The AI and Robotic Process Automation revolutions are in full swing with record growth in both sectors as well as an explosion of new startups in this space. Not to be left behind, existing vendors are rushing to heed the call of AI and automation by sprucing up there existing product suites with conversational interfaces and smart AI-driven assistants. This session will cover the impact of these emerging technologies on the IAM product space and what can be expected in the near future.
Can everybody hear me? Is it broadcasting?
It is. Okay. Great. Well, thanks everybody. I'm Patrick Parker. I'm the CEO and one of the co-founders of empower ID. Happy to be here again, see many familiar faces after many years. So I'm here to talk about how artificial intelligence and technologies like robotic process automation, chat bots, or change identity and access management over the next three to five years. Change the products, change, how organizations use it change even what you think the realm of identity and in access management is about. So it's exciting times we're in the midst of a paradigm shift. And one, one example. So how many of you saw Google's demonstration of their new duplex technology? Okay. Some of you, quite a few of you. That's awesome. So it's kind of a mind blowing moment for me. Google's artificial intelligence, digital assistant, able to call and schedule a haircut call and schedule a reservation at a Chinese restaurant, which we know that can be difficult sometimes.
And it could handle all the nuances of a conversation and react to it. And the human on the other side was completely unaware that they were talking to a bot. So, you know, you're on the verge of a paradigm shift. When you take everything you've been building, like let's say the, the, the identity assurance, the knowing a user, the verification of identity, the multifactor. Well, and you kind of say, okay, now how does that apply? If most of the tasks in the future will be done by personal assistance and bots, everything we've been building for multifactor for assurance for user behavior analytics kind of goes out the window. If, if, if I'm gonna delegate to a bot, it's not gonna be a human it's, it's acting on my behalf. So that really, you know, you see that, okay, everything we've been building really doesn't apply.
So I'm here to talk about robots, not the, you know, this type of robot, but more the software robot. So kind the, just the history, the origin of robots, the fir the term robot and feudal medieval Europe was meant to mean forced work or hard labor. In 1920, a check writer wrote a play Ross's universal robots were for the first time we had a robot as a manufactured being, and they even had a robot uprising. So the first start of all of dystopian sci-fi novels that we all love today. So in popular society, we have the friendly robots that do work for us. Everybody loves C3 PO and R two D two. We have the bad robots that are usually the enemies that we're fighting against. So, and, and then, so what are the real robots today? What are the advances in artificial intelligence what's really going on?
So Google again, Google's really on the cutting edge of artificial intelligence. They have something called auto ML or auto machine learning. And recently they tested it to where auto machine learning actually is machine learning that writes its own machine learning. So it's, it's AI that programs AI and they, the recent results was they programmed auto ML wrote an AI to recognize images. And it was actually 4% more accurate than its human peers who are working on AI to recognize images. So we've reached the point where the AI is more better at writing AI than the human beings who write AI. Another one that just blew my mind was that scientists have been trying to solve for 120 years. How, when you cut a flatworm of Pollin, how it can regenerate its spotty parts, tons of scientific literature intends to study. So scientists at Tufts university said, we're gonna stick AI on this, let it crunch and analyze all the previous research and let it iterate through possible results. And in three days it solved the mystery. It found, I think it was what two proteins and one other molecule that actually showed them how that animal was able to regenerate body parts. So the implications for medical science are, are vast. So some of the not so good advances, you know, humans bring their biases and their bad habits with them, recent robot, where it could do machine learning, but they were still kind of treating the robot, very condescendingly, teaching it, things that weren't appropriate.
So all of these, you know, the, the artificial intelligence, a big scare with Google duplex, how do you know it's a human being? What are the ethics of having an AI operate on your behalf? How independent can it be? How much under your control does it need to be? And do you need to reveal that they're dealing with an AI and not a human being? So the ability of these robots that's led to, you know, in popular culture, a big job scare that, you know, at some point in the future, none of us will have jobs. The elite will live in a little bubble. Everybody will be outside and the, all the work will be done by robots. So, and it is hard to compete with a robot cap, Gemini study, you know, a robot can work 24 7, it can parallel process. It never needs a vacation.
