Inspired Execution

Chet Kapoor: Jerzy, welcome to Inspired Execution. 

Jerzy Biernacki: Hello, great to meet you. 

Chet Kapoor: You have your PhD in computer science and have spent the last 9+ years at Miquido. 

Jerzy Biernacki: Yes, almost 10 now.

Chet Kapoor: What's one thing about your career journey that most people do not know? 

Jerzy Biernacki: One thing only. Well, I think most people don't know how it all started. So I think it all started when I was like around 14 years old. I was on a school trip to the academic computer center of our local university. And I remember being like utterly mesmerized by rows of computers and server racks. And I felt like really, this was my natural habitat. So from that moment, I knew I wanted to dive into the world of computer science. 

Chet Kapoor: That's awesome. So has it been something you regretted or you're very grateful that you started? 

Jerzy Biernacki: Oh, I'm really grateful that I started. 

Chet Kapoor: So if I was a student today, what advice would you give me today, right, 2024? 

Jerzy Biernacki: If you would be a student, right, still studying. I think... 

Chet Kapoor: Yes, a computer science student. 

Jerzy Biernacki: Yeah, I think I would give you advice to start working in the industry. Because by the second year at my university, I already was working in the industry. I spent like... I tried to do both studies and work. And I think that was the great advantage that I have. After completing studies, I also had experience in the industry. So yeah, I think that's the best advice I can give. 

Chet Kapoor: You know, it's funny, Jerzy, you say that, right? I had a... When I was going through my computer engineering degree here, I had to work because that's how I paid for school. And I had a small little company where we were doing programming, Lotus 1-3 and a dbase and all that stuff. This was a long time ago. But I think that shaped me probably a little bit more than the courses I took. 

Jerzy Biernacki: Yeah. 

Chet Kapoor: Right? It's really interesting. The real world experience. You know, we talk about internships. People don't realize how much internships actually shape the way we think about our professional lives. 

Jerzy Biernacki: Yeah, definitely. 

Chet Kapoor: Let's talk about AI, right? Our point of view is very simple. 2024 is about production AI, right? And 2025 is about transformational AI. What does that mean? That means every customer we talk to will have some kind of AI app in production in 2024. It could be really small. It could be a pilot. It could be an internal app. Doesn't matter. But everybody's going to do one. 2025 is when they start thinking about people with agents versus people without, right? And start thinking about really things that affect the top line and the bottom line of the company, right? Yeah. And try to transform the companies around it. Would you agree with that? And if not, how would you... What would your point of view be? 

Jerzy Biernacki: Yeah, I think like Generative AI is definitely going to change the industries and it's going to be like everywhere. Because it just increases the efficiency of the processes. So every time we implement it, we gain some value and we get ROI and therefore, why not use it? I think it will change more and more and it will be present in every industry. And in broader terms, AI, like machine learning, it's also becoming cheaper to implement. And we're seeing these niche examples of machine learning being implemented across different industries. So I think it's going to be more common for industries that previously were considered less technical, like, I don't know, agriculture, for example, to implement AI and get value from it. Because it's like the tools are present there. You can use it. You can build cheaper, faster, and get return on investments much faster. 

Chet Kapoor: Have you found that... Back to our timeframe discussion, Jerzy, have you found that people are going to put things in transform in this production AI, this calendar year? And do you believe that by the end of this, if we were having this conversation a year from now, would you have seen enough interesting new use cases that we are not seeing today that actually changes companies and industries? 

Jerzy Biernacki: I think companies are still struggling with some challenges. And I think they're not... I don't see enough use cases, not as much as I would expect by now. Many companies still struggle with... Many struggle with finding use cases. That's really strange for me, because I see use cases everywhere, but many still struggle. Many are still afraid of AI, of regulations, of hallucinations, of bias, and many other challenges. And they are just afraid to start. And I think that's a big mistake. 

Chet Kapoor: Do you think they're afraid to start because they think that AI is going to replace them? Do you think there's an underlying sentiment here? Or do you think the hallucinations and things like that are real? Or is it both? 

Jerzy Biernacki: Well, when it comes to AI replacing humans, I think for now AI is going to more like to augment us and not totally replace us. But definitely it will reduce the time we need to complete certain tasks. So in the short term, I predict that there will be some downsizing in the companies and then people would need to change the way they work. But in the long term, usually every change, every technological advancement made us more happy and then as a society gave us more time to actually live our lives. So I think this is a change for good. And still people are afraid of it, right? I think it depends on who in the company you ask about it. Because the management is not afraid of it. They probably want to do it. In terms of they're not afraid to downsize, right? They want to make the most of their resources. But when you, I don't know, for example, ask regular employees, they might be afraid. But on the other hand, these regulations and hallucinations and all the concerns regarding the challenges, this is mostly a concern of the management and the decision makers. 

