commonly used to that of surgical robots. And then as you go through each segment, you have really fundamental questions around risk, reward, access whereas the surgical robot, what are the alternatives if there is no surgeon around. So these trade-offs are quite profound for autonomous vehicles. There will be 40,000 folks in fatalities or injuries nationally, and that’s a significant challenge.
Similarly, you have quite stark challenges around access to healthcare, and I believe a key item here in going through these scales is the interface and triage. Interface with humans and triage of autonomy. So let me give an example, Uber and Lyft are actually poised to use autonomous vehicles well before the mass consumer market. And that works because, as we discussed earlier, there’s the criticality of the matching algorithm, the routing algorithm, but also the triage algorithm.
So as a user, if an autonomous vehicle only works 10% of the time, that’s not really super helpful for the premium that would be committed. However, if you are a ride sharing company and you have the ability to say: Oh, 10% of my requests in a city are easy, they’re straightforward, the paths are clear, I can dispatch an autonomous vehicle, as opposed to a human. There’s autonomy ratchets or you can shift the ratios. If it’s medium you’ll send it half the time.
Similarly with surgical robots, similarly with insurance claims, similarly with drug safety, this gets to that second point around interfaces. How do you have risk adjusted the triage for work for expert operators as autonomy picks up? And as we’re seeing the actual, the highest performance systems are actually a combination of the two, whether the human is a fail safe or there is a specific instruction from the human or the more rote aspects of the operation.
So, this is so interesting when we’re thinking about surgical robots, insurance claims, drug safety. It sounds that there’s a parallel that we can make between the five levels of autonomous vehicles, because it’s not just for vehicles, but autonomy everywhere. And so I want to dive into that because for listeners on the show, not everyone may actually know what are the levels of autonomy. And in the tech world we have been talking about it quite a bit, especially with Lyft and Uber, so let’s dive into them on each level. I’d love to hear, of both of our thoughts, on where we think it is. So level zero, the beginning, this is like prior to data science and AI. This is no automation. So this means you’re just driving the car.
This means, maybe, you’re using cruise control, perhaps. Or you are using a stick shift or you’re just driving and there’s no signals, no nothing telling you that something’s going wrong other than: Oh, your gasoline is low. Or, there’s an engine failure. So, that’s still where most cars are today. Anything prior to the probably 2015 or 2016
Some items in terms of automatic braking, lane distance, getting into the higher ones by a large volume of cars. But until a few years ago, we’ve been at zero.
We’ve been at zero. And we can say that up until a few years ago, the same is in medical. Whether it’s insurance claims or drug safety, or even surgical robots. The robots will be completely controlled by the human. Or the drug safety, you’d have many analysts making sure that everything’s by the books. Or even insurance claims it’s being completely processed by a human. But we’re starting to see that shift, that transition. And so when we move from level zero, which is no automation, to level one, which is driver assistance, that’s a lot of where we’re at today. So the human is still in control, but there are certain automation capabilities like lane assist technology.
So if you’re driving at night and you like, suddenly are dozing off, the lane can take you back into the center. Or, as you mentioned, if you’re in the winter and it’s snowing or the ice and the braking system can better apply itself, or can better move so that you don’t skid, or if you get close to a car, it breaks faster. So the driver assistance, that’s really what we’ve been looking at, perhaps, in almost every car maker in the past few years.
Exactly. And tying it to different industry areas like in this to a kind of alerts. So in financial crime or anti money laundering, simple rule based alerts. And there’s so much to go beyond that.
That makes a lot of sense, and that’s probably what we’re starting to see, also, with surgical robots. A surgical robot now could have a different sensor and it can indicate where the temperature’s changing on the human or fluids need to be replaced or whatever specific sensor. So, we’re starting to see that also in medical. So level ones, in 2019, it’s really where we’re
Level two is the first big shift that we’re starting to see. And this is where companies like Tesla really got in and they said: We’re going to be at level two or above, and most vehicles can do that. So level two means you’re having the driver assistance, but it’s advanced. So it means it can control braking and predict for you before you’re going to crash into a car, or it could steer without your hands on the wheel, or it can adjust the cruise control speed based on the traffic of vehicles around you. So these are almost like little smarter features based on a lot of sensors. And level two is where most cars are beginning to get in 2018, 2019, and perhaps all new cars will start being at level two next year.
