Welcome to the Real Intelligence podcast episode on ChatGPT and generative AI! In this exciting episode, a panel of experts from RXA @ OneMagnify and Ready Signal will take you on a journey to explore the revolutionary impact of ChatGPT and generative AI on businesses worldwide. This game-changing technology is reshaping the way companies operate, empowering business intelligence, and democratizing data science like never before. Get ready to uncover the potential of ChatGPT and generative AI to drive revenue growth, enhance decision making, and revolutionize the field of data science.

Interview Transcript

Introductions

[Anna Schultz]: You’re listening to the Real Intelligence podcast presented by RXA, a leader in business intelligence and data science consulting services. We’re here to bring attention to the unique stories, perspectives, challenges, and success. The individuals in the data industry face at every career stage.

Welcome to the show. Thank you for tuning into the Real Intelligence podcast. I’m Anna Schultz, Marketing Manager at RXA, and joining me is the CEO and founder of RXA, Jason Harper. Today, we have an extra special episode planned to explore the revolutionary impact of ChatGPT and generative AI on businesses across the globe. This game changing technology is redefining the way companies operate, elevating business intelligence, and democratizing data science like never before. It’s not just the next big thing. It’s the thing that will shape the way businesses thrive and compete moving forward.

 So we’re thrilled to have an extraordinary lineup of experts with today who will demystify ChatGPT and generative AI, reveal how you can harness its potential to drive revenue growth, enhance decision making, and make data science more accessible than ever. Joining us are Mikayla Penna, Sales Account Executive at RXA, and our go to authority on generative AI’s role in the business landscape, and Jacob Newsted, Data Engineer at RXA, who will break down the technical wizardry behind ChatGPT. We also have the brilliant Matt Schaefer and Megan Foley from Ready Signal, who will share their insights on practical business applications for this groundbreaking technology.

What is ChatGPT?

[Anna Schultz]: So let’s jump into the extraordinary world of ChatGPT and generative AI to empower your business with the knowledge to capitalize on this transformative technology. To start us off, we know that business leaders are keen to understand how ChatGPT and generative AI can revolutionize their operations, business intelligence and data science. But first, I think it’s helpful to truly understand what is generative AI. Jacob, can you walk us through the general definition and more specifically how ChatGPT works?

[Jacob Newsted]: Of course. So I think really the easiest way to go about this is starting just at the words ChatGPT and working backwards. GPT, we can start with that. Stands for generative, pre trained transformer. Going a little bit deeper, we’re gonna have a lot of steps back is transformer. It’s a specific machine learning architecture that kind of takes a human intuition to looking at data.

Transformers have a mechanism called attention. Where when you look at, let’s say a sentence, like the cow jumped over the moon, certain things like the and cow are related to each other. And so the machine learning algorithm kind of pieces these little things together, and it processes data, not just as a collection of words, but as a collection of relations. This is really important when we kind of go together onto a higher level thing, which is the generative part.

So going back to GPT, generative pre trained, the way GPT models are specifically made is their input a series of text, and it is cut off at a certain point. And the machine learning model is then asked, okay, input the next word. And it trains itself kind of in an unsupervised way, although that some people might be shouting at me in the background, but it’s easy to think about it as

unsupervised as you just give us some text and you say complete this sentence. Word by word by word. It forms these relations using attention. And eventually, it generates a complete sentence.

Now take a step back and do that over the entire corpus of Wikipedia. Every public source book. And all sorts of text all over the entire world multilingual, Now even with GPT-4, we have images, audio, video, all sorts of things. And it generates these kinds of very general ideas of relationships between words, sometimes images, and it’s able to generate piece by piece, some sort of predictive… this is what should come next.

That’s what GPT is, GPT two, GPT three, GPT four, we’re gonna keep going and they’re gonna probably keep the same architecture with small changes, which would be a completely different topic. But then we really have to talk about the big game changer here, which is the chat part. Chat actually comes from the idea that rather than training a model, but just on predicting the next token or next word depending on what model you’re using, we score how well it outputs things.

So if you’ve ever used GPT before, you know that it can come up with very different outputs to the same prompt. Depending on, like parameters and stuff like that. It’s basically how they sample the next word, so it doesn’t always do the same thing. So they leverage that They say, okay, GPT come up with five different statements to this one prompt. And then a human says, okay. I’m gonna rank these one to five, how pleasing it is it for me the human to read. And then it updates itself based on that over and over and over. And they actually automate it. They cheat… They actually train a model to reward it for them so that they can automatically do this, but more or less. It’s a human ranking things from one to five and updating itself based on that.

So that’s where we go from just predicting the next… word or token to making it human pleasing and easy to interface with. And that’s why chat GPT is so big. Because it takes this kind of scientific approach to predicting the next word or token and gives it a sort of human touch something that’s easier for us to interact with. And that’s why this has been such a big deal is that now we have this almost human like, I hesitate to say that, but almost human like model that anybody can interact with with very little knowledge about what GPT expects. It’s what a human expects now.

[Jason Harper]: I think I think one of the things you said in there too that is just…I don’t like the secret sauce, I guess is that there’s a human and loop in this process that human ranking. I know there’s some nuance to how that’s actually occurring, but it’s that that human ranking where it’s still utilizing this human in the loop for training that has completely made this game changing. Right?

 So when you’re using it, it does it does have that feeling of, like, this is, you know, a person. I thank it. I’m just I feel compelled to say, oh, great, thank you. Like, give it praise and, like, work with it in this, like, positive, like, environment, and it rewards me back with, also like praise like, oh, that’s great. Yeah. Great idea. Oh, this would be really fun to work on. Good luck with your project Like, it’s really interesting. Like those little nuances to that that really make it feel like I’m working with someone. Yeah. I mean, maybe that is creepy and is bad, but I do feel like I’m working with someone when I’m in there playing with this tool.

[Jacob Newsted]: Yeah. And that’s like one of the big breakthroughs. There’s a couple of, like, big kind obstacles we have to go through, but just interacting with these models. I mean, it’s you. It’s anybody. Like, I’m the nerdy dude that kind of is like, oh, hey, numbers cool. But everybody else is interacting with these models. So it’s very huge that people get this kind of feedback. It makes a big difference.

[Anna Schultz]: Absolutely. Thank you, Jacob for walking us through that. Now that we have this general understanding of the program. I think it’s important that we pivot and discuss how businesses can actually apply ChatGPT today.

Optimizing ChatGPT and Generative AI for Business

[Anna Schultz]: So a question for the larger group, maybe Mikayla, Matt and Megan. Can you walk us through how you’ve seen ChatGPT be used to optimize business operations?

[Matt Schaefer]: I’ll go first. So I think that, to Jacob’s com… The way you summarize, it was fantastic, especially for someone on this call that’s not a data science practitioner. I spent my entire career in the world of data and analytics, but I think that what I see here is a fundamental game changer making kind of this world of data science much more approachable to the masses, but for those that can really harness this as a utility in its current form.

 I think can be a fundamental game changer from an efficiency standpoint and things like content creation. It’s it’s things as simple as linkedin posts, blogs, email prospecting, you know, I think that there’s a ton of application today, and it’s only in those types of rounds at least from my vantage point, that we’ve only scratched the surface. And it’s amazing. It’s a little bit terrifying, but extremely exciting on, I think what this will do and how this will reshape how you know, businesses run and operate and scale?

[Mikayla Penna]: Awesome, Matt. To add to that, for me, being a sales woman, my entire career. For me, what’s exciting and what I’ve seen a little sliver of to help me so far is just being more efficient and more productive in my role specifically. Like I’m able to give GPT a few prompts and have it spit out my notes from client calls so I can, you know, put together contracts and follow up on emails to, you know, eventually get deals across the finish line. And for me, spending less time on what I like to call my admin tasks in my role and more time building deep connections with clients, is a game changer, and I’m just excited for the future to see where this could go from there.

[Megan Foley]: Yeah. I Totally agree with you Mikayla. I totally feel that every single day and  it’s really cool being in a small company that your small team can do the work of like six to ten people now. And it’s not even just small companies that are, like, really seeing the effects. If anybody from, like the big, like Whole Foods and the Amazon are also seeing machine learning algorithms. Really just take their processes and run with it. There’s so many cool new outputs that really run on chat. And it’s really cool to see how they’re working their way into your everyday sales processes. Anything from, like a copilot that’s running on like, Microsoft suite and just helping you with Linkedin prospecting? So to really just writing Linkedin posts, doing marketing, anything from the visual aspect to the copy itself. And it’s really just making it a lot easier for the little guy compete too.

[Mikayla Penna]: Absolutely Megan. Totally agree.

[Jason Harper]: I wonder some of that note like so obviously Microsoft made a big investment into OpenAI in particular and Microsoft is incorporating it into Github and their tools, but I hadn’t really pieced it together that they they also own Linkedin. Right? So, like, are we seeing are we seeing that level of integration occurring within Linkedin Like you are. Can you talk about that? I don’t know that I’ve actually seen that yet.