Chet Kapoor: Yeah, for sure. You know, it's really interesting, Jerzy, it's that every technology wave I've been part of, right? Client server, web, mobile, cloud, there's been resistance from people, right? It's always been there, no matter what you do, right? But I feel like this time around, I think it will be deeper. The resistance will be deeper than digital transformations, for example. And the reason is because I think one is Gen AI is moving really quickly. And the second thing is, it is more human-like than anything else we've ever done from a technology point of view in the history of technology. 

Jerzy Biernacki: Yes. Yes, definitely. 

Chet Kapoor: Would you agree with that? 

Jerzy Biernacki: Yeah, definitely. And you can see examples of that, like Klarna, because their system, like it basically replaced 700 employees. So it all changes the way we work.

Chet Kapoor:  Yeah, for sure. For sure. I want to come back to this in a couple of minutes, but let's talk about AI regulation, right? You've said AI is not a silver bullet, but a tool that requires careful and responsible use. 

Jerzy Biernacki: Yeah. 

Chet Kapoor: At the same time, I believe innovation is key. How do you think about, and by the way, the EU has done, you know, the state of California has passed a bill. The EU has done a lot of work. I was in Davos. I heard the EU folks and the UN folks talk a little bit about this. So people have been thinking about this quite a bit. What's your perspective on regulation, governance, and innovation? 

Jerzy Biernacki: So I think it's a good direction to make sure that the tools that we're using are secure and we consider ethical aspects and we consider transparency, fairness, and accountability in all the projects that include AI. And some security measures are required to do that. And I think the more transparency we have, the better, especially in these high risk models. I have some concerns regarding the amount of additional work, like paperwork, documentation on top of what we already provide that is going to add to the cost of developing AI and therefore make the development less appealing. But I think it's a good direction to make sure that the solutions are secure. 

Chet Kapoor: I have, I'm generally, I've actually said, as I've gone through all the different waves, I've always said, let technologists do what technologists do, which is innovate. And then as things get more mature, you can start regulating. This time around, I've changed my position. I've said, yes, this technology called Gen-AI will have negative aspects to it, and it'll happen a lot faster. By the way, all technology waves have a negative aspect to them, right? And it's true for all of them. But this time, I think we need regulators involved a little sooner. My concern becomes the balance, like you said, paperwork, the balance of who is regulating, the people who are regulating, do they actually understand the technology and the capabilities, or they're purely coming at it from a political issue of we're going to save jobs, right? So that's one. The second thing is, you've got to regulate, you have to have some guardrails, but not too much, like you said, the paperwork, because otherwise, engineers like you and me and others that we work with will just go and do something else because if you cannot freely decide what you want to do, you're just not going to do it. Would you agree with that? 

Jerzy Biernacki: Yes, yes, I agree. I think why people are afraid of AI is that in traditional IT systems, we control programs’ operations using a set of instructions in the form of code. It's like a really strict language, and these systems are deterministic by nature, and there's no element of randomness. And so for the same set of instructions and data, you get the same output. With generative AI, it's not that simple because systems are probabilistic, and this makes them inherently non-deterministic. So that's why they are creative, right? But on the other hand, it's hard to control them. Another aspect is that you use natural language to communicate with them. So it's not like a strict language like code. 

Chet Kapoor: Yeah, it's not HTTP, for example, or HTML, or whatever else it might be. 

Jerzy Biernacki: Yeah, so it's like communication, and most of the problems you have at work and relationships in families, it comes from bad communication. So unfortunately, here, we have communication issues with the models. And there are, like many I would say, popular cases of AI failures, and people are afraid of being on the spot and being pointed out as the one who failed. And precisely because of this non-deterministic nature and the communication in natural language, solutions utilizing large language models require additional deterministic controls and security mechanisms. So there are many different challenges here, like hallucinations are just one thing, right? But you have to also take into account protecting against attacks like prompt injection, you need to ensure compliance with regulations, you need to ensure compliance with social norms, company policies, and business rules. So all these require deterministic mechanisms and the guardrails to make sure that the solution is safe. 

Chet Kapoor: Yeah, for sure, for sure. Very, very, very well said. Really, really well said. So let's talk about your company. You've been there now for 9+ years, as we talked about it. You know, customers are, at the end of the day, solving every situation we're in. We are solving for thought scarcity and time scarcity, right? With people and processes. But like you said, people are just a little scared. What has been your approach at Miquido? 

Jerzy Biernacki: About approaching AI? 

Chet Kapoor: Yeah, approaching Gen AI, right? When you talk to developers and you're talking about use cases, when you're talking to executives, right? What has your advice been? 

