Accelerating Learning Processes with AI and ML – Dan Ciruli, D2iQ

This is Techstrong tv. Hello, I'm Amanda Ani and I'm here today with Dan Oui. He is the Vice President of Product Management at D two iq.

How are you doing today? I'm doing great, Amanda. How are you doing today?

Doing well. So can you tell me a little bit about D two iq? Sure.

D two IQ is a leading provider of Kubernetes management platform. We have a platform that helps companies, enterprises, and government as they adopt cloud native, uh, development and deployment. Ok, wonderful.

And, um, share a little bit about your background and experience with our audience. Uh, sure. I have, uh, started my career as an engineer and spent about a decade doing that before moving into product management.

Um, and since being in product management, I have always worked on enterprise products that help technical people do their jobs better in one sense or another. Um, I've been at D two I Q now for about a year prior to that I was, uh, leading the platform team at a company called Zuora that does subscription management, um, software. But I spent almost a decade before that at Google where I worked on Google's a p i infrastructure.

And then as Google Cloud became a thing, I worked on, um, uh, a variety of things in open source. I was, uh, one of the founding members of the OPI Open API initiative. I then sat on the is Istio, uh, governance board, um, and helped with, as we defined, you know, how services communicate in the cloud native world.

Wonderful. Lots of great things there. I know.

Um, open source has really changed, uh, how people are able to work and produce. So going back, you said you've had a lot of interest in ai, uh, machine learning and technology that helps people in the workplace. So I know we have some, uh, generative AI platforms that have really just taken the world by storm, like chat G P T recently and then chat GT four rolled out.

Um, and I think more people than ever outside of the normal traditional tech industry are really noticing this technology now. So how do you think that AI and also machine learning are gonna help people in the workplace now? Well, I think, I think it's, you can almost say that in the present tense, not even in the future how they are going to, like, how are they today?

They, they're, they're literally helping people today and, and they're help. What's interesting to me is it's helping non-technical people and it's helping technical people. And I have plenty of examples of both.

One thing is, is, uh, and how it's helping non-technical people, um, generative ai, I, I I, I don't really love the term AI because I don't think it's really artificial intelligence. I think it's simulated, but one thing it does very well is it writes very well. And, uh, many, many people are using it to help produce copy or to help in brainstorming.

And, and that is happening all over. You know, I would guess literally that the quality of press release, the quality of the writing in press releases, I I would specify that has improved in the last year in the world because of generative ai. Because these, these machines, uh, you know, what they're very, very good at is sequencing words and they can just phrase things very well.

So I shouldn't say that the quality of the press relief has release has gotten better. It's a form versus content thing. They can help with the form.

It can't necessarily help with the content if you don't have a good idea for your press release, you know, if you don't have a good product that you're writing about, generative AI isn't gonna make it better, but it can literally help you with the phrasing, you know, and, and I think it's helping people compose emails that are succinct, that are well phrased as well. And then there's what it's doing on the technical side, which I also think is super interesting. Yeah.

Um, so go into that and then, and then I have some additional questions, but yes. How is it helping on the technical side? So let me give a specific example.

Um, uh, when I was at Google, as I said, I worked on the API infrastructure, and at the time, Google had somewhere around 200 public APIs, and we supported seven different client library languages, right? We supported, uh, you know, Java and Python. Um, maybe Ruby was one of 'em, Java script for sure.

And, uh, we had, so you've got 200 APIs, you've got seven languages you support if you do a sample in every language that's 1400 samples. Um, and even for a company of Google's size that was way more than we could staff, we didn't have people who were experts in those languages, who were also users of those APIs who could craft samples and then keep those samples up to date. It was just impossible.

And as a result, we didn't have samples for most of those combinations. Um, it was a difficult problem and, and it was one that we didn't really have a solution to. So when I saw chat, G P T could generate code, and I'm not a Google anymore, but literally the first thing I did was I, I went and said, show me how to call, uh, the Google translate a p i from Python.

And boom, it generated a sample. That was not only was it, was it well done, it also gave me things like, here's where you go to get your I A P I key in order to, to plug in here. And then I said, well, now, uh, you know, I, I asked it to make some changes to how it called it, it did that, and then I said, uh, and now can you combine it?

So I could say, translate something from Spanish to English and then put it through this sentiment analysis, a p i, and it could do that immediately. So, so I'm thinking, you know, if, if I were to Google now, I would just say, yeah, we can have a sample for every API in every language, because we can have a machine generate them. You could say, comment the code and they would be well commented.

So it was pretty amazing to think that this thing that you, you know, at Google a few years ago was just completely out of the realm of the possibility now anyone could do. That's amazing. So going back to the writing aspect and then the technology aspect, both of those, in my mind, I'm thinking just from hearing you speak, how could we maybe harness AI to actually educate and um, level up skills?

Um, because it feels to me like by utilizing these platforms, you're also learning at the same time and improving your skills a lot more quickly. And in, in reality, I think they're best for, for that learning phase. What I would not do today is use chat G P T to generate any code I expected to put in production.