[Megan Foley]: Yeah. So I would say that there Is probably about like, six or seven tools that are trying to be the next big thing. And really what it comes down to is they’re still playing around still trying to figure out who like the best one is and who’s gonna to get that Microsoft kinda tie in. But what it comes down to is, all these tools are scraping somebody’s page figuring out what they say their sales are their skills, all these different things, scraping it, putting together a blurb, and you can explain what your company does, and then what’s like your main key product takeaways, and it will scrape their page and see how you match up to make a tailored message to them for you.

[Mikayla Penna]: And to add to that, Megan, that’s something you know that I feel like is a huge game changer because spending time tailoring a small social media, like Linkedin message for each persona you’re talking to is incredibly time consuming. And I used to, you know, spend three, four hours of a block in my day to do this. And now I could do this in thirty minutes, which is amazing and then get back to the work that’s meaningful, and at the end of the day, produces revenue. So I love it.

[Jason Harper]: That’s though, that’s been that’s fantastic. I love it too from that perspective for sure. I would say one of the when… And thinking about the comment about creating the post in that. Like, I just started using Microsoft designer, which I think runs on DALL-E, DALL-E. And I don’t know what y’all experience have been with that, but I started playing with it this weekend and found it to be Length technical term is rubbish. Like, I didn’t find a lot of success in its ability to actually create something I could use. It was entertaining and like, I enjoyed the results from, like a, you know, made me giggle, but I didn’t find anything from the image perspective that was actually useful. Curious what you guys have different experiences with that?

[Matt Schaefer]: Yeah. Sorry. Not as advertised, I think is what the the way that I would sum it up. I think that there could be a ton of really interesting application for it in future versions of this, but I would agree, not as advertised underwhelming a little bit.

[Megan Foley]: Yeah. And to kinda add to Matt’s point, I’ve even expanded it outside of the Microsoft Designer to other like platforms. So like Canva has their own called Canva Magic. And it seems like this really cool feature that’s it’s like, oh wow. It’s gonna place Adobe Photoshop, all this sort of stuff and Even Adobe has their own with Firefly. And really there on definitely like the beginning cusp that I feel like if that didn’t get released when it got released, they would have been a couple years out from it. And really, I felt like they jumped on the bandwagon and threw it out here, and it really shows that humans right now kind of have that that cap space pretty well like managed in that it can’t produce the same things that I feel like people can right now. But who’s we’ll see in a couple of years? Maybe it will be completely different. It’s definitely the beginning.

Quality of Generative AI

[Jason Harper]: Yeah. Yeah. I think it’s been underwhelming and I think you’re right. They just jump the gun. I don’t know why they’re trying to push so fast when it’s just… So you can’t help but compare the quality of output that we’re getting from, like GPT-4 and think, like, how amazing this is, and then my expectations. Have been completely transformed, for what I expect from any of these. So when I get something back from them that maybe a year ago, I would have thought Wow. This is really cool. Look at how good they did. It’s like it’s… My expectations are completely different, and it’s just, yeah, rubbish, I will stand by that.

[Mikayla Penna]: Yeah. I feel like the bar has been set like super high with every iteration of GPT. And now it’s like, the race to the top to get that visual generative AI for several companies, not just open AI and I wanna see who gets there, Like, who can do it.

[Jason Harper]: Yeah. I think the the only place I’ve seen it successfully applied has been in that sort of, like, using…Stable diffusion, where you’re really training your own, and you’re really you’re feeding it quite a lot of your own bespoke image, data, if you will, then it can actually be really powerful, but it’s it’s just not starting from what feels like scratch with with DALL-E or these other these other sources.

[Jacob Newsted]: I kinda feel like a lot of these models, nothing was tailored at first. We we had a lot of stuff that was like, oh, I can generate images based on ninety billion different images on the Internet. But like, to actually put it into a product and say, hey, you can use it with our product x is a completely different expectation because I can ask for a sunny meadow view in the appalachia or something like that. And stable diffusion can come up with something really good. Maybe even DALL-E, I will admit as well that DALL-E tube is pretty rough. But I think the problem is is a lot of people just started throwing spaghetti at the wall, and it’s not tailored to anyone product yet. I think it’s gonna be a little bit, but I am really looking forward to seeing, like, how they navigate that. It’s gonna be kinda interesting.

[Matt Schaefer]: It’s it’s an awareness and an education opportunity right now. Right? I mean, because the I think about this like, who the hell thought Youtube influencer would be a thing, you know, a decade plus ago. And now you’ve got, you know, new opportunity for to be like the the maestro of prompts and like what it can be produced, that’s gonna be a pretty desirable skill set here in not so in not a very long period of time. You, I think that companies are already thinking about this both from an education and enable, but who I’m hiring for and the Job reps that we’re producing, it’s just in it’s influencing everything. I think it’s super fascinating. Right? And it’s… Those who don’t embrace are the ones that are gonna get left behind.

[Jason Harper]: So it’s funny I was having this conversation at a birthday party. On Saturday with a bunch of ten year olds and the parents. And so we were sitting around talking. And of course, since I was there, I brought this up and we were talking about GPT and AI and stuff. And

talking about that, in particular, Matt, the usefulness of the skill set of writing a prompt and how important of a skill set that is to be able to actually write really effective prompts and looking for that as an employer and things like that. And for me, like, I I’m may be wrong about this, But I feel like that’s gonna be an insanely useful skill set for, like, twelve months. And then it’s gonna actually be using GPT and GPT like solutions actually write the prompts. And so your ability to create these really well constructed prompts and the nuance that goes into that is critical, but it’s super short lived because it’s really about create. Like, it really is more about pulling what you’re looking for and stuff. But so I really this whole space. I guess, I just say that it’s a…I don’t know where this… I guess, from a skill set needed to take advantage of this, don’t I don’t have my head around exactly what the right skill set is in the long or even medium term?

AI’s Effects on the Workforce and General Public

[Mikayla Penna]: Yeah. Right now, I see it as in the short term companies hiring for roles like AI project manager or AI data science manager, people that can, you know, have the skills to prompt it and have worked on that and are ahead of the pack in that game and then eventually, from what you were saying Jason, pivoting to…Now I run the whole department within my company. That’s in charge of AI and I managed these tools that we’re using for AI. And I think that’s really exciting and also when I talk to like, my friends and my family that are outside of, like, this whole tech space we live in. They don’t even know what ChatGPT is yet. Right? They’re like, Oh, it’s just, like scary robot stuff. That’s gonna take over the world. Like, no offense mom and dad, but their their generation just doesn’t get it. And I I just think it’s really important to know, like, if you’re not a robot because I’m not yet. Right? Is you’re like, what about, like, soft skills and empathy what about all these things and I don’t think you should be scared of that?

Like, job skills such as, like, communication and empathy and having a deep knowledge of, like, who you’re talking to, now where we stand and in my opinion into the future. That’s not something that AI is going to replace. But if you have the hard skills right of understanding AI and understanding how to prompt it or even how it works like Jason dick or not… I’m sorry. Not Jason. Delete that, Anna. Like Jacob explained at the beginning of this call. I think there’ll be huge skills in the job market and I know everyone in this call that I’m on with today, like, we’re excited that we’re into it and we’re talking about it and I think when the greater public you know, finally embraces AI in general, whether that be DALL-E or ChatGPT. I think we’ll be ahead of the game. We’ll be ahead of the curve, you know, knowing about something and being excited about it and teaching yourself knowledge before the general public goes wild, You know, makes me really happy and excited for where I’m at.

[Jason Harper]: I think that’s a good point. I sometimes forget too. I get I forget that people don’t know what this is. Like, and I’m for my every now and again, like, at this birthday party. I was like, oh, you’ve never I’m trying to describe it it as the everyone knows I was like, you haven’t seen this. I’m like, oh, like, pull up my phone. I’m showing them for the a first time. This thing that I’m, like, yeah. I don’t know. I thought everybody used this. I’m confused by this conversation. Give me a moment to reset my brain to back to reality because not everybody is using this, you know, all day every day and as, you know, it’s like you know, it is. And I think too it, it’s imperfect and there’s lots of issues with it. So it makes sense that most people haven’t fully adopted this because it’s wrong a lot and you have… To like, be willing to deal with this thing that’s, you know, It sounds like it’s super smart and super right, and you can trust if for everything, but it’s you know, it’s not. It’s still so early. I sometimes forget that forget that.

[Mikayla Penna]: Exactly. And I don’t think we’ll see, like, full, like, public use of this until it it is trained to be more accurate. Right?

[Jason Harper]: Mhm.

[Megan Foley]: I feel like the biggest conversation at kids birthday parties with it is about cheating probably Right, Jason.