We’re seeing parallels in all at this level in the area of customer service. So this mapping, David, as you’re highlighting, is absolutely essential and we view it in just that way. When we do a collaboration with the Fortune 100 company. Sometimes our collaborations, our deals will last three, five, seven years.
So we’ve very much look to the evolution of technology and what can be done, particularly as we can supply for these market initiatives thousands of people, on a single site for a single customer, and then expand that out globally.
So customer service at this level of autonomy comes to mind, whereby instead of it just being a phone call, as it was years prior to that, for the apps being increasingly self-service, the proactive. You may recall the need, and many people still do call in their credit card saying: Hey, I am traveling to this country, don’t block my card with the geo location.
Adding alternative data, even a small item, like: Hey, you can ping my phone and get my latitude and longitude once a week for not to block my card when I’m on vacation or business travel. We have instances at this level of autonomy where a little bit of extra data goes a long way for the user experience.
It’s super interesting. I know John Meda out in the Seattle Portland area, and he talks a lot about how the customer service industry evolved into a customer experience industry. So we have this new term CX, which has been a lot of that evolution. Both of us, who we travel a lot and when I’m in different countries, there’s almost this expectation today that my credit card company knows that I’m going to travel.
So, do I, as a human, need to really call you and tell you between dates X and Y that I’m going to be here, so please don’t block my credit card. Is there some smarter AI assisted intervention to say: Oh yeah, based on David’s profile, that seems quite normal. So, we’re not going to flag that, we’re not going to put them in this situation. But, we’ll still have a human, he double check that. The human will augment, they’ll assist, they’ll be there to ensure everything is going smooth.
And it sounds like that’s where we’re moving, not just in the cars,where you could potentially let the machine take over a little bit, but also in these systems with your credit cards.You could let the machine help run the processes, but then you’ll still have humans for anti money laundering and KYC and all these other processes that you got to ensure are checked with compliance.
These cases are abundant, from supply chain to areas in finance and accounting, like order to cash, these very disciplined process focused areas that are in some ways the backbones of industry that really do matter for getting value from data and a focus on customer experience.
So I couldn’t agree more with your comments, David. And that is indeed a key part of our thesis and our go to market. Just that. Value from data as per digital transformation, as well as a focus on customer experience. We’ve gone through lengths of research with executives around the world on this topic to structuring our organization around these key competencies.
We have an initiative, for example, called Genome. Where we took a skills inventory of our 90,000 staff, and we’ve mapped out learning paths along these lines from analytics, machine learning, design thinking, agile methodology. So as a large workforce, we can be ready to serve our clients that much better while bringing to bear advanced digital technologies. So all the focal points are absolutely around data,value from data, and customer experience.
Now, I would love to take a fun tangent with you because you just brought up learning paths. And for myself learning, and development and design thinking and pedagogy, and all these topics are super fascinating.
Our lives did not intersect, but they could have, actually, been with one of your former clients. You’ve done a lot with Booz Allen Hamilton. So a few years back, I also built out a lot of their learning paths to grow their team into data science and data analytics, and it’s so amazing to see digital transformation.
Whether it’s a company like Booz Allen Hamilton, or now with Genpact, I’m sure you’ve seen the opportunity of what digital transformation looks like for all your different employees and I want to hear about how do you think with your thesis on digital transformation, to get each to every person to the future of work, or to be future skilled and prepared whether that’s through re-skilling or up-skilling from today into 2025 and beyond?
Certainly, David. And one of the initiatives we had was around the field guide to data science who wrote with a number of esteemed colleagues. It was to enable folks working with algorithms or making decisions with algorithms to have a bit of a roadmap, part of the book, actually, had an algorithmic guide.
