In this episode of Generative AI In The Enterprise, Zach delves into generative AI with guest Mark McKelvey, Co-Founder of Stacked Analytics. The two talk about the staying power of generative AI, with Mark providing insights into its various applications.
From using large language models for data extraction and categorization to exploring novel ways of utilizing generative AI in decision-making systems, Mark shares how he incorporates these tools into his consulting work to drive efficiency and innovation for businesses. The discussion also touches on the risks associated with automation and the importance of societal readiness for the evolving landscape of AI technology.
View This Episode On:
- YouTube: https://youtu.be/cKFgjOIgKh4
- Apple Podcasts: https://podcasts.apple.com/us/podcast/mark-mckelvey-co-founder-of-stacked-analytics/id1730289289?i=1000654058985
- Spotify: https://open.spotify.com/episode/5e0ly1zAUdpTCzRS5qYGxl?si=s4RjeBimT9aV1zTAyVYWOw
- … or wherever you get your podcasts!
About Guest Mark McKelvey:
Mark has many years of experience working with cloud service providers (AWS, GCP, Azure). He gets a kick out of improving decision-making with data. The problems he most enjoys involve building, leading, and working within teams to optimize systems, automate where prudent and uncover value in data stores. The bulk of his experience and expertise lies in designing and implementing data systems including sourcing data, ETL, warehouse design, data analysis, algorithm selection, model tuning, and automation.
About The Generative AI In The Enterprise Series:
Welcome to Keyhole Software’s first-ever Podcast Series, Generative AI in the Enterprise. Chief Architect, Zach Gardner, talks with industry leaders, founders, tech evangelists, and GenAI specialists to find out how they utilize Generative AI in their businesses.
And we’re not talking about the surface-level stuff! We dive into how these bleeding-edge revolutionists use GenAI to increase revenue and decrease operational costs. You’ll learn how they have woven GenAI into the very fabric of their business to push themselves to new limits, beating out competition and exceeding expectations.
See All EpisodesPartial Generative AI In The Enterprise Episode Transcript
Note: this transcript section was created using generative AI tools like YouTube automated transcripts and ChatGPT. There may be typos, slight content changes, or character limits for brevity!
[Music]
Zach Gardner: Ladies and gentlemen, welcome to the Future. My name is Zach Gardner, and I’m the Chief Architect at Keyhole Software. About five or six months ago, I embarked on a journey after hearing a lot about Chat GPT. I initially thought it might go the way of blockchain—a fad that hits the world and then fizzles out. However, it seems generative AI is here to stay.
To form a well-rounded opinion, I wanted to explore this technology beyond my application background and talk to people from diverse fields. So, I searched the far corners of the worldwide web and found some interesting folks willing to chat with me. Today, I’m excited to have Mark McKelvey, co-founder of Stacked Analytics, with me.
Zach Gardner: Mark, how’s it going?
Mark McKelvey: Life is good. Thanks, Zach.
Zach Gardner: Great to have you here. And just a quick disclaimer, the views and opinions expressed are those of the participants and do not necessarily reflect those of their employers, trade organizations, yacht clubs, or loyalty programs. We’re just two guys talking and having a good time. So, Mark, for those in the audience who don’t know you, could you give us a bit about your background? How did you get into technology, and what was your first computer? Also, what excites you about using data to solve real-world problems?
Mark McKelvey: Sure. My first computer is a tough one to recall. My mom worked for King Radio and then AlliedSignal, so we got a second phone line and a modem when I was about 12, which was 30 years ago. I don’t remember the exact computer, but I was online via DOS prompt back then.
I started my career in the military doing signals intelligence analyst work. We used models to predict the source of frequencies and other aspects of signals floating through the air. After that, I got a degree in computer science, did some web development for the Army at their Analytics Center, and got interested in operations research, which was kind of a precursor to data science.
When data science and big data became popular, I transitioned into the business world. I started as a data scientist, led a software engineering team building analytics software, and now I’m consulting, helping businesses use data and machine learning technologies to find efficiencies or develop new products.
Zach Gardner: That’s fascinating. Thank you for your service. I think your diverse background gives you a unique perspective on generative AI. Have you come across any interesting or novel use cases for this technology, either personally or professionally? Are there any aspects people aren’t talking about that they should be?
Mark McKelvey: Mostly, my focus now is on data and analytics. I see generative AI as a tool to build decision-making systems and help companies understand their business. For example, I use these tools to read restaurant reviews or old reports, extracting data to categorize and analyze customer feedback without manual effort. This helps build dashboards showing trends, like slower service on certain days at specific locations.
I’ve applied typical uses like summarization, retrieval-augmented generation, and natural language search to improve user experiences and make complex information more accessible. Personally, I use Chat GPT and DALL-E for LinkedIn posts, tuning my writing and generating images.
Zach Gardner: Yes, tools like Chat GPT are great for overcoming writer’s block. Starting with a blank page can be intimidating, and AI can help outline ideas. You mentioned retrieval-augmented generation, which helps mitigate hallucinations by providing a set of documents for the AI to reference. I’m curious if you’ve found other ways to reduce hallucinations or understand why they happen.
Mark McKelvey: Hallucinations are a byproduct of the probabilistic nature of these tools. Creativity in AI requires some level of hallucination to generate new content, not just repeat the same answers. There are great articles online explaining why hallucination is necessary. Retrieval-augmented generation is crucial for combating this, especially when building products. If people wanted a generic answer from Chat GPT, they could go directly to it.