[Jason Harper]: That does come up. That does certainly come up. Yet. I mean, we’re seeing that I guess, and especially at the college level, like, being able to… Just everyone seems to be using it to write everything. So I don’t know how college is really gonna react to this. I feel like it is sort of like a this is a lean in opportunity to figure out how to incorporate this into everybody’s, you know, educational experience versus you know, I’m already seeing, like, places trying to put the guard walls up to, like, combat and fight the use of this technology and whatever. But I feel like that’s a fool errand, but yes, I could talk about that separately.

[Megan Foley]: The willingness to learn and the adaptability really. I feel like that’s the soft skills needed to really take it on the challenge.

[Mikayla Penna]: Yeah. And I think there’s so many, like, different businesses that can be made out of this, whether it’s a business that detects if someone wrote a paper or took a test. With AI help or, you know, hospitals using AI to make better informed decisions to go the way of human error. There’s so many things that get my wheel spinning, but I’m like, oh, my gosh. So many new businesses. Not even just in tech are gonna come out of this era.

[Jason Harper]: I feel like. Well, I think Mikayla too as you’re talking about that. What are the implications of this talking to a friend of mine last week sort of about some of the implications of some of the the visual and audio components of this and its ability to kinda like, fake things in real time. I think a corollary some of the results of this too will be sort of like, kinda what the opposite of scaling is descaling, I think is the wrong word, but, you know, bringing things back where you’re not able to do…We… We’re gonna rely more on being in person and being in real life with people and being able to… Like, it’s gonna place more value on being in person and talking to people or going to the park, or doing, like, these sort of things, like having conversations that because the amount of, you know, curated automated video content where you’re talking to things or interacting with things that aren’t necessarily actual human on the other side. It’s gonna make phone calls feel less real. Right?

So, like, you know, and sort of I’ve I’ve seen and felt that we’ve changed from, like, came having phone calls feeling very real to then having really video calls now is really my normal defaults today for a lot of conversations that I’m having and that sort of the thing. And I think there’s gonna be another change where it’s like, I just wanna be in person to actually see you and be around you and in your space, and I think same thing with colleges and testing and things like that. Like, and it you’re not… It’s gonna reduce our ability to scale and do these big broad things and actually focus on no, like, more interpersonal re reactions or relationships with teachers, with peers and things like that, which I think that feel is it may slow some things down, but be a true net benefit to help people interact with one another.

[Matt Schaefer]: Oh my gosh. I… That wouldn’t that be an amazing byproduct at this moment in time. I mean, many of us have young kill young kids. We talk about screen time all the time in my house and, like, the video games and, like, how do I…It’s like to an unhealthy level. And I mean, that’s an interesting perspective of the future that this actually drives both innovation and more human connection, then than anything else. I’d… My chips are on that one, Jason. Yeah.

[Jason Harper]: Yeah.

[Anna Schultz]: Yeah. Absolutely. Wouldn’t that be a good a good outcome of all this. And it’s funny. I think a lot of people think it’s, you know, the opposite. It’s driving all towards this AI towards this more technology, but maybe it will have the effect. So we’ve gotten a lot into the

aspects of ChatGPT helping with business operations, maybe image generation, content creation, things like that. One of the other areas I know we’ve all seen great potential for generative AI is in enhancing business intelligence and data driven decision making. So can we kind of steer the conversation and talk about how ChatGPT or generative AI in general can help in generating actionable insights from data?

[Mikayla Penna]: Sure. I can speak to just something that’s kinda driving me and keeping me up at night is how businesses will use this to be more accustomed to their customers. Again, sales girl talking here. So I’m thinking about the money always. And for me, it’s like, okay, In today’s age, say I wanna go and buy a new pair of sweatpants. Right? I’m like, okay, I’m gonna go to like, my five go to stores price is gonna be a factor, the type of pants I’m gonna look for gonna be a factor. And then I kind of, like price shop across the Internet. Right? And now I feel like if companies can incorporate this, let’s like say a big one like a target, for example. If I can go on to target and have GPT like white labeled on their site and ask it questions about what I’m trying to find, and it spits out what they sell to me right away without having to, like, go through their search and do all this. I’m gonna buy it right then instead of just putting it in my shopping cart and waiting half the time, I’ll put thousands of dollars of clothes in my shopping cart, and I’ll never buy them. It’s just like Window shopping on the Internet. And I feel like it could lead to making faster customer decisions, which, you know, lead to more revenue for businesses, whether it’s small and large, and that’s something I think about all the time on all levels of business, specifically retail,

[Anna Schultz]: Yeah. Absolutely.

Generative AI’s Effects on Data Science and Analytics

[Matt Schaefer]: I think from my perspective, like, as you think about the world of BI, data science, analytics. I mean, it still is a very kind of like…I guess, it’s a world of that’s kind of under the wrapper of super technical and I think that this, you know, kind of helps to… I hate saying demystify, but I think really makes that world much more approachable and some things that we’re thinking about like at Ready Signal is playing in this world external data, leading indicators, purpose built forecasts, being able to provide these in a way that they’re both consumable, but also interpretable becomes extremely important in our world, both through, like, trust and actual embrace of the application, I.E, the forecast that I’m gonna rely on to run my business and you know, and drive some of the next four or five decisions that I’m gonna to make today. I think that just the the way that we’re incorporating that to have that human kind of readable summarization of something that’s super tech technical again, not the practitioner on the call, but, you know, I benefit from it, my business benefits from it and my customers benefit every single day and I just really think that that’s been an incredible game changer from my perspective at least from my vantage point. I I don’t know, Jason If you share that or maybe to the to the rest of the panel here. If there’s anything that you comment or challenge in in that statement?

[Jason Harper]: No. I mean, I think, I think it’s very similar to the way that we’re we’re seeing things. I don’t…I mean, I I know Guess what it say. I look like Jacob, you’re gonna chimein? I mean do you have any thoughts on this. Go ahead, Jacob.

[Jacob Newsted]: Yeah. Sorry. Well, I I use it. In my work life right now, we recently had a project that had a large table of customer information, feedback, sentiment, all sorts of things in a structured way, but it’s like text and lots of other things. It’s hard to go through everything and summarize it all. So we used ChatGPT and shoutout to this Python library if any techies are listening, Llama Index, makes it super easy. We are able to ask questions to this table if that makes any sense, and it gives us summarizations. It gives us feedback. What what’s the overall sentiment during this time frame or during this event or something? And it just gives us human readable insights into it, which is huge because whereas where I used to do random data science, like probabilistic analysis on things, I don’t need to do that. All we do is we have an app that a person asks a question and the data answers itself back to you. Which is huge. Because now it’s not me that needs to do this. I make the app. Everybody else asks the questions, which is huge. It’s transformative, and it gives us the ability to do even more crazier things to hopefully answer everybody’s questions.

[Jason Harper]: I’ll say having been having used that specific library, all weekend and including…

Several hours of today. It is amazing. There are lots of pitfalls though to using it. And on the technical side, number one, expect to read in a folder or not a file. I spent two hours trying to figure out why it wasn’t working. And so that’s… Thank you, Tom for helping me with that this morning. Yeah. No. It it happens. But so what’s also interesting is so I was able to process in around three hundred thousand comments in a specific kind of another text customer experience I thing we’re looking at. But what’s interesting is, it’s still not. It doesn’t work on the first try. Right? So so learning, like, even training and training training samples and that. So, like, this is this is not magic. Like, it might sound and put a magic wrapper around it, but like, getting getting actual insights out of data still requires a lot of thinking and time. It’s not as simple as just throwing a dataset set over the wall and then coming back with this magical thing that you can ask questions of it. It requires really intelligent, an intentional organization of the data, looking at it, reacting, responding to it and changing it, right. And so incorporating, we’ve… We’ve had to make, you know, several iterations of changes to this. And I constantly have my little usage screen up to, like, look at my, like, oh, that was four dollars. Oh, that was eight dollars. Oh, that was oh, that was fifty bucks oops. Like, just watching, like, the training of these things, it’s very, you know, it’s interesting just it’s very complicated. So I’d say, like just… That’s what I’m the vision of what you’re doing there and what, you know, what we’re trying to do and the sum of the stuff is, like, just it’s gonna be magic. It’s gonna look like magic to the end user, but like to get there, there is… Oh, there’s a lot of smoke and mirrors and putting things in place. Actually make this thing happen. It’s actually interesting. So speaking to that and specifically Llama index since to Unfortunately, I’m sorry. You struggled with that one.

