AI Risks & Remedies - Podcast Episode 1

Transcript:
Just teaching employees, [...0.8s] you know, obviously, they know how the process works and they know how automation works in their business, but really educating them on how AI can impact and help them in the business is the most important thing, right? Is AI a real opportunity for JD Edwards, or is it just the latest buzzword? And what should you actually be doing today to prepare your business for it?In this kickoff episode of not your Grandpa's JD Edwards, we're unpacking how AI is being integrated into JD, the risk you need to watch out for, and the steps you can take to stay ahead.Whether you're skeptical or all in this episode delivers practical insights to help you make smarter decisions in an AI powered ERP world. [...4.7s]Welcome to not your grandfather's JD Edwards. I'm your host, Nate Bushfield, and today we're joined by an AI advisor from ERP suite, Drew. Rob [...0.7s] drew, thank you for joining us today. Uh, just gonna ask you a few questions about the topic of AI risks and the remedies behind that. Um, so let's just dive in [...1.2s] also, Nate, sounds good. Glad to be here.Excited to talk with you about AI risking remedies and kind of our perspective at ERP Suites with how we kind of mitigate those risks and get you ready for AI and on your AI journey and your company.All right. So on the topic of AI and the risk behind it, what are some of the biggest risk is using [...0.5s] an AI ERP systems like JD Edwards and what mitigation strategies can businesses kind of implement?Yeah, it's a great question, Nate. So, so a lot of, a lot of strategies and a lot of things that, you know okay, she's with, you know, employing AI in the JDL, which is the biggest thing that comes to my security, right?Security compliance. You and your company already have, you know, security structure in place, you have policies in place, [...0.6s] you have a whole security making sure your data doesn't get in and out of your business, right?And the biggest thing is just understanding that, [...0.6s] um you know, [...0.4s] there's a way to mitigate that risk, and, um you know, have a good security strategy around AI as well.Um, whether your date is outside of JD Edwards or [...0.7s] using other third part, like, using other third party systems, whether it's inside JD, I was just that connection between JD.I would security as well as ERP security is very important, right. And there's, and there's other risking remedies will rather risk will get into, but that, that's probably the biggest one that comes up.Um, another one that a lot of people think about is, it's really job loss, right, so [...0.8s] in your business, right, especially with automation, right, robotic [...0.4s] automated processes is one that comes up. Um, a lot of people at the company are just worried about losing their jobs, right.Um, really what you got to think about with AI and an AI in your business is actually enhances your jobs, right. It allows you to make more strategic decisions when you're, when you're working your jobs.Um, it allows you to be poor proactive, right, more predictive when, when allowing making recommendations and solutioning, right, [...0.4s] um, and actually providing recommendations on your data right. Um, we like to think is just, you know, leveraging AI is helping you in the business. So there's a, kind of, big two [...0.7s] that come to mind initially.So, [...1.4s] yeah, just, just believe it that and, kind of, take you wherever you feel fit after this. Yeah, of course, yeah, we can dive in a little bit deeper into the job loss and automate AI automation if you want to go that route.So how does [...0.7s] AI [...0.5s] automation impact jobs and in ERP environments and even it [...0.4s] environments, and, like, what steps can businesses take to reskill employees? So get them ready for this.Yeah, so that's, that's an absolute great question. And what in the biggest thing is, is really just around, you know, AI education, AI adoption, right?Just teaching employees, [...0.5s] you know, [...0.6s] obviously, they know how the process works and they know how automation works in their business, but really educating them on how AI can impact and help them in the business is the most important thing, right?And how they can use it to leverage, um, their future jobs, right, and what they do and allow them to take more increasingly significant roles in the business. Um, that's probably the biggest thing.So just, just don't be scared of that, and just [...0.6s] start to be, you know, start to gain knowledge around AI and how we can use it in your business, and really start to understand how it can actually help you in your job or that rather that hurt you, right?That, that's probably the biggest thing with the job wash portion. So, [...1.3s] yeah, I love it. I mean yeah, we have to be training, have to be at the forefront, completely understand that.I mean, [...0.5s] gotta take it from any normal human being out there that is competing with AI. They, they should really embody that and really start training themselves to out compete, maybe, but [...0.4s] to work with it and to [...0.6s] really come together and move AI into a different category than what it is right now.Um, to circle back a little bit [...0.4s] and go into [...0.4s] mitigating AI bias, um, and going a little bit towards that direction. So how can organizations detect and reduce AI bias and their JD Edward systems to ensure fair and [...0.6s] accurate decision making?Yeah, this, this is a really good question, um, made that you bring up. It's AI bias is well out there, right?Um, the biggest reason