It never needs a raise. And the cost of a robot is dramatically less than even offshoring. So offshoring is probably one of the areas that I have the, the biggest initial impact and actually is, is happening today in the financial services, injury industry and the insurance industry. So this is not our first scare. There have been many scares going back. You know, the Luddites is a famous one where they destroyed the looms and the presses. There was a scare in agriculture in the United States in the early 19. At that point, farm work was a major profession. There were 30 million people employed in the United States doing manual agricultural work. So with the automation and the invention of farm machinery, there was a big scare that they were all gonna lose their jobs. And actually the farm employment dropped in 1900 to 1990. It went from 30 million to 3 million. So in the, the population tripled. So, so there was about a 99% loss in farm labor, but during that same period, the average employment grew. So there wasn't, although all of these people lost their employment in farms, they actually shifted. And society took that as an advantage, more people were fed better standards of living. And one of the things that society did was that they instituted Manda, mandatory schooling. So since there weren't jobs on the farm, those, those children now would go to school to learn, to do the new ways of work.
So there's a lot of kind of robot angst out there, you know, humans saying, well, what can't a robot do? What can a robot do? And just this general angst about robots and what, and how we compete with them. So a good article by Harvard business review is thinking through. So in the future, how will work change? Is it as simple as everybody says that, you know, we're, some jobs will be completely replaced by robots, or is there a more complex story? What do the business leaders need to know? And how do they need to think about it to take advantage and transform their organizations using artificial intelligence and bots? So one of the things they said was that you have to cut through the hype too much hype, you know, technology, the very fashionable industry. There's always a lot of hype about the latest technology.
So you have to look at it very practically. And what they found is that the jobs themselves would not be replaced. Instead, if you think about a job, almost like a supply chain with different parties involved or different tasks involved to produce an end result, you just have to deconstruct the job and say, okay, which parts of this process can be most efficiently done by whom or by what? So, looking at a different part of the process, some of it will be human driven. It needs our intuition, our creativity, it needs our judgment. Some of it's just manual drudgery that does not make sense for a human. So you can O you can optimize the process by deconstructing it into its parts and outsourcing those parts to the most capable parties, whether they're inside the organization as humans or artificial intelligence or software, or whether they're outside the organization.
So the organization itself is to produce a good or a service and how it assembles that it's a capital source that assembles all the appropriate resources has a better business process to deliver a better product at a better price. Using whatever means are most efficient along the way. So the, the main thing that they found was that thinking about replacing jobs was the major barrier to leveraging this technology really have to look at a process, look at your onboarding process, your joiner process, your mover process, your lever process, and say, at which points do we really need human intervention? And which points could a, a non-human or artificial intelligence eliminate the step or perform the step for us. So you cut down the cycle time, you cut down the cost and you optimize the process.
So if you look at it, instead of it being the total work is being replaced, it's more like a relay race, where at some point in the relay, it's a human being running and doing the task. They're handing off to an artificial intelligence to do the next step, and then maybe back to a human. So it's, it's a mesh of a, a process being quicker, faster, and allowing a human being to do more, produce more output, to be in control of more processes than they could if they were involved at every step in the way. So we'll either be assisted by robots at different steps in the process. And then even the, the steps in the process that we do, we will be augmented. So we'll have more intelligence. One recent example, I heard from a person working in robotic process automation is that when the cable TV repairman shows up at your house, they hit a button and there's a robotic process, automation.
It goes and scans all the systems. It get it interrogates all of your devices in your home. It interrogates your bill. It gets all the information from all these systems. And all of a sudden on their iPad pops up a summary of all the information they could have when they walk into that house. So they know everything about your devices. They know everything about your bill, and they know everything about your problems, all in one instantaneous step. So that way the human's work is augmented. They're going to use that information in their judgment. They experience their creativity to solve your problem quicker and provide better service, but they didn't have to go do all those manual steps, interacting with all those systems. So they were augmented by the technology.