[Jacob Newsted]: Literally, the thing about that is it creates it your prompts, your questions and everything is not just gonna be tailored to GPT anymore. It’s actually tailored to your dataset set. Because I don’t know if you I don’t wanna get too jargon, if you tried the vector store, python kind of class, basically, it takes your data and it creates…I don’t know. It creates kind of themes and ideas around it and it stores them in these kind of like semantic ways. So even just asking questions of your data, one dataset is not going to have the same kind of semantic relationships between every single row as the next one. So you’re not just fighting against GPT and prompting anymore. It’s really gonna be an art of. And like you said, there’s tons of tuning. It’s lots… Of different things that are just gonna…It’s not magic. Well, we hope to make it look like Magic.

[Jason Harper]: Right. But it isn’t. Yeah. It should feel that later to the users. I I would add two to that. One thing I’ve I’ve felt myself like, so I I… Was a coder at one point in my life. I think I can claim that. There’s some people who would take massive exception to that. With Lary listening, Lary is like you’re not a coder Jason. That’s okay, Lary with one r by the way. But so at any rate, the… I think for me, like, it’s empower me to go write coding guys and, like I’m spending it ton of time, like, writing python. And so the thing is…But that’s the thing. Right. I’m not I’m not a code. Admit it. And And so, like, I run into these little issues that even ChatGPT can’t diagnose because it doesn’t know the issue. And I spent hours on this thing. And this morning, meeting with an actual code. Showing him my cody. He’s like, oh, yeah, that should be a directory not a file, like, within seconds. And I was like, oh, got it. Okay. Thank you. But so I think like, there is sort of, like, this like, fun exploratory stuff where there’s gonna be a lot of skin of the knees. And I think that we’re at a phase now where it’s just not it’s super approachable and easy to access, but to actually get the true results out of it, It’s still really hard. Like, it’s really, really hard and requires like actual skills that are developed when people over a long period of time.

[Anna Schultz]: Thank you all for those answers on that. I think it’ll be really interesting to see how that all develops and at RXA? That’s our bread and butter right? So that’s kinda what we’re we’re working on figuring out.

Automation’s Effects on the Workforce

[Anna Schultz]: One final question for everybody to kinda keep things in a different direction. As these types of generative AI become more integrated into the business world, the skills required for professionals are evolving as we touched on earlier. Do of you guys have any additional thoughts or elaborations on the importance of candidates who can effectively utilize AI tools? Maybe how it’ll impact the future of the job market. Additionally, maybe how companies can work to upskill their current employees to leverage these technologies more effectively.

[Megan Foley]: Yeah. I guess I can start. So I feel like to kinda say, like, the fear of automated work. Is not something that’s novel. It’s been around since the eighteen hundreds believe it or not. And there’s a really interesting passage that I was reading from like a BYU professor who pretty much said that this is not a new idea. And that every single year, we kinda come out with a new thing that this is gonna replace the work. Like, we’re not gonna work anymore, really when you look past, like, past the add it, it just really made everyone more efficient. And, like, we’re were talking about data, so I might as well bring a data source into this. So The world economic forum said that there was eighty-five million jobs that are going to be replaced by AI one day. But if you look at that same statistic, they say that ninety-seven million are going to be created. So work really doesn’t get replaced. It’s just the work that isn’t efficient that people don’t want to do. We didn’t we didn’t wanna to plow the fields ourselves. We built a tractor of thing. So pretty much does kind of what a lot of this has come down to, and you’re gonna get left behind if you don’t adapt. But at the same time, kind of if you adapt and you become more efficient, you’re just not gonna really I feel, like see it in the future, you’re just gonna be looking back at it one day and you’re like, hey, the iphone existed, it didn’t exist twenty years ago. To it does today I can’t imagine my life without it. That’s really where I feel like this kind of conversation is going.

[Mikayla Penna]: That was really well said, Megan, and I complete completely agree with everything you just mentioned, and I just feel like it’s really going to not only automate the workforce, even more because like, that’s what we’ve been doing for the past twenty years is automation, what however you wanna look at it. But for me, it’s like, it gets people to doing the jobs that they were hired for and the skills that they shine at and that they’re good at and then takes away, you know, those hard admin tasks that are just nobody wants to do. Right? And at the end of the day, you could look at it. It saves your, like, overall workforce Payroll, you might not need to hire somebody to do x y and z. If these employees you have have the skills to know how to use Ai, To Make Them More Efficient And Work Faster And Better And Smarter.

[Megan Foley]: Yeah. And Think About all the Jobs That Didn’t Exist Twenty Years ago. Even like, four years ago. Like, a computer programmer or just stuff like this. So I feel like it’s such a cool experience that we get to see the future kinda get built here, Mh And there’s jobs that haven’t even existed yet today that are gonna be like, someone’s job in the future. So, I just feel like that’s a really cool thing. And it just shows you that if your company is willing to adapt, like, ready signal, we talk every day and we’re like, okay, what’s the new like, technology that we wanna see if we can work into our workflows. And we just kind have conversations and balance back and just adapt. It’s a really cool way to jump on the train for the future and really just get more efficient.

[Mikayla Penna]: And even if this doesn’t pan out to be, like, unicorns and rainbows, like, us and this call think it’ll be. At the end of the day, you’re teaching yourself how to be more efficient. And how to use these skills, whether companies embrace it or not, just you as a person can benefit from learning how to prompt and learning how generative AI works.

Wrapping Up and Final Thoughts

[Jason Harper]: Yeah. I don’t know. I’m curious if anybody else feels the same way, but… I use it so much. I am noticing that sometimes I’m talking to people and asking questions. And it’s… I think it’s changing the way that I actually ask questions and talk to other people I don’t I don’t know I’m the only one, but like, I’m definitely noticed. I was like, oh, alright. That’s different.

[Megan Foley]: We prompted him like chat.

[Jason Harper]: Just like how I’m just like, I think like how I I honestly like with my wife, like, we’re very different people are wired very differently. She’s extremely. She’s a… Pharmacist and so she’s very focused and detail oriented. And I think I think it’s actually improved my communications with her because Like, I I I think I ask questions more clearly now or in a different way that actually explains things better. So I don’t… May. Maybe on the only one. I don’t know.

[Megan Foley]: I definitely thought you were gonna talk about. Oh, sorry, Matt. But I definitely thought you were gonna, like, say, act like Pharmacy expert for me and then just go off in different prompts that you’re doing it. But I can definitely see how you have to be more clear and concise because the robots can just you know, totally Pharmacy come out of a different direction if you aren’t asking a very light clear answer to them. I think I’m leaving some assumptions out when I’m asking questions. I’m actually, like just being a little bit more clear and not assuming quite as much because I’ve learned through this rapid,

[Jason Harper]: like you know, query, you know, prompt response, prompt response and like having to hone my questions, so instead of asking. Leave just just less assumptions. Right? So being a little bit more detailed, not a ton, but a little bit more in my speaking, I don’t know. I think it’s it’s doing something in my brain.

[Mikayla Penna]: You’re turning into a robot.

[Matt Schaefer]: See at least it can still be the butt of a lot of jokes in the office around the water guilty of it myself. I I think I just think this is a really an incredible, like moment in time to, like watch evolution happen in real time. And I think that it’s really opportunistic from an efficiency standpoint when you think about us. And our our greatest resource is time and it’s finite. And if I’m not spending eighty or forty or fifty percent of my time, whatever it may be on some of these things that are menial tasks that could be automated, it’s an incredible opportunity. And like when you put that into the scope of work life balance and just effectiveness and throughput that my teams can generate that translates to value and they actually embrace it. I think it is pretty awesome. Right again, this is like maybe the Rosy lens version of what this is, but I think it’s incredibly fascinating. I think that, you it’s just fun to watch this play out because I think we all can benefit and continue to help influence this so that the robots maybe don’t take over.

[Anna Schultz]: Right. Absolutely. Thank you all. Go ahead, Jacob.

[Jacob Newsted]: No. Sorry. I was just gonna say that I find this kinda interesting because, like, previous technological like, influences kind of didn’t affect me. It was a big… Oh, it doesn’t affect. Me at all. And AI affects all of us in every way. Yep. So I kind of hope that whereas it didn’t affect me before me well, maybe now people take fifteen minutes out of their day to just read what the newest stuff is. Like, I’ve done that before as like a researcher. I read papers, but that doesn’t need to be your way of reading this kind of new AI news. Read a blog post, Read whatever the newest thing, Google shoved at you is, and I just hope people try and learn things more often and react to this rather than just have it hit them like a truck. So because it they could happen. It could it could happen. If you’re not aware and ready to read all these things and learn. So that’s all.

[Anna Schultz]: Absolutely to add that. I mean, I think just playing around with the Open AI ChatGPT has been really helpful, you know, just going in there and asking different prompts and kind of learning through trial and error to, you know, seems to be personally a great way to kinda play around with it and test it out and and get better at it.

So Absolutely. Cool. Well, I guess there you have it, an incredible deep dive into the captivating world of ChatGPT and generative AI with very a ton of invaluable insights from our guests. We hope you all found this episode as enlightening and inspiring as we did. And our heartfelt thanks goes out to our experts for sharing their time and our knowledge with us today.