you come of it, and if you're working with AI right now, and you haven't really established it inside of your business is working a lot with, you know, products such as software, such as, like, Chat EBT, right, so you're out there, you're trying to get it quick information, right?It's not sensitive data, that's the important thing, right? You just, kind of get [...0.5s] quick information. Sometimes it's not, you know, it's not accurate, it's not correct. Um, the data meet may be misleading, it may be, [...0.8s] um, outdated, right? Um, and, because, you know, software like that is just going out and grabbing whatever data is, is out there, right? Um, and really what we're getting back to is having reliable data, right?And I'm not talking about just having data inside of your JD Edward system, but really reliable data wherever you go out and get it right up to date data, that's clean, right? The duplicates are limited, [...0.4s] um, and, and it's accurate, right?And the thing with bias is, is you just really don't want it to influence your business because you want it, yeah, cause you want it to be [...0.4s] reliable data and that, and that's the biggest way to kind of negate that. It's just, it's just making sure that you're not just going out there, right.And another big thing too is you got to think about biases, having human inside, having a human interacting with your data, having a subject not an expert who actually knows that data that you're currently working with. Um, and it's regularly refreshed. If you're working with AI models, the models are trained well, right.So [...0.5s] that's, that's kind of the biggest thing, it will help, and, you know, when you train those miles that the data isn't as skewed. Um, [...0.6s] of course, it's, it's working with your data repository, your internal data sources, [...0.6s] right.And a lot of people would say that your subject matter expert on this, so it's great to hear from you directly from the source, um, and learn a little bit more, but to stay on that data train, there's a lot of concerns about [...0.4s] AI and data privacy, government compliance, how should organizations approach AI security in their ERP system?Yeah I mean, the biggest thing you gotta, you gotta focus on there is, is you're gonna have your policies and procedures in place. Now you really gotta take around, not round about way, but you really gotta take a step back and really look inside of your business before even implementing AI.Where's your current security right now, right? Whether using different technologies, whether using JD Edwards, is it in a good spot to even implement AI, um, into your business? Right?Um. And, you know, there's different AI tools out there, you know what we talk about and I'm trying not to get too much in the RP suites, but we talk a lot about insetted AI inside of the JD Edwards system.And that works a lot with JDL, with security, which if you're JDL, it's customer that's awesome because, you know, you work with that every day, you're familiar with it, your security teams aware of it, [...0.5s] um, and you understand how it operates.Now when you start to talk about, you know, security and governments and leveraging other tools out there, you know, a lot of cloud based tools at ERP suites, we use [...0.6s] Oracle Cloud Infrastructure, um, for AI services, but there's other cloud infrastructures out there, just such as [...0.7s] AWS, right, as your, right.Um, and as you start to use these services and start to implement them, like, how are you [...0.8s] throwing out good security policy, policies, good, I am, I am roles, good encryption, right, the biggest thing is, is moving your data from one place to another.You know, your data doesn't have to be still in the cloud percent. It can be on tram, [...0.4s] but how do you access that data, right, is in the biggest thing is it just knowing that a lot of these bigger technology, these AI services out there, whether be AWS is your OCI, [...0.7s] um, they provide an encryption, right.And as you start to implement and add these services into your business and start to connect them to your various different ERP systems, JD hours, [...0.4s] other data sources, um, just be mindful as a security team that you are [...0.6s] giving the right roles of people, right? People are accessing certain information that they can, right?I mean, I'm not trying to get too, too technical to into the week, but someone who's been working manufacturing on a digital assistant shouldn't have access to someone who's working in sales, right? There should different goals of who access who and who's using different processes in that space. So [...0.4s] that's really important.Um, of course, it's really important piece there. So, yeah, just really have no overall data government strategy, um, and compliance with whatever business you're in, especially with medical, right, compliance around that, that area. I mean, that's, that's the overarching thing that we really [...0.5s] love to do there. It's just having a really good strategy in place [...0.5s] before the internet.Yeah, of course. Um, so would you say that, that also goes with data privacy, [...0.5s] not only security but privacy and making sure that data just isn't out there in the open where anyone can possibly see it.So what are some sound steps [...0.8s] ensure data privacy? And, well, obviously security, but more the privacy side of it. [...1.5s]Yeah, the biggest thing is, I mean, we go back to the, you know, the data encryption piece, right, especially for working with sensitive data, like making sure your data is secure before you running it through any