And that ties into what we discussed at the top of the call. Regarding AI for good, we launched the Data Science Bowl. And what we see with initiatives like these is that dovetailing with open source, the pace of the evolution of these packages, software packages that are available, that an individual can say: Yes, I want to learn. Yes, there are MOOCs massively online open courses I can engage in. And it’s an unprecedented opportunity for individual learning.
As an organization we also consider collective intelligence. So this idea of upskilling individuals, but what does that mean also for a sub team or department or initiative? And there’s all sorts of fascinating patterns around how a set of individuals upskilling a certain way on a team can materially boost the team performance, If the upskilling of the individuals happens in a particular way. So examples range from launching a web application to doing an alternative risk model on a financial institution where you need folks to stretch just a little bit to disrupt the status quo and doing things.
So, I’m quite passionate about education, quite often. And what really comes down to is we’ve, years prior we talk about computer literacy is around digital literacy. So one of the things we see when we bring to bear, for example, supervised machine learning, classification based tasks, as an example in industrial processes. What we find is that the domain experts that see that output, for example, the machine learned features, which at first can be somewhat counterintuitive, actually to a domain expert could mean something profound in a customer service context that could mean: Oh, because of the way the machine is learning the patterns here, I spot a process improvement need. And so this idea, it’s not that everyone’s going to be writing algorithms, but part of digital literacy will be around.
Hey, This recommendation engine for a product is acting this way. It probably means there’s an opportunity if we launched this product, or if we redesign our processes in another way. And that skill is, in some ways, the inverse of customer experience; it’s around organizational building or the experience of a line operator or staff within a company. And so it’s quite a human high empathy activity coming up with these massive rants and automation.
That is so fascinating, because talking about having a machine do a specific task, over and over, to a human capable level, so 95% accuracy or better. Then to have humans supervise that and say: Oh, this is interesting, the recommendations said X. So are we good with that? Do we want to change that or tune that recommendation to improve? That is where we are right now. That is this evolution in this new wave of AI, where a lot of investment, a lot of opportunities, a lot of ventures have been growing, a lot of clients have been saying: This is where will it get to.
I shared earlier on our episode today about now in the Brooklyn Navy yard in New York, there’s now a startup that’s using these shuttles that can move from point A to B. And it’s this repeatable process, controlled environment, where there’s very little deviation or risk for noise. And that, in our model of getting to level five for driving, is actually known as level three. So that’s level three, which is the conditional automation. It’s where we have these tasks of automation that do require sensors and certain processes that are being tracked, where a human does not necessarily need to supervise every minute, but then the human can take over or they can adjust the model.
And what they’re learning and providing is adaptive programming for your different employees and clients; or particularly you have, as you said, a recommendation engine that you’re going to change, the human could leave it alone or they can move in to change that over time.
And, to highlight this, what does it mean for humans when different industry areas hit level five, if we were to go even before automobiles to two horses. As an example, horses served fantastically for hundreds or thousands of years. At some point it came to this need that we needed, we wanted, we desired to go faster and further, then the nature of transportation changed. Indeed there is nostalgia, something like horseback riding. Urbanization, cost pressures, all these items came where transportation was changed and we had automobiles.
What’s happening now is actually quite similar to that. The nature of transportation is changing, so that if we have level five vehicles, well, that means the human time. When a human interacts with something it’s really precious, and in this case, it will have nothing to do with driving the car, it’ll be everything to do with working, spending time with family, entertainment and learning.
And so transportation will then shift again from: Hey, you’re on a horse. It’s really slow. It’s not serving your time. Well, it’s going to change. You’re going to miss the horse because of the nostalgia. That exact same framing for automobiles will hold, and it could be that for automobiles, it could be the nostalgia that gets people back on track.
We have profound safety needs around our roadways around the world. And so what this culminates to is that when a human touches something that is a precious moment in time, and often it’s framed exactly the opposite. Hail it just match the human automation level and call it a day.