So another cap, Gemini. So if you look at a robotized process, basically, you're gonna say, this is an agent entering an order and fulfillment. So at some points in the process that human being will be interacting. There'll be entering data at other points in the process, which especially when you're interacting with multiple backend systems, many of them legacy, many of them, not an ability to easily integrate, to get data from one system into another. And most business processes require you to enter data into many different systems. So if an API is not possible, or if it's cost prohibitive, then a robot can mimic the actions of a human being. They can be much quicker. They can be less error prone, you'll ensure data consistency. So the human interest of data, the robot processes it through the systems and gets a quote. The agent provides their human reviewing. They know the market, they know the customer, they can review the quote. The robot creates some processes, the order, if they're any exceptions. So you can have rule a rules engine, or even a rules engine augmented with artificial intelligence to see if there's anything else the human needs to evaluate in the process. And then you process the order. So you're cutting down a manual process that might take hours into something that can be very, very quickly automated at lower cost.
So, one thing to think about is our jobs may depend in the future on being bought whisperers. So being able to better train a bot to do the portions of our work that we do not want to do so we can, we can work on higher value tasks. We can use our, our human skills and allow the, the bots to perform the intermediate task, to produce the results. So each worker will be able to do work quicker and be able to produce more output. Another very interesting study from Harvard business is how will this be applied? So we know that we're gonna be deconstructing jobs. We know that we're gonna be rethinking work more like a supply chain optimizing each point in the supply chain, but what are the technologies from a business perspective that are involved in this? And they broke it down into three main categories, which pretty much fits with everyone else's definition, but one was robotics and cognitive automation, which a lot of people call robotic process automation.
It's a technology where the robot is doing the work of a human being, not interacting with systems from an API perspective, but actually entering into data just like a human would be human would do taking data in one system or from an Excel file and entering it into the next system, getting the output and entering it into the next system. So it's actually very, non-technical typically programmed by business users and not really program. It's more of a training exercise. The next is cognitive insight. So it's using machine learning and artificial intelligence to crunch through large volumes of data, either data on your systems or streaming data or events coming in to identify problems, risks, anomalous behavior, to optimize who should have access to what in the organization to detect outliers things that a human being could never really analyze or ingest that amount of data to produce, but, but AI can be running and churning on it all the time.
And then the third one is the cognitive engagement, which everyone's familiar with chatbots, the chatbots, becoming the new user interface. We'll see that now. So, and, and these can all be in combination. So the machine learning or the cognitive insight can analyze and detect an event that can fire off a chat bot that interacts with the user interrogates them tries to assist them. And then based upon the chat bot, figuring out what they're trying to do, what's their intent using artificial intelligence. Then the robotic process automation can go do the work. So the three can work in parallel or in tandem together. So what is robotic process automation? How does it differ from traditional workflow? Technology is business process automation technologies. Again, they're the very blurry line which we'll look at, but the idea is that it is workflow automation, typically programmed or trained by business users, non-technical users.
And it interacts with applications not by not modifying them, but by performing the actions that a human being would do exactly what a user would do on a keyboard, the robot does for them with no human involvement. So there's no programming. And then of course you can apply artificial intelligence with us. And you end up with bots that are more mission centric. The bot has an end goal, and you didn't necessarily program exactly how it needs to get to the end goal. It can automate, it can learn. It can think to get to the end goal. So what's, there's a blurry line, which we'll see. This was from Ernston young, describing typical robotic process automation. And it's a blurry line, definitely for us in identity access management, because you see the typical process across the top, you know, the user requesting from the health desk, it going for approval, the admin team.
And then it splits off into the database team, the active directory team and the app team performing the work and the robotic process automation. The admin team is replaced by a robot, but then each of the, the database team, the ad team and the app team are replaced by robots as well. Now in our world, we, we would consider those connectors and a lot of people are using robotic process automation as a, a low tech connector. You know, it's a connector where you don't have to build a connector to a system or where you can't build a connector to a system because every system has a user interface that somebody could type into. So in some cases, a connector makes more sense because the application user interface is changing or there is a good API. And in other cases, robotic process automation is a good gap filler because it doesn't have an API and maybe it's a static legacy app.