AI models, right.And, and it goes back to bias as well. I mean you know, as you start to train your data, and it all comes back at data is gold, you want good historical data to train your models, cause you wanna get correct results in the end, right?Um you know, we talked about bias before, and that's the inability to get, you know, correct results, um, if you train your data, right, cause the data is skewed or not match what you want. So I think the biggest thing here is just, [...0.5s] you know, being able to, [...0.6s] uh, have reliable data, right.And, and it's a forefront of it, like, you have to focus on that being important, right, [...0.4s] um you know, obviously security is important, but, you know, not having duplicate sources, duplicate, uh, records, um you know, in your data, um, not having missing data, right, um you know, and definitely just working on that, you have good historical data to feed the models.Um, and, yeah I mean, as far as data privacy, it's again having those good, and then encryption connections, right, [...0.5s] um, and, and understanding again, as, like, I, kind of, talked about before. It's just that you have certain roles in about role set up, whether it be JD Edwards or see I role set up.So you're not, [...0.4s] you know, you're not going to someone else's, um you know, what date is not moving from one place to another, where and shouldn't be going where it is, right?Everybody who's working in a company is working in their space, so, [...1.3s] yeah I mean, seriously, thank you again for those insights.I mean, I know that that's one of the main things that I worry about is data security and privacy, and making sure that everything stays where it needs to stay, and we're not leaking any sensitive information or even personal information.Um, obviously, that's something that everybody really values, but it's definitely something that I hold very close to my heart. So again, thank you for answering that one, um, to kind of ship back a little bit to [...0.4s] talking about job loss maybe, but more of the human intelligence side of it.Yeah, absolutely. So, [...0.4s] so is AI changing the way humans work in their day to day jobs, and or have they become too reliant on AI solutions? I know myself, I, I utilize a lot of chat EPTs and different [...1.0s] AI solutions out there right now. Um, so is that something that you've seen or [...0.6s] what can you tell me about that?Yeah I mean, absolutely, it's it's, it's really just, yeah, when you go on the job loss, right, that's, that's definitely a fear for people losing their job due to automation.AI is definitely more an automation, but, [...0.4s] and I've mentioned this before, like, AI being the ability to enhance your job, but from your lens, what you're talking about here is [...0.4s] the loss of human intelligence.And, and honestly, like, you've even talked about using other platforms such as chatty BTI, [...0.4s] I use it daily, just answer simple questions, you know, it's good to help build a presentation to, like, if you really think about, it can start to do so many things for you. Is it gonna be perfect?No, and that's when the bias came in before, right, like, it's not gonna give you all the information. You're gonna have to use your knowledge to actually build it out. But, [...0.5s] yeah, let's say, like yeah I mean, you're using those things, and that becomes your daily work, you reliant on them, right?You don't wanna be relied on that stuff. We don't wanna lose our human intelligence. Why we got, [...0.5s] you know, we're working these jobs in the first place, making those strategic business decisions for your company.Um you know yeah, and again, you can use it to build, you know, a Powerpoint for if you really wanted or, or like you said, answering a simple question. A lot of developers out there really go out there and say, hey, can you fix this syntax for me?Well, what it might do when it fixes a syntax is it might fix that one error that you asked it to, but it might create 50 others. Um, so just we're not relying fully on AI is what I'm getting at, right?Like um, for example, like, if you're gonna build a, a simple, you know, document understanding or reading contracts, like, that's fine.It can read for you the miles good enough to do that. Um, but if you're gonna start, like, asking questions about, you know, just various aspects of that contract, um you know, and it can do that for you, but what I'm saying is just don't rely on, on it to have those conversations or, you know, [...0.6s] about it.So, like yeah I mean, that's definitely the biggest thing is you wanna grow with AI is what I'm getting at, like, loss of human intelligence. You wanna grow and you wanna learn about it.We don't wanna be left behind and have AI rely to make all our decisions, right, and that's where, that's where the training comes in the play, right.I mean, we talked too early about data and security, and heaven training around that, and how to, you know, [...0.4s] be prepared for implementing AI, um, by enhancing your data, current data infrastructure and security, but also you gotta enhance your employees.You gotta, you gotta teach them and educate them [...0.5s] on, you know, just not even just simple AI 1, AI 1.0. 