To our audience, if this episode sparked curiosity and got you thinking about the incredible potential of ChatGPT and generative AI for your business. I’d be sure to hit the like button and subscribe to the real intelligence podcast webcast. We’d love to hear your thoughts, so don’t hesitate to reach out and engage with us. And if you’re ready to explore how ChatGPT can revolutionize your business, our team at RXA and Ready Signal are here to guide you on this transformative journey. Please reach out to us at learn@rxa.io to Io and let’s unlock the full potential of generative AI together.

What is ChatGPT?

[Anna Schultz]: So let’s jump into the extraordinary world of ChatGPT and generative AI to empower your business with the knowledge to capitalize on this transformative technology. To start us off, we know that business leaders are keen to understand how ChatGPT and generative AI can revolutionize their operations, business intelligence and data science. But first, I think it’s helpful to truly understand what is generative AI. Jacob, can you walk us through the general definition and more specifically how ChatGPT works?

[Jacob Newsted]: Of course. So I think really the easiest way to go about this is starting just at the words ChatGPT and working backwards. GPT, we can start with that. Stands for generative, pre trained transformer. Going a little bit deeper, we’re gonna have a lot of steps back is transformer. It’s a specific machine learning architecture that kind of takes a human intuition to looking at data.

Transformers have a mechanism called attention. Where when you look at, let’s say a sentence, like the cow jumped over the moon, certain things like the and cow are related to each other. And so the machine learning algorithm kind of pieces these little things together, and it processes data, not just as a collection of words, but as a collection of relations. This is really important when we kind of go together onto a higher level thing, which is the generative part.

So going back to GPT, generative pre trained, the way GPT models are specifically made is their input a series of text, and it is cut off at a certain point. And the machine learning model is then asked, okay, input the next word. And it trains itself kind of in an unsupervised way, although that some people might be shouting at me in the background, but it’s easy to think about it as unsupervised as you just give us some text and you say complete this sentence. Word by word by word. It forms these relations using attention. And eventually, it generates a complete sentence.

Now take a step back and do that over the entire corpus of Wikipedia. Every public source book. And all sorts of text all over the entire world multilingual, Now even with GPT-4, we have images, audio, video, all sorts of things. And it generates these kinds of very general ideas of relationships between words, sometimes images, and it’s able to generate piece by piece, some sort of predictive… this is what should come next.

That’s what GPT is, GPT two, GPT three, GPT four, we’re gonna keep going and they’re gonna probably keep the same architecture with small changes, which would be a completely different topic. But then we really have to talk about the big game changer here, which is the chat part. Chat actually comes from the idea that rather than training a model, but just on predicting the next token or next word depending on what model you’re using, we score how well it outputs things.

So if you’ve ever used GPT before, you know that it can come up with very different outputs to the same prompt. Depending on, like parameters and stuff like that. It’s basically how they sample the next word, so it doesn’t always do the same thing. So they leverage that They say, okay, GPT come up with five different statements to this one prompt. And then a human says, okay. I’m gonna rank these one to five, how pleasing it is it for me the human to read. And then it updates itself based on that over and over and over. And they actually automate it. They cheat… They actually train a model to reward it for them so that they can automatically do this, but more or less. It’s a human ranking things from one to five and updating itself based on that.

So that’s where we go from just predicting the next… word or token to making it human pleasing and easy to interface with. And that’s why chat GPT is so big. Because it takes this kind of scientific approach to predicting the next word or token and gives it a sort of human touch something that’s easier for us to interact with. And that’s why this has been such a big deal is that now we have this almost human like, I hesitate to say that, but almost human like model that anybody can interact with with very little knowledge about what GPT expects. It’s what a human expects now.

[Jason Harper]: I think I think one of the things you said in there too that is just…I don’t like the secret sauce, I guess is that there’s a human and loop in this process that human ranking. I know there’s some nuance to how that’s actually occurring, but it’s that that human ranking where it’s still utilizing this human in the loop for training that has completely made this game changing. Right?

 So when you’re using it, it does it does have that feeling of, like, this is, you know, a person. I thank it. I’m just I feel compelled to say, oh, great, thank you. Like, give it praise and, like, work with it in this, like, positive, like, environment, and it rewards me back with, also like praise like, oh, that’s great. Yeah. Great idea. Oh, this would be really fun to work on. Good luck with your project Like, it’s really interesting. Like those little nuances to that that really make it feel like I’m working with someone. Yeah. I mean, maybe that is creepy and is bad, but I do feel like I’m working with someone when I’m in there playing with this tool.

[Jacob Newsted]: Yeah. And that’s like one of the big breakthroughs. There’s a couple of, like, big kind obstacles we have to go through, but just interacting with these models. I mean, it’s you. It’s anybody. Like, I’m the nerdy dude that kind of is like, oh, hey, numbers cool. But everybody else is interacting with these models. So it’s very huge that people get this kind of feedback. It makes a big difference.

[Anna Schultz]: Absolutely. Thank you, Jacob for walking us through that. Now that we have this general understanding of the program. I think it’s important that we pivot and discuss how businesses can actually apply ChatGPT today.

Optimizing ChatGPT and Generative AI for Business

[Anna Schultz]: So a question for the larger group, maybe Mikayla, Matt and Megan. Can you walk us through how you’ve seen ChatGPT be used to optimize business operations?

[Matt Schaefer]: I’ll go first. So I think that, to Jacob’s com… The way you summarize, it was fantastic, especially for someone on this call that’s not a data science practitioner. I spent my entire career in the world of data and analytics, but I think that what I see here is a fundamental game changer making kind of this world of data science much more approachable to the masses, but for those that can really harness this as a utility in its current form.

I think can be a fundamental game changer from an efficiency standpoint and things like content creation. It’s it’s things as simple as linkedin posts, blogs, email prospecting, you know, I think that there’s a ton of application today, and it’s only in those types of rounds at least from my vantage point, that we’ve only scratched the surface. And it’s amazing. It’s a little bit terrifying, but extremely exciting on, I think what this will do and how this will reshape how you know, businesses run and operate and scale?

[Mikayla Penna]: Awesome, Matt. To add to that, for me, being a sales woman, my entire career. For me, what’s exciting and what I’ve seen a little sliver of to help me so far is just being more efficient and more productive in my role specifically. Like I’m able to give GPT a few prompts and have it spit out my notes from client calls so I can, you know, put together contracts and follow up on emails to, you know, eventually get deals across the finish line. And for me, spending less time on what I like to call my admin tasks in my role and more time building deep connections with clients, is a game changer, and I’m just excited for the future to see where this could go from there.

[Megan Foley]: Yeah. I Totally agree with you Mikayla. I totally feel that every single day and  it’s really cool being in a small company that your small team can do the work of like six to ten people now. And it’s not even just small companies that are, like, really seeing the effects. If anybody from, like the big, like Whole Foods and the Amazon are also seeing machine learning algorithms. Really just take their processes and run with it. There’s so many cool new outputs that really run on chat. And it’s really cool to see how they’re working their way into your everyday sales processes. Anything from, like a copilot that’s running on like, Microsoft suite and just helping you with Linkedin prospecting? So to really just writing Linkedin posts, doing marketing, anything from the visual aspect to the copy itself. And it’s really just making it a lot easier for the little guy compete too.

[Mikayla Penna]: Absolutely Megan. Totally agree.

[Jason Harper]: I wonder some of that note like so obviously Microsoft made a big investment into OpenAI in particular and Microsoft is incorporating it into Github and their tools, but I hadn’t really pieced it together that they they also own Linkedin. Right? So, like, are we seeing are we seeing that level of integration occurring within Linkedin Like you are. Can you talk about that? I don’t know that I’ve actually seen that yet.

[Megan Foley]: Yeah. So I would say that there Is probably about like, six or seven tools that are trying to be the next big thing. And really what it comes down to is they’re still playing around still trying to figure out who like the best one is and who’s gonna to get that Microsoft kinda tie in. But what it comes down to is, all these tools are scraping somebody’s page figuring out what they say their sales are their skills, all these different things, scraping it, putting together a blurb, and you can explain what your company does, and then what’s like your main key product takeaways, and it will scrape their page and see how you match up to make a tailored message to them for you.

[Mikayla Penna]: And to add to that, Megan, that’s something you know that I feel like is a huge game changer because spending time tailoring a small social media, like Linkedin message for each persona you’re talking to is incredibly time consuming. And I used to, you know, spend three, four hours of a block in my day to do this. And now I could do this in thirty minutes, which is amazing and then get back to the work that’s meaningful, and at the end of the day, produces revenue. So I love it.