So, having a voice and an opinion that comes from content that you are comfortable with or have produced yourself is important. Make sure that your responses include references to that content so you can provide links to more information about X, Y, or Z. If you’re trying to sell something, embed ads or references to products your company offers right in those responses. For example, you could say, “This is the reason you should buy this weed killer” or whatever it might be.
Zach Gardner: Yeah, for sure. I think my lawn service makes for a great weed killer and a great grass killer too. No shade being thrown, I will not name names, but we are not returning as customers.
You mentioned it briefly before, but I’m curious if you can elaborate more on how you are using these tools in your professional life. Are you using Chat GPT to generate more content? Have you tried using it for creating PowerPoints or running it locally? That’s one of the things I think will really blow up as consumer-grade hardware becomes more powerful. Have you tried running Llama locally yet?
Mark McKelvey: I have no interest in running anything locally. I try to do everything on cloud servers for a few reasons. It might be a fun little activity, I suppose. Professionally, I use it to help me write and clean up content. I’ve seen entire presentations produced by uploading some bullet points to one of these tools. I use it a lot for my clients since I’m a consultant.
For example, I work with a company that’s been around since the 70s. They’ve always stored the output of their work in Word documents or WordPerfect documents. They can’t efficiently understand if they’ve already done work in a specific area. Using these models, I extract particular data points and store those in a data warehouse so that we can set visualization tools like PowerBI on top of them. This is somewhat of a novel use case.
Zach Gardner: WordPerfect? I haven’t thought about WordPerfect in quite a while. That’s a blast from the past. I read that Walmart has their own internally trained language model. They faced similar challenges. It was difficult for people to understand the vast corpus of resources they had from all these different documents and departments over the years. They used retrieval-augmented generation to load up everything, making it easier to search and find specific policies or references.
As you mentioned, it’s smart not just to ask for the answer but also where the answer was found so humans can curate the response. Never just copy and paste directly from the AI—except for images. If it’s text, give me ideas, but I want to be the one to write it.
Mark McKelvey: Absolutely. I wish I could push the pixels around in an image. I need to learn Photoshop or something because sometimes the images I get back aren’t quite what I want. It’s difficult to tell the AI, “This is the part I don’t like,” without getting something completely different.
Zach Gardner: One thing we haven’t touched on yet is the risks. I’d like to dive into what you see as the biggest risks with this technology. I work with an organization focused on expanding career pathways, and I think a lot about what my children are going to do in a future where we have robots learning to fold shirts or doing dishes. What do we do in a world with advanced automation?
Mark McKelvey: That’s a great question. I was looking at a chart showing jobs replaced by industrialization and automation and the jobs created as a result. Until the 90s, the paths followed each other. But when computers became more common, the number of jobs created started to decline. Generative technology and robotics will likely widen that gap further. As a society, we need to think about what we want humans to do. This question might come with a lot of pain if we don’t address it proactively.
Zach Gardner: Very insightful point. Rosie was her name—the robot maid from The Jetsons. It’s the name of our little robot cleaner, too—a Roomba.
Mark McKelvey: Yeah, Roomba.
Zach Gardner: Mark, this has been great. Thank you so much for your time and for agreeing to come on the program. It’s always a pleasure talking to you.
Mark McKelvey: Likewise, thanks for having me.
Zach Gardner: And ladies and gentlemen, we will catch you in the future.
[Music]
Subscribe on
Latest Blog Posts
Blog Topics
- .NET
- .NET Core
- Agile
- AI
- Angular
- Apache
- API Development
- Architecture
- Articles
- ASP.NET
- Automation
- AWS
- Azure
- BackboneJS
- Blazor
- Blazor Server in .NET 6 Series
- Blockchain
- Business Intelligence (BI)
- C#
- Chat GPT
- CI/CD
- Cloud
- COBOL
- Community
- Company News
- Consulting
- Conversational Apps
- Creating an FHIR API
- CSS & HTML
- Data Management
- Data Science
- Databases
- Design
- Development Technologies & Tools
- DevOps
- Docker
- Educational Event
- Effective Automated Testing With Spring Series
- Flutter
- Gen AI In The Enterprise
- Git
- Go
- Google Cloud Platform
- GraphQL
- Groovy
- Heroku
- HTML5
- Hyperledger
- Infrastructure As Code (IaC)
- Intro to Spring Batch Series
- Java
- JavaScript
- JavaScript Debugging Series
- JHipster Series
- Kansas City
- Keyhole
- Keyhole Creations
- Kubernetes
- Learning Svelte
- Machine Learning
- Microservices
- Mobile
- Modernization
- moderntoolingseries
- MongoDB
- MySQL
- Next Level
- Node.js
- OpenShift
- openshiftseries
- Opinion
- Podcasts
- PostgreSQL
- PowerBI
- Programming
- Project Management
- Python
- RAG
- React
- React Native
- REST
- Scaling PHP Apps
- Security
- Service Fabric
- Single-Page Application
- Soft Skills
- solidfoundationsseries
- Spring
- Spring Batch
- Spring Boot
- SQL
- SQL Server
- Tableau
- Testing
- Testing React Native Series
- Tutorial
- TypeScript
- UI/UX
- Unity3D Series
- Unity3D Series 2
- Videos
- Vue.js
- Xamarin