1, but how to use it, how to use it effectively. You know, you're not just gonna implement AI, just implement AI in your business. There has to be good, [...0.5s] um you know, efficiency to gain oil I to gain. So there's, there's reasons behind that as well.So [...0.8s] of course, um, but what you say that there are anyone, like, any companies that you've worked with, any people that, you know, that have maybe become a little bit too relying on AI solutions. And [...0.6s] how would they maybe, [...0.7s] I don't know, change that about themselves?Like yeah, obviously you said training or something like that, but, like yeah, is there, [...0.6s] is there a little bit, any more specific way of, like, training their brain to [...0.7s] not use AI as much, or is that a good thing to use AI a lot? [...1.2s]Yeah, and that's, honestly, that's, that's a really good question, Nate. And, and I understand where you're coming from. It's it's it's it's, it's getting to the point where I think as we grow, and as AI starts to grow and become more a part of our lives [...0.7s] is, is not just training the employees on what the systems are, where they, where the technologies were using, how do we use them, what use cases can we attach them to, [...0.6s] but really start to train, um, and start to learn how to, yeah, as you said, like, how to make it how to, how to make it applicable, applicable on your job and how to use it effectively, right.And it gets back to the point of, like, we just don't wanna use AI, just to use AI, we don't wanna just implement AI, no business just cause we have it on a roadmap.There has to be a reason behind it, right? I mean, there's no reason to implement AI and spend all that money if you're not gonna get any proficiency and productivity gains in the end. So that's probably the most important thing, but it's, again, it's it's, it's growing with AI.And again, every [...0.7s] right now, you know, haven't really seen it with a customer using AI too much. Because it, what you really got to think about is, it's sort of a phase approach, right?You start with, kind of, mitigating automation, right, that's kind of the initial piece, and then you wanna go into predictive analytics, and then eventually autonomously RP with a machine learning, and to talk about digital systems that you're reaching out and doing things for you, right.And again, again, that's good for, like, doing remedial task. Um, a lot of times we pull out digital assistance, we talk about, you know, it being a college level intern and doing those kind of things for you. But again, it's still not gonna be small enough, especially at this point to do.There's very, um, very detailed, very [...0.7s] strategic based work. Um, so again, you can, kind of, see how some of the stuff is tying together, cause then this ties back to [...0.4s] loss of human intelligence job automation.But just kind of the remedies behind that. Like, again, it's definitely all about growing and [...0.4s] with the AI and and and, and the new, uh, the new features that come out every single day.Um yeah, [...0.7s] yeah, so you gotta look out for that. I mean, it wasn't too long ago that I was, uh, college intern. Uh, so definitely, uh, as for utilizing AI even as those types of roles and everything like that, it could really [...0.5s] save you in a lot of places.I'm sure, but also it all, I mean, it's cutting edge right now. Everybody's talking about AI, so I'm sure, um, but talking more a little bit deeper about AI, [...0.5s] uh, what are AI hallucinations? You know like, how can JD Edwards users prevent inaccurate or misleading AI generated insights?Yeah, that's, that's a great question. And, and this comes up a lot when we talk to customers. And it's really just when a mile generates a response, that's like false and misleading.So when you think about this, you're really talking about if you're looking at JD Edwards data in particular, [...0.5s] um, if flagging a transaction, let's just talk about that.That is a threat, right? And it's really not right, um you know, the AIA agent runs through information about, you know, company policies and returns information that may be inaccurate, right, um, and what this really comes around at is that do models are not trained as effectively as they need to be.And it starts just gramming data, just to grasp data. It's not correct, right, and it's not, you know, they're inaccurate, yeah, they're just, they're all inaccurate results, right, [...0.4s] um you know, a company stock may have increased when actually didn't, like, I don't know, I mean, you start looking out, and that's what you get when you, you start using some of these tools out there.And, and always comes back to the point of, like, [...0.4s] wherever your data is coming from, make sure it's reliable, make sure it's flushed out, make sure [...0.8s] you have the all correct, all the correct information in there, because [...0.4s] your models only as good as the data it's trained on.And that's the biggest thing here. So to prevent, who is your nations to prevent incorrect or misleading data, especially to your business users or maybe working with the digital assistant, right, and asking various questions, you don't want them giving incorrect information from that digital assistant and passing it on to a customer. That's the biggest thing is, is to be very preventive about it.Be proactive about [...0.6s] when you've done out the model, not trying again, not trying to get too technical, but [...0.7s] make sure it's the right model that's training with the correct data.That's the biggest thing, right? And then we're gonna prevent those solutionations as misinformed data because in the end you're using your rose reliable data sources and house.It could be third party but [...0.5s] talking just whatever data source you use, it's it's, it's gone through by you, it's it's it's, it's good reliable data that you can use, and it's not [...0.8s] on that chatty BT out there.It's not just out there, right? And it's not just grabbing data that may be