[Jason Harper]: That’s though, that’s been that’s fantastic. I love it too from that perspective for sure. I would say one of the when… And thinking about the comment about creating the post in that. Like, I just started using Microsoft designer, which I think runs on DALL-E, DALL-E. And I don’t know what y’all experience have been with that, but I started playing with it this weekend and found it to be Length technical term is rubbish. Like, I didn’t find a lot of success in its ability to actually create something I could use. It was entertaining and like, I enjoyed the results from, like a, you know, made me giggle, but I didn’t find anything from the image perspective that was actually useful. Curious what you guys have different experiences with that?

[Matt Schaefer]: Yeah. Sorry. Not as advertised, I think is what the the way that I would sum it up. I think that there could be a ton of really interesting application for it in future versions of this, but I would agree, not as advertised underwhelming a little bit.

[Megan Foley]: Yeah. And to kinda add to Matt’s point, I’ve even expanded it outside of the Microsoft Designer to other like platforms. So like Canva has their own called Canva Magic. And it seems like this really cool feature that’s it’s like, oh wow. It’s gonna place Adobe Photoshop, all this sort of stuff and Even Adobe has their own with Firefly. And really there on definitely like the beginning cusp that I feel like if that didn’t get released when it got released, they would have been a couple years out from it. And really, I felt like they jumped on the bandwagon and threw it out here, and it really shows that humans right now kind of have that that cap space pretty well like managed in that it can’t produce the same things that I feel like people can right now. But who’s we’ll see in a couple of years? Maybe it will be completely different. It’s definitely the beginning.

Quality of Generative AI

[Jason Harper]: Yeah. Yeah. I think it’s been underwhelming and I think you’re right. They just jump the gun. I don’t know why they’re trying to push so fast when it’s just… So you can’t help but compare the quality of output that we’re getting from, like GPT-4 and think, like, how amazing this is, and then my expectations. Have been completely transformed, for what I expect from any of these. So when I get something back from them that maybe a year ago, I would have thought Wow. This is really cool. Look at how good they did. It’s like it’s… My expectations are completely different, and it’s just, yeah, rubbish, I will stand by that.

[Mikayla Penna]: Yeah. I feel like the bar has been set like super high with every iteration of GPT. And now it’s like, the race to the top to get that visual generative AI for several companies, not just open AI and I wanna see who gets there, Like, who can do it.

[Jason Harper]: Yeah. I think the the only place I’ve seen it successfully applied has been in that sort of, like, using…Stable diffusion, where you’re really training your own, and you’re really you’re feeding it quite a lot of your own bespoke image, data, if you will, then it can actually be really powerful, but it’s it’s just not starting from what feels like scratch with with DALL-E or these other these other sources.

[Jacob Newsted]: I kinda feel like a lot of these models, nothing was tailored at first. We we had a lot of stuff that was like, oh, I can generate images based on ninety billion different images on the Internet. But like, to actually put it into a product and say, hey, you can use it with our product x is a completely different expectation because I can ask for a sunny meadow view in the appalachia or something like that. And stable diffusion can come up with something really good. Maybe even DALL-E, I will admit as well that DALL-E tube is pretty rough. But I think the problem is is a lot of people just started throwing spaghetti at the wall, and it’s not tailored to anyone product yet. I think it’s gonna be a little bit, but I am really looking forward to seeing, like, how they navigate that. It’s gonna be kinda interesting.

[Matt Schaefer]: It’s it’s an awareness and an education opportunity right now. Right? I mean, because the I think about this like, who the hell thought Youtube influencer would be a thing, you know, a decade plus ago. And now you’ve got, you know, new opportunity for to be like the the maestro of prompts and like what it can be produced, that’s gonna be a pretty desirable skill set here in not so in not a very long period of time. You, I think that companies are already thinking about this both from an education and enable, but who I’m hiring for and the Job reps that we’re producing, it’s just in it’s influencing everything. I think it’s super fascinating. Right? And it’s… Those who don’t embrace are the ones that are gonna get left behind.

[Jason Harper]: So it’s funny I was having this conversation at a birthday party. On Saturday with a bunch of ten year olds and the parents. And so we were sitting around talking. And of course, since I was there, I brought this up and we were talking about GPT and AI and stuff. And

talking about that, in particular, Matt, the usefulness of the skill set of writing a prompt and how important of a skill set that is to be able to actually write really effective prompts and looking for that as an employer and things like that. And for me, like, I I’m may be wrong about this, But I feel like that’s gonna be an insanely useful skill set for, like, twelve months. And then it’s gonna actually be using GPT and GPT like solutions actually write the prompts. And so your ability to create these really well constructed prompts and the nuance that goes into that is critical, but it’s super short lived because it’s really about create. Like, it really is more about pulling what you’re looking for and stuff. But so I really this whole space. I guess, I just say that it’s a…I don’t know where this… I guess, from a skill set needed to take advantage of this, don’t I don’t have my head around exactly what the right skill set is in the long or even medium term?

AI’s Effects on the Workforce and General Public

[Mikayla Penna]: Yeah. Right now, I see it as in the short term companies hiring for roles like AI project manager or AI data science manager, people that can, you know, have the skills to prompt it and have worked on that and are ahead of the pack in that game and then eventually, from what you were saying Jason, pivoting to…Now I run the whole department within my company. That’s in charge of AI and I managed these tools that we’re using for AI. And I think that’s really exciting and also when I talk to like, my friends and my family that are outside of, like, this whole tech space we live in. They don’t even know what ChatGPT is yet. Right? They’re like, Oh, it’s just, like scary robot stuff. That’s gonna take over the world. Like, no offense mom and dad, but their their generation just doesn’t get it. And I I just think it’s really important to know, like, if you’re not a robot because I’m not yet. Right? Is you’re like, what about, like, soft skills and empathy what about all these things and I don’t think you should be scared of that?

Like, job skills such as, like, communication and empathy and having a deep knowledge of, like, who you’re talking to, now where we stand and in my opinion into the future. That’s not something that AI is going to replace. But if you have the hard skills right of understanding AI and understanding how to prompt it or even how it works like Jason dick or not… I’m sorry. Not Jason. Delete that, Anna. Like Jacob explained at the beginning of this call. I think there’ll be huge skills in the job market and I know everyone in this call that I’m on with today, like, we’re excited that we’re into it and we’re talking about it and I think when the greater public you know, finally embraces AI in general, whether that be DALL-E or ChatGPT. I think we’ll be ahead of the game. We’ll be ahead of the curve, you know, knowing about something and being excited about it and teaching yourself knowledge before the general public goes wild, You know, makes me really happy and excited for where I’m at.

[Jason Harper]: I think that’s a good point. I sometimes forget too. I get I forget that people don’t know what this is. Like, and I’m for my every now and again, like, at this birthday party. I was like, oh, you’ve never I’m trying to describe it it as the everyone knows I was like, you haven’t seen this. I’m like, oh, like, pull up my phone. I’m showing them for the a first time. This thing that I’m, like, yeah. I don’t know. I thought everybody used this. I’m confused by this conversation. Give me a moment to reset my brain to back to reality because not everybody is using this, you know, all day every day and as, you know, it’s like you know, it is. And I think too it, it’s imperfect and there’s lots of issues with it. So it makes sense that most people haven’t fully adopted this because it’s wrong a lot and you have… To like, be willing to deal with this thing that’s, you know, It sounds like it’s super smart and super right, and you can trust if for everything, but it’s you know, it’s not. It’s still so early. I sometimes forget that forget that.

[Mikayla Penna]: Exactly. And I don’t think we’ll see, like, full, like, public use of this until it it is trained to be more accurate. Right?

[Jason Harper]: Mhm.

[Megan Foley]: I feel like the biggest conversation at kids birthday parties with it is about cheating probably Right, Jason.

[Jason Harper]: That does come up. That does certainly come up. Yet. I mean, we’re seeing that I guess, and especially at the college level, like, being able to… Just everyone seems to be using it to write everything. So I don’t know how college is really gonna react to this. I feel like it is sort of like a this is a lean in opportunity to figure out how to incorporate this into everybody’s, you know, educational experience versus you know, I’m already seeing, like, places trying to put the guard walls up to, like, combat and fight the use of this technology and whatever. But I feel like that’s a fool errand, but yes, I could talk about that separately.

[Megan Foley]: The willingness to learn and the adaptability really. I feel like that’s the soft skills needed to really take it on the challenge.

[Mikayla Penna]: Yeah. And I think there’s so many, like, different businesses that can be made out of this, whether it’s a business that detects if someone wrote a paper or took a test. With AI help or, you know, hospitals using AI to make better informed decisions to go the way of human error. There’s so many things that get my wheel spinning, but I’m like, oh, my gosh. So many new businesses. Not even just in tech are gonna come out of this era.