These transformations are actually about if a human makes a judgment on a documents, on a piece of data, that must be learned, that must be generalized. In the same way, If a machine has churned through a million documents transactions, then a human should take inspiration and say: what does that mean for new business opportunity, learning opportunity? So I believe really strongly organizations that view it in this way. The preciousness of human time will be the ones that succeed and organizations that view time as a commodity are missing actually the most formidable mark of the AI revolution.
Sure, that makes sense. And thinking about level five,that sounds amazing. I can’t wait for level five, especially as researchers that we both are in the industry, it does create so much opportunity, but we’re not there yet. And that begs the question what’s level four. The level three is this assisted and level five is like pure bliss, human time available.
Level five is pure bliss, human time available. Level four is exactly what you just described. So, the machine is running these documents, millions of documents with a predefined task with the opportunity to come up with recommendations. But it’s still, if you will, geo-fenced within a certain border.
So it’s not going to go off kilter and start becoming some bot or reporter that starts writing jibberish or inventing a new language or coming up with something very abstract, but it’s going to have tasks they can work on. And based on all that training and compute and storage effectively solve the entire task, whether it’s driving from destination A to B or writing a report on a stock in it’s annual report and what that means for the company next year. it can go through that entire life cycle of the task without any human interaction at all. But the human could still take back control if they want. So, whether it would be is that this geo-fence, or this task would be limited so that if you’re outside that task, you can’t perform it.
And we’re moving to level four, mostly because of privacy, fear and concern, and not really an understanding of AI and automation. And that’s like how you just mentioned Armen. What type of organization are you going to be: one that’s all about the human and giving back time in this precious resource, so we are thinking of expanding into new industries and new opportunities and new business, or you’re going to commoditize time. And the world is thinking about both of those a lot right now. And we’ve started seeing that with data, with GDPR in Europe and these other initiatives in the US.
And we’re starting to see now that even with certain AI applications like facial recognition and how much do we want the machine to do that versus not. I just, for the first time, a few weeks ago took a flight, and that flight, particularly, had me have my face scanned. My face was scanned and it said: Oh, you’re approved. No need to talk to the agent. You can board the flights.
So that in essence is a level four. And you might ask why is that level four? Well, I didn’t need to speak to the agent, but If it did accidentally not recognize me, or if I wanted to, I still had the opportunity to talk to the gate agents. So that would be level four. And car wise, we’re getting to maybe get there. I know Tesla, Ford and GM and a few others are working on it, but that’s still a tough part to achieve.
David, as you’re highlighting here, we’re at a very exciting time and It’s really around data use, data rights and also aspects related to culture. These examples are quite striking and as a society, we need to be very thoughtful in terms of how we want our data use. The example you just gave around transportation, specifically flights, there’s throughput implications. And then there’s, as you’re leading to, there’s a risk around identity.
A lot of the most striking AI use cases are around things that are quite foundational like that of a sorting of work. So let’s take our email as an example, we’ve had email filters for quite some time that were working better over the years, and we have things like priority inbox or focused inbox, depending on what email application you use.
Now, most of industry is actually not geared up that way. There’s whole triage steps of which insurance claims should be processed and what order. There’s timestamps for financial transactions. Items associated with potential planning, challenges and supply chains. And what’s quite striking here is that in many industry areas, these work items are done first in first out.
Namely a human on a team of a thousand picks up a shared work queue, which is like an email shared email inbox, does that analogy, and they’re just sort of working on it as it comes in. Then there’s a lag and when a document is touched and, even on a personal level, we know that must be odd, because when we open our email in the morning, we just don’t look at the first one in our spam folder, we sort of triaged by the subject name, who sent it.