old, right or incorrect and trying to throw it together to get you a good, good response, [...0.5s] right.So we just got to be mindful of that [...0.5s] and just again, it just hoping on using reliable data sources, wherever this may come from up to date and in train data sources.Yeah I mean yeah, you're completely correct about that thing. You never really know what's happening behind the walls of a lot of these different chat bots, and, yeah, different, yeah, different AI solutions that are out there.So, I mean, it's definitely reliable if you know where the data source is actually coming from. So thank you for those insight seriously, and it definitely comes back real quick just to add one more point.It definitely comes back to, you know, just again, having those subject matter experts, you know, working with these models, right.They understand the data better than everyone, having them in the loop when you are training these models and understanding what you're training. They will understand the data more in any developer, data scientist or anyone putting together these models.So having their feedback and understanding what data [...0.5s] should be displaying when you, when you, you know, you run a model, and you get a report back, or you're talking with the digital system, and you're getting insights back, making sure that stay is, is very correct.So it's a lot of just overlaying fact checking, right, [...0.8s] yeah you know, a human could be a part of it the beginning, and then again, as is the model grows, [...0.5s] you know, stronger, more complete correct, right, gets more confident, cause that's what it's gonna be a lot of training.Then you can start to weed off that, that human that's been working with it every day, and they can start making decisions on soon. So talk again about how, you know, jobs can be influenced, um, with AI rather than, you know, [...1.7s] her by them.So, yeah yeah, well, incredible, I mean um, so past this, obviously people will be able to see this online and everything, but is there another event or something like that that [...0.5s] your piece weeks is putting on that maybe [...0.8s] someone that is trying to learn about AI solutions or maybe dive in a little bit deeper into what AI and JD Edwards kind of looks like?Is there something out there that, oh, yeah, we can sign up for, register for something like that?Yeah, absolutely. Yeah. And there's a, there's an AI, um, AI Week event, um, put on by European Suites. Um yeah, so what's great about that is, is anybody can really join, whether it be the sea level, the directors, right, the business alas, the heavy core [...0.4s] developers, right, that work with JD Edwards every day. It's for really everyone. Um.There's this few different tracks along that. So right when we talked about today was kind of one of the tracks of getting started learning about the AI risk and remedies. But then you can also talk about, you know, different AI 1.0. 1 components if you really just starting with AI, because what you really want to think about is people start in different areas, right. Some might start looking at AI, might start working with technologies, but they don't know where to apply it.And then others might be starting from ground zero, right, so they need to know that AI 1. 0. 1 as well as building out use cases of another example of just kind of that getting started track. Um, but there's other ones that are great.There's hands on workshops to work with various tools and technology. I, kind of, how it done before, but just understanding contracts and feeding contracts to an [...0.5s] AI document understandings one of them, [...0.5s] right.There's other ones that use generate of AI to read, you know, compliance documents or [...0.6s] CSR for trade training, CSR reading those documents, um, as well as other functional tracks, right, like where can you implement I in your manufacturing, in your inventory and your [...0.4s] sales finance, when can you start to implement those?Um, so that's, that's also another track, and then there's a few other [...0.4s] keynote speakers are very important from Oracle height dot Io is not a point of ours, and as well as the ERP suite.So you'll hear from a lot of different people as well, and, kind of, various solutions and products are going out. But really again, how to get started, how to, how to really start doing hand, say I and using your business.So [...0.4s] a great event, by the way, didn't mention the game. It's free, right? So anybody can join online virtual free event, so just wanted to, yeah, wanted to put that out there for you guys as well as I think that's a great place to start, yeah, of course, [...0.5s] on your AI journey.It will help you, [...0.4s] yeah, I'll make sure to add the link below for registrations and everything like that, but I also heard that you had Paul Howe Cooper there, um, incredible mind, can't wait to learn more from him, and especially with that stuff.I mean, [...0.5s] he, he knows this business, he knows JD Edward, so very excited to hear what he has to say during his keynote speaking.I mean, [...0.7s] he's definitely a great, uh, definitely a great man obviously, but he's very knowledgeable when it comes to everything, when it like JD Edwards, Oracle and AI, so very excited about that.Um, but thank you for joining us [...1.0s] seriously, really does, [...0.5s] really do appreciate it, having a close friend on the first podcast, which is incredible.And please, please like, subscribe whatever you want to do more content coming soon. Thank you absolutely, yep, thanks for having me, Nate, you can put up 30 this weekend. [...5.7s]