[Jason Harper]: I feel like. Well, I think Mikayla too as you’re talking about that. What are the implications of this talking to a friend of mine last week sort of about some of the implications of some of the the visual and audio components of this and its ability to kinda like, fake things in real time. I think a corollary some of the results of this too will be sort of like, kinda what the opposite of scaling is descaling, I think is the wrong word, but, you know, bringing things back where you’re not able to do…We… We’re gonna rely more on being in person and being in real life with people and being able to… Like, it’s gonna place more value on being in person and talking to people or going to the park, or doing, like, these sort of things, like having conversations that because the amount of, you know, curated automated video content where you’re talking to things or interacting with things that aren’t necessarily actual human on the other side. It’s gonna make phone calls feel less real. Right?

So, like, you know, and sort of I’ve I’ve seen and felt that we’ve changed from, like, came having phone calls feeling very real to then having really video calls now is really my normal defaults today for a lot of conversations that I’m having and that sort of the thing. And I think there’s gonna be another change where it’s like, I just wanna be in person to actually see you and be around you and in your space, and I think same thing with colleges and testing and things like that. Like, and it you’re not… It’s gonna reduce our ability to scale and do these big broad things and actually focus on no, like, more interpersonal re reactions or relationships with teachers, with peers and things like that, which I think that feel is it may slow some things down, but be a true net benefit to help people interact with one another.

[Matt Schaefer]: Oh my gosh. I… That wouldn’t that be an amazing byproduct at this moment in time. I mean, many of us have young kill young kids. We talk about screen time all the time in my house and, like, the video games and, like, how do I…It’s like to an unhealthy level. And I mean, that’s an interesting perspective of the future that this actually drives both innovation and more human connection, then than anything else. I’d… My chips are on that one, Jason. Yeah.

[Jason Harper]: Yeah.

[Anna Schultz]: Yeah. Absolutely. Wouldn’t that be a good a good outcome of all this. And it’s funny. I think a lot of people think it’s, you know, the opposite. It’s driving all towards this AI towards this more technology, but maybe it will have the effect. So we’ve gotten a lot into the

aspects of ChatGPT helping with business operations, maybe image generation, content creation, things like that. One of the other areas I know we’ve all seen great potential for generative AI is in enhancing business intelligence and data driven decision making. So can we kind of steer the conversation and talk about how ChatGPT or generative AI in general can help in generating actionable insights from data?

[Mikayla Penna]: Sure. I can speak to just something that’s kinda driving me and keeping me up at night is how businesses will use this to be more accustomed to their customers. Again, sales girl talking here. So I’m thinking about the money always. And for me, it’s like, okay, In today’s age, say I wanna go and buy a new pair of sweatpants. Right? I’m like, okay, I’m gonna go to like, my five go to stores price is gonna be a factor, the type of pants I’m gonna look for gonna be a factor. And then I kind of, like price shop across the Internet. Right? And now I feel like if companies can incorporate this, let’s like say a big one like a target, for example. If I can go on to target and have GPT like white labeled on their site and ask it questions about what I’m trying to find, and it spits out what they sell to me right away without having to, like, go through their search and do all this. I’m gonna buy it right then instead of just putting it in my shopping cart and waiting half the time, I’ll put thousands of dollars of clothes in my shopping cart, and I’ll never buy them. It’s just like Window shopping on the Internet. And I feel like it could lead to making faster customer decisions, which, you know, lead to more revenue for businesses, whether it’s small and large, and that’s something I think about all the time on all levels of business, specifically retail,

[Anna Schultz]: Yeah. Absolutely.

Generative AI’s Effects on Data Science and Analytics

[Matt Schaefer]: I think from my perspective, like, as you think about the world of BI, data science, analytics. I mean, it still is a very kind of like…I guess, it’s a world of that’s kind of under the wrapper of super technical and I think that this, you know, kind of helps to… I hate saying demystify, but I think really makes that world much more approachable and some things that we’re thinking about like at Ready Signal is playing in this world external data, leading indicators, purpose built forecasts, being able to provide these in a way that they’re both consumable, but also interpretable becomes extremely important in our world, both through, like, trust and actual embrace of the application, I.E, the forecast that I’m gonna rely on to run my business and you know, and drive some of the next four or five decisions that I’m gonna to make today. I think that just the the way that we’re incorporating that to have that human kind of readable summarization of something that’s super tech technical again, not the practitioner on the call, but, you know, I benefit from it, my business benefits from it and my customers benefit every single day and I just really think that that’s been an incredible game changer from my perspective at least from my vantage point. I I don’t know, Jason If you share that or maybe to the to the rest of the panel here. If there’s anything that you comment or challenge in in that statement?

[Jason Harper]: No. I mean, I think, I think it’s very similar to the way that we’re we’re seeing things. I don’t…I mean, I I know Guess what it say. I look like Jacob, you’re gonna chimein? I mean do you have any thoughts on this. Go ahead, Jacob.

[Jacob Newsted]: Yeah. Sorry. Well, I I use it. In my work life right now, we recently had a project that had a large table of customer information, feedback, sentiment, all sorts of things in a structured way, but it’s like text and lots of other things. It’s hard to go through everything and summarize it all. So we used ChatGPT and shoutout to this Python library if any techies are listening, Llama Index, makes it super easy. We are able to ask questions to this table if that makes any sense, and it gives us summarizations. It gives us feedback. What what’s the overall sentiment during this time frame or during this event or something? And it just gives us human readable insights into it, which is huge because whereas where I used to do random data science, like probabilistic analysis on things, I don’t need to do that. All we do is we have an app that a person asks a question and the data answers itself back to you. Which is huge. Because now it’s not me that needs to do this. I make the app. Everybody else asks the questions, which is huge. It’s transformative, and it gives us the ability to do even more crazier things to hopefully answer everybody’s questions.

[Jason Harper]: I’ll say having been having used that specific library, all weekend and including…

Several hours of today. It is amazing. There are lots of pitfalls though to using it. And on the technical side, number one, expect to read in a folder or not a file. I spent two hours trying to figure out why it wasn’t working. And so that’s… Thank you, Tom for helping me with that this morning. Yeah. No. It it happens. But so what’s also interesting is so I was able to process in around three hundred thousand comments in a specific kind of another text customer experience I thing we’re looking at. But what’s interesting is, it’s still not. It doesn’t work on the first try. Right? So so learning, like, even training and training training samples and that. So, like, this is this is not magic. Like, it might sound and put a magic wrapper around it, but like, getting getting actual insights out of data still requires a lot of thinking and time. It’s not as simple as just throwing a dataset set over the wall and then coming back with this magical thing that you can ask questions of it. It requires really intelligent, an intentional organization of the data, looking at it, reacting, responding to it and changing it, right. And so incorporating, we’ve… We’ve had to make, you know, several iterations of changes to this. And I constantly have my little usage screen up to, like, look at my, like, oh, that was four dollars. Oh, that was eight dollars. Oh, that was oh, that was fifty bucks oops. Like, just watching, like, the training of these things, it’s very, you know, it’s interesting just it’s very complicated. So I’d say, like just… That’s what I’m the vision of what you’re doing there and what, you know, what we’re trying to do and the sum of the stuff is, like, just it’s gonna be magic. It’s gonna look like magic to the end user, but like to get there, there is… Oh, there’s a lot of smoke and mirrors and putting things in place. Actually make this thing happen. It’s actually interesting. So speaking to that and specifically Llama index since to Unfortunately, I’m sorry. You struggled with that one.

[Jacob Newsted]: Literally, the thing about that is it creates it your prompts, your questions and everything is not just gonna be tailored to GPT anymore. It’s actually tailored to your dataset set. Because I don’t know if you I don’t wanna get too jargon, if you tried the vector store, python kind of class, basically, it takes your data and it creates…I don’t know. It creates kind of themes and ideas around it and it stores them in these kind of like semantic ways. So even just asking questions of your data, one dataset is not going to have the same kind of semantic relationships between every single row as the next one. So you’re not just fighting against GPT and prompting anymore. It’s really gonna be an art of. And like you said, there’s tons of tuning. It’s lots… Of different things that are just gonna…It’s not magic. Well, we hope to make it look like Magic.

[Jason Harper]: Right. But it isn’t. Yeah. It should feel that later to the users. I I would add two to that. One thing I’ve I’ve felt myself like, so I I… Was a coder at one point in my life. I think I can claim that. There’s some people who would take massive exception to that. With Lary listening, Lary is like you’re not a coder Jason. That’s okay, Lary with one r by the way. But so at any rate, the… I think for me, like, it’s empower me to go write coding guys and, like I’m spending it ton of time, like, writing python. And so the thing is…But that’s the thing. Right. I’m not I’m not a code. Admit it. And And so, like, I run into these little issues that even ChatGPT can’t diagnose because it doesn’t know the issue. And I spent hours on this thing. And this morning, meeting with an actual code. Showing him my cody. He’s like, oh, yeah, that should be a directory not a file, like, within seconds. And I was like, oh, got it. Okay. Thank you. But so I think like, there is sort of, like, this like, fun exploratory stuff where there’s gonna be a lot of skin of the knees. And I think that we’re at a phase now where it’s just not it’s super approachable and easy to access, but to actually get the true results out of it, It’s still really hard. Like, it’s really, really hard and requires like actual skills that are developed when people over a long period of time.