So, we can definitely focus on these use cases around privacy. I’m most excited about the ones around getting our industrial backbones to just work better. And that involves quite mundane things like documents sorting and a triage. So it’s certainly David. This item around, personal data, government data, we’re going to see this be flushed out in the coming years. And increasingly this requires companies to engage with governments, so that industry can advance, but it’s also done by doing right by a society. And this is more aligned to the approach of humans plus algorithms working in creative ways with data, as opposed to just saying: Oh, this is the automation level. So, absolutely non-trivial challenges you bring it up, David.
That’s incredible. And whether it’s with businesses or with governments, as you’ve mentioned, to get to full automation, one of the themes you’ve been sharing over and over throughout our session today is alternative data. How important is to have access to data and large amounts of it, so that whatever system you’re building you can create the best possible system. And that requires different sources, different quantities of data. And this alternative data industry has been growing so much. I had the opportunity in the late spring to attend an alternative data conference in New York city.
I didn’t even know this conferences existed, conferences just for, would you like to buy data? Would you like to buy geolocation, GPS data? Would you like to access this for your service for your product? So you can better create automation that provides a better customer experience, so that you can better serve your clients and ultimately give us more time to get to level five. Which would be in that car sitting, drinking a mimosa or whatever is your preferred beverage, reading a book, playing Fortnite, having an Oculus Rift on, playing virtual reality while the car drives in. Taking your conference calls, anything, and your time is freed up, liberated to work on more cognitively challenging advanced tasks.
And that process, that thought is like, that’s amazing. Like there are billions of hours globally that could be freed up if we can get to level five full automation. And it’s not, if we get there, it’s when we get there, and we’re in that process. And I choose to believe in the optimism path that you take as well, that humans plus machines, humans plus algorithms, the fourth industrial revolution.
This is an exciting time where anything is possible. And we’re going to move there and want to find out also from yourself, from what you’ve started seeing and how everything is evolving for your industry and your clients. What do you see as some of the takeaways or new trends that are emerging, that people should start thinking about today, so they could be a part of this movement from level zero to level five?
David, I believe this hits on something we’re talking about learning and in many large organizations, governments, companies, nonprofits, in terms of skill need traditional analysis is based on quantitative data. What I’m highlighting here is that Bringing even a little bit of unstructured data is non-trivial, like text or images, pulling out structured information, useful information from that unstructured data. So is the geolocation geo tagging example. All sorts of value to be had around getting that data engineering. So for organizations to be quite disciplined about doing that fusion of, we spoke about a credit card example. Most people are comfortable with that trade: Hey, I use my credit card or banking app anyway. I definitely don’t want my card to be blocked or I will be stranded somewhere. But for that to be done properly, agreements need to be in place.
The terms and conditions need to be set and organizations need to execute technically, get partners that can help them execute while keeping the customer experience focused. So I believe this is one of the most exciting things we’ve seen in these open source communities, Is this working with unstructured data And the bindings with other data sets. So, one thing I’m definitely seeing across organizations is that because it’s technically non-trivial to merge in even a little bit of unstructured data to where traditional businesses are done.
That’s a major blocker for it to happen. Let alone items about launching a new product or items around business strategy. Well, folks, citizens should be engaged with society governments that represent for fair use of the data, fair trade data. If you will, in a consumer goods context, we very much care how our products, what we eat and what we drink, are sourced ethically. I believe we’re on the… But cost was something quite similar with data, and that is something significant individuals around the world can do. And that is something that organizations can do, whether they’re business to business or business to consumer working with data.
That is all super cool. And you’ve heard this here first on the HumAI podcast. We’ve talked about it over several episodes, but we are seeing the emergence of data as a service, and data science as a service, and the evolution of being able to work with analytics, and digitally transform. We’ve learned so much today, Armen such a pleasure to walk us through level zero to level five. That is a lot of fun and I’m so optimistic for the future. So thanks so much for being with us today on the HumAIn podcast.
My pleasure, David
Hey humans, thanks for listening to this episode of HumAIn. My name is David Yakobovitch, and if you’d like HumAIn, remember to click, subscribe on Apple podcasts, Spotify or Luminary. Thanks for tuning in. Join us for our next episode. New releases are every Tuesday