Jerzy Biernacki: I think my general advice to implement this mechanism, these techniques to combat these challenges, to neutralize the risks. And there are many different techniques, like the proper design, knowledge management, access control, evaluation, validation, and testing. And each of these mechanisms allow us to get more control over the result. And yeah, I think the most important thing that we've done at Miquido to actually combat these issues is that we developed a framework. We call it AI Kickstarter. That enables rapid creation of robust and secure solutions based on large language models. And it's not just another tool for building PLCs and prototypes. It's a tool that generally speeds up the work and provides easy control over the results. And it includes all these advanced mechanisms for data protection, access control, result validation, evaluation, and testing. And a significant portion of this framework is already published as open source. So you can find it on GitHub under MIT. So you can actually use it commercially. So I encourage you to test it out and build your own applications using cloud language models. On GitHub, you can find examples, instructions, and even a video course. So yeah, you can try it. I personally think and believe that this is better than BlankChain and Llama Index and other frameworks that are out there. 

Chet Kapoor: That's awesome. And as long as you're helping developers build applications sooner rather than later, right? I mean, you know, the word that we keep talking about is agility. Yeah. How do you get them to do it a lot? It doesn't, you know, the GitHub stars make a difference, but it's still early in the market, right? What matters is, are you making the life easier for an average developer? Not a data scientist, right? And, not a data engineer, but an average JavaScript or Java developer, or even a Python developer. 

Jerzy Biernacki: Yes, exactly. 

Chet Kapoor: That's awesome. All right. This has been super interesting. So let's go to the rapid fire round. So I'll ask a bunch of questions. I have five of them and answer as quickly as you possibly can. What's the most interesting use case you've seen for Gen AI? 

Jerzy Biernacki: I've seen so many different use cases. Hard to pick one. Most interesting one. I don't know, the first one that comes to my mind is like from the entertainment. Actually, the first one that I've seen, even like before the chat GPT and all this, sort of the first one that inspired me was an RPG game called Dungeon something. And I don't recall the exact name, but it allowed you to actually live your story and you could start a narrative and then it generated a few paragraphs of text. And then you could actually write what happens next and the story continued. And that was actually the story that inspired me to really dive deep into large language models, their capabilities. And that's why we admitted to large language models. And yeah, I think it's one of the things that inspired me a lot. 

Chet Kapoor: That's great. That's great. What, which household chore would you want to be, would you want to be automated? 

Jerzy Biernacki: I think this one is, it's interesting because what we imagined I don't know, like 30 years ago is that household chores would be the first ones to be automated, right? And right now it occurs that they're the most difficult ones to actually automate. We automate creative things, like generating the content, generating music, images, but we can't really automate I don't know, doing laundry. And it's, if I were to choose one thing, I guess it would be like ironing my shirts. 

Chet Kapoor: Ah, that's kind of cool. By the way, I've asked this question of a lot of guests and every one of them has something to do with laundry, right? Around laundry is where the problem comes in, not dishwashing, but laundry, or vacuuming. What global problem do you want AI to solve first? 

Jerzy Biernacki: Well, global problem. I think my wish would be for the wars to end, but I'm not sure how AI could solve that. If I knew, I would be already working on that. 

Chet Kapoor: And by the way, the war is close to you, right, Poland? So it's not like it's very far away. So that's very well said. One bold prediction for the future of Gen AI? 

Jerzy Biernacki: I think it's going to replace most of our current tasks in, I don't know, 15 to 20 years. So we'll be doing something completely else. I don't know what exactly, but it will definitely replace most of the tasks that we're currently doing. 

Chet Kapoor: I would, my favorite, by the way, would be somebody who can go through my email. 

Jerzy Biernacki: Yeah. 

Chet Kapoor: Like 200 messages or 300 messages a day, go through them and say, I've looked through all of them, 80% were updates, 10%, you need to go off and respond this way, and the other 10%, you need to think about it. So now I've only got six emails to think about, and now I'm going to, or I just say, FIIs, approvals, and then creativity. Those buckets would be awesome. 

Jerzy Biernacki: Yeah, I think it's closer than we think. I think it's quite possible to implement it in the near future. 

Chet Kapoor: All right, we'll have to talk about that for sure. Okay, one last question. A few words or a phrase that best describes great leaders.

Jerzy Biernacki: I think a great leader needs to know their limits, what they know, and admit that they don't know something. And I think they need to work with the team, right? Because a leader without a team is not a leader. So working with a team and motivating them, engaging them, it's like the most important part, like bringing this creative factor in them. Yeah. 

Chet Kapoor: It's awesome. No, that's very well said. Jerzy, this has been awesome. Thank you very, very much for joining us. I'm sure our listeners are going to enjoy this. We appreciate your perspective on AI. Clearly, I can see scars, right? You've done this before. And so having people share this from you is going to be awesome. Again, thank you very, very much.

Jerzy Biernacki: Thanks for having me. It was a pleasure.