[Anna Schultz]: Thank you all for those answers on that. I think it’ll be really interesting to see how that all develops and at RXA? That’s our bread and butter right? So that’s kinda what we’re we’re working on figuring out.

Automation’s Effects on the Workforce

[Anna Schultz]: One final question for everybody to kinda keep things in a different direction. As these types of generative AI become more integrated into the business world, the skills required for professionals are evolving as we touched on earlier. Do of you guys have any additional thoughts or elaborations on the importance of candidates who can effectively utilize AI tools? Maybe how it’ll impact the future of the job market. Additionally, maybe how companies can work to upskill their current employees to leverage these technologies more effectively.

[Megan Foley]: Yeah. I guess I can start. So I feel like to kinda say, like, the fear of automated work. Is not something that’s novel. It’s been around since the eighteen hundreds believe it or not. And there’s a really interesting passage that I was reading from like a BYU professor who pretty much said that this is not a new idea. And that every single year, we kinda come out with a new thing that this is gonna replace the work. Like, we’re not gonna work anymore, really when you look past, like, past the add it, it just really made everyone more efficient. And, like, we’re were talking about data, so I might as well bring a data source into this. So The world economic forum said that there was eighty-five million jobs that are going to be replaced by AI one day. But if you look at that same statistic, they say that ninety-seven million are going to be created. So work really doesn’t get replaced. It’s just the work that isn’t efficient that people don’t want to do. We didn’t we didn’t wanna to plow the fields ourselves. We built a tractor of thing. So pretty much does kind of what a lot of this has come down to, and you’re gonna get left behind if you don’t adapt. But at the same time, kind of if you adapt and you become more efficient, you’re just not gonna really I feel, like see it in the future, you’re just gonna be looking back at it one day and you’re like, hey, the iphone existed, it didn’t exist twenty years ago. To it does today I can’t imagine my life without it. That’s really where I feel like this kind of conversation is going.

[Mikayla Penna]: That was really well said, Megan, and I complete completely agree with everything you just mentioned, and I just feel like it’s really going to not only automate the workforce, even more because like, that’s what we’ve been doing for the past twenty years is automation, what however you wanna look at it. But for me, it’s like, it gets people to doing the jobs that they were hired for and the skills that they shine at and that they’re good at and then takes away, you know, those hard admin tasks that are just nobody wants to do. Right? And at the end of the day, you could look at it. It saves your, like, overall workforce Payroll, you might not need to hire somebody to do x y and z. If these employees you have have the skills to know how to use Ai, To Make Them More Efficient And Work Faster And Better And Smarter.

[Megan Foley]: Yeah. And Think About all the Jobs That Didn’t Exist Twenty Years ago. Even like, four years ago. Like, a computer programmer or just stuff like this. So I feel like it’s such a cool experience that we get to see the future kinda get built here, Mh And there’s jobs that haven’t even existed yet today that are gonna be like, someone’s job in the future. So, I just feel like that’s a really cool thing. And it just shows you that if your company is willing to adapt, like, ready signal, we talk every day and we’re like, okay, what’s the new like, technology that we wanna see if we can work into our workflows. And we just kind have conversations and balance back and just adapt. It’s a really cool way to jump on the train for the future and really just get more efficient.

[Mikayla Penna]: And even if this doesn’t pan out to be, like, unicorns and rainbows, like, us and this call think it’ll be. At the end of the day, you’re teaching yourself how to be more efficient. And how to use these skills, whether companies embrace it or not, just you as a person can benefit from learning how to prompt and learning how generative AI works.

Wrapping Up and Final Thoughts

[Jason Harper]: Yeah. I don’t know. I’m curious if anybody else feels the same way, but… I use it so much. I am noticing that sometimes I’m talking to people and asking questions. And it’s… I think it’s changing the way that I actually ask questions and talk to other people I don’t I don’t know I’m the only one, but like, I’m definitely noticed. I was like, oh, alright. That’s different.

[Megan Foley]: We prompted him like chat.

[Jason Harper]: Just like how I’m just like, I think like how I I honestly like with my wife, like, we’re very different people are wired very differently. She’s extremely. She’s a… Pharmacist and so she’s very focused and detail oriented. And I think I think it’s actually improved my communications with her because Like, I I I think I ask questions more clearly now or in a different way that actually explains things better. So I don’t… May. Maybe on the only one. I don’t know.

[Megan Foley]: I definitely thought you were gonna talk about. Oh, sorry, Matt. But I definitely thought you were gonna, like, say, act like Pharmacy expert for me and then just go off in different prompts that you’re doing it. But I can definitely see how you have to be more clear and concise because the robots can just you know, totally Pharmacy come out of a different direction if you aren’t asking a very light clear answer to them. I think I’m leaving some assumptions out when I’m asking questions. I’m actually, like just being a little bit more clear and not assuming quite as much because I’ve learned through this rapid,

[Jason Harper]: like you know, query, you know, prompt response, prompt response and like having to hone my questions, so instead of asking. Leave just just less assumptions. Right? So being a little bit more detailed, not a ton, but a little bit more in my speaking, I don’t know. I think it’s it’s doing something in my brain.

[Mikayla Penna]: You’re turning into a robot.

[Matt Schaefer]: See at least it can still be the butt of a lot of jokes in the office around the water guilty of it myself. I I think I just think this is a really an incredible, like moment in time to, like watch evolution happen in real time. And I think that it’s really opportunistic from an efficiency standpoint when you think about us. And our our greatest resource is time and it’s finite. And if I’m not spending eighty or forty or fifty percent of my time, whatever it may be on some of these things that are menial tasks that could be automated, it’s an incredible opportunity. And like when you put that into the scope of work life balance and just effectiveness and throughput that my teams can generate that translates to value and they actually embrace it. I think it is pretty awesome. Right again, this is like maybe the Rosy lens version of what this is, but I think it’s incredibly fascinating. I think that, you it’s just fun to watch this play out because I think we all can benefit and continue to help influence this so that the robots maybe don’t take over.

[Anna Schultz]: Right. Absolutely. Thank you all. Go ahead, Jacob.

[Jacob Newsted]: No. Sorry. I was just gonna say that I find this kinda interesting because, like, previous technological like, influences kind of didn’t affect me. It was a big… Oh, it doesn’t affect. Me at all. And AI affects all of us in every way. Yep. So I kind of hope that whereas it didn’t affect me before me well, maybe now people take fifteen minutes out of their day to just read what the newest stuff is. Like, I’ve done that before as like a researcher. I read papers, but that doesn’t need to be your way of reading this kind of new AI news. Read a blog post, Read whatever the newest thing, Google shoved at you is, and I just hope people try and learn things more often and react to this rather than just have it hit them like a truck. So because it they could happen. It could it could happen. If you’re not aware and ready to read all these things and learn. So that’s all.

[Anna Schultz]: Absolutely to add that. I mean, I think just playing around with the Open AI ChatGPT has been really helpful, you know, just going in there and asking different prompts and kind of learning through trial and error to, you know, seems to be personally a great way to kinda play around with it and test it out and and get better at it.

So Absolutely. Cool. Well, I guess there you have it, an incredible deep dive into the captivating world of ChatGPT and generative AI with very a ton of invaluable insights from our guests. We hope you all found this episode as enlightening and inspiring as we did. And our heartfelt thanks goes out to our experts for sharing their time and our knowledge with us today.

To our audience, if this episode sparked curiosity and got you thinking about the incredible potential of ChatGPT and generative AI for your business. I’d be sure to hit the like button and subscribe to the real intelligence podcast webcast. We’d love to hear your thoughts, so don’t hesitate to reach out and engage with us. And if you’re ready to explore how ChatGPT can revolutionize your business, our team at RXA and Ready Signal are here to guide you on this transformative journey. Please reach out to us at learn@rxa.io to Io and let’s unlock the full potential of generative AI together.

About RXA

RXA is a leading data science consulting company. RXA provides data engineers, data scientists, data strategists, business analysts, and project managers to help organizations at any stage of their data maturity. Our company accelerates analytics road maps, helping customers accomplish in months what would normally take years by providing project-based consulting, long term staff augmentation and direct hire placement staffing services. RXA’s customers also benefit from a suite of software solutions that have been developed in-house, which can be deployed immediately to further accelerate timelines. RXA is proud to be an award-winning partner with leading technology providers including Domo, DataRobot, Alteryx, Tableau and AWS.

Website: https://www.rxa.io

Twitter: @RXAio
LinkedIn: linkedin.com/company/rxa.io