So it's a better option. So this describes RPA versus workflow. RPA operates at the presentation layer. So again, it's interacting with the user interface. A lot of people call it, light it because it doesn't really touch systems. It doesn't modify systems at all. It's just kind of interacting the same way user would. And then business process automation or workflow integrates at the business logic layer or at the data layer. So it's actually integrating into the back end of systems, automating via API. And it's much better for systems that have an API or for bulk processes where you want to send a thousand users to provision. That might be quite a bit slower if you're doing a robotic process automation where it's typing. And the two really, I see them blending together because robotic process automation is really just an extension of workflow with the ability to interact in the user interface level.
So it's taking the technologies that we're typically used for automated QA testing, selenium and others that drive the user interface for automation testing. It's making those more user friendly and putting them on a, a workflow scheduler that can run in parallel. So I see the, the two merging together. One key factor. Every the, the, the younger people, especially, they only want to type, they don't want to click. They don't want to interact with anything. They wanna pull it up on their phone. So the, the conversational user interface is becoming the new user interface. If you can have a chat bot, if they can type, it's very quick to say, refit my password and talk to a bot and have the bot run you through and refit your password, or to request a bot bot to send you a link to single sign on into an app or even user provisioning.
Why have to log into a web app, have to go click to a page, load a form, fill out the information. If I can very quickly just film it on my, on my phone. So the, the CUI you'll see, and your user community driving you into being that more of the UI. So your team spent again, another paradigm shift, again, all of this technology to build this perfect UI framework, the perfect interactivity themeing design. And now that's kind of going out the window. You really only want to interact with UI when you wanna visualize something, when it's, you know, when you, when you have to. So the, the new, the conversational user interface is really the one that the users are eyeing and, and not the one that you spend all your time on.
So what's the right approach to start again. Harvard business study said a moonshot is not your best bet. It's best to take a low risk approach. So, you know, MD Anderson, they had a moonshot project to try to use AI to help solve rare types of cancer. They spent a lot of money. It was too ambitious to, and it didn't achieve their initial goals. Now, at the same time, they had the it department doing simple tasks, helping patients with families, scheduling hotels, helping them have a better experience, huge ROI, huge, huge cost reduction from the employees, from the human involvement and increased satisfaction and a better feeling from their, from their patient community. They got better service because of the bonds. So key takeaways develop a portfolio of things that you'd like to accomplish. And then pilot learn gradually do pilot, pilot, pilot. They actually get out there and test it, I guess is saying, is that, you know, you never, no one's ever Dr.
Hit oil by drilling through a map. So you really have to get your ideas out there and test them, test them and see if it's gonna work. See if your plan is gonna be functional and I'll skip ahead. So just some things to, yeah. 43 seconds. So where, where I see the future, I see that identity management, the overlap of identity management with artificial intelligence and robotic process automation, that that's really the sweet spot. That's where the, the, the greatest benefit to an organization. That's what you can use to drive and change, reinvent your organization, change the nature of work for the future.
Okay. Thank you, Patrick.
For a great keynote. And I think the fastest five final slides of all the keynotes, you know, you, interestingly, I had some memories of certainly two 70 terminal alienation screen scraping. Yes.
It's the same technology. Yeah.
So it's nothing new under the sun, isn't it? Nope, Nope. To speak. So that was something which came up. Maybe we have a quick look, at least there was one question or there are a couple of more, but so maybe the second one, do you think an AI can be tricked to break security? And if so, how will be secure? Can you give a short answer to that
Question? I'd say yes. I mean, we had Microsoft had Tay, which was their first chat bot and very quickly the internet trolls had TA saying the most racist, sexist things possible. So yes, AI can be tricked it. It's working on an incentive. And if the incentives aren't optimized to wait against, you know, bad, bad outcomes, then yes, you can trick it. Okay. Super.
How can we help you