You know, I, I wouldn't, I would never trust code that came out of it to go straight into production. You need someone just like code that a human writes someone needs to review. It needs to make sure that it's doing the right thing and that it's observing best practices.

However, if you're trying to learn how to do a thing, it's fantastic. And the thing that makes it better than a, like a prepackaged tutorial is that you can ask it questions. You can say, Hey, do this.

Well, what if I wanted to do that? Which you can do if you have like a, an instructor sitting with you, but you can't do, if you're just observing a tutorial, if someone's crafted a tutorial and that interactivity, you can get with it as you're learning something out. Now I want to add this.

How about if I try this? It's fantastic. I, you know, another project that I worked on is, is this Istio project.

It's part of the cloud native landscape. It's very complicated to get started. And that was another thing I said, help, help me do this.

We use a file called the YAML file, which can be really complicated and difficult for people to understand how the concepts work together. I asked Chad g P T to help with that, and it did right away. And then I said, now tweak it.

So it does this and it can do that. And, and the fact that it can be an interactive learning tool makes it fantastic for learning. It's not the same as having an expert there, but it's, uh, it is a lot easier than having an expert there.

It's almost like a great on the job learning tool and this buffer zone where, I mean, everyone's busy. Everybody has a lot of work. And so rather than going and asking someone else a person for all this help, you first have this tool that you can ask for help and then maybe figure it out yourself so that you don't have to take up somebody else's time.

And, and often you don't have access to that expert. You know, a student at Lear, a student at home today who wants to learn how to, uh, you know, solve a particular programming, programming problem in Python can just go do that. Just ask this thing and get started and then continue to, and, and that's an important part is continue to ask it more questions cuz it has retains context.

Yes. Um, definitely in the classroom, I see so many use cases for, it mean people that struggle with math, they're utilizing AI to help solve difficult problems, but not just give them the answer, it actually lays it out for them. Um, so I feel, um, and then just improving the writing skills, um, as you said, so, so many, so many ways to just upscale level up and learn more quickly.

But then also, let's talk about digital transformation. Of course, that's what our platform covers and most businesses are deep in digital transformation efforts. So how can we utilize these tools to help augment and speed up these processes?

One of the interesting things that happens in digital transformation, I think is, is you want to become much more data-driven. All of a sudden all your systems are digital and you're gathering all this data. Um, we right now need to hire data scientists.

If I wanna know something about my user base, um, we're gonna have da data scientists who are going to figure out how to get data into a data lake. And then I say, well, I need a dashboard that does this, or I want to know this about my, my customer base. Someone needs to figure out how to go write the correct query that's gonna query the correct data sources to, to pull these answers together.

One thing I think we will see in, in these, uh, ml slash ai systems is the ability to ask the system, Hey, I need, uh, you know, a breakdown, ask this in English, a breakdown of our customers, you know, broken down by quarter, separated by product, and, uh, see the growth over time. Ask that in English, right? And type that query and have it return with, oh, can you make that a bar graph and instead go back eight quarters instead of four.

Right. And just interact with the systems in English. And, and that's gonna change, you know, if you, we think about a tool like Excel or, or Google Sheets, how complicated it can be.

You know, you, some people are wizards and they know how to use pivot tables and they know how to write select queries and these, these complicated things in spreadsheets, when really someone started that by saying, Hey, I need you to break down the data like this. Well, these AI systems in, in a, in a digitally transformed business will be able to just take those queries in English and, and return that data. Now there's a scary time where we're not sure if it's doing it right.

Like you're gonna have to, you're gonna have to see kind of the proof are, are we sure it's writing those queries correctly? But the, the, uh, accessibility of of getting those, those, uh, answers, instead of sending your data scientist off for weeks to do it, uh, is, is going to be in very interesting. And then your data scientist will ha still have a job, but it's gonna be how do you get the right data into the system?

How do you ensure that the system knows how to interpret data? So the, uh, I don't think of this so much as a, as a, um, a, you know, job eliminator as rather it's gonna transform what, what the jobs are and what people have to do. Mm-hmm.

I'm wondering if this is actually gonna open up the ability for more, um, non-traditional, um, tech professionals maybe that don't have as much tech skill or background will now be able to come in and do some of these jobs with these tools. E every generation of of technology transformation does that, right? E every, everything does Excel itself was just a landmark.

Like it used to be very, very, very difficult because you had to be able to program right. Computer language in order to kind of mash data around. And Excel was like, no, you can just, you can just do this with pluses and minuses and equals signs, right?

That in itself, we don't think of it that way anymore, but in, in a way that was functional programming and declarative programming and, and it, and it did and it made it, so you went from like, literally there was a, you know, a hundred thousand or a million people on earth who knew how to, who knew how to program to, there's a hundred million and this will be the same thing that it, it does absolutely give those abilities to people who, who, you know, weren't able to learn how to run a data pipeline in the past. And I think that's fantastic. Yeah.

You no longer, maybe in the future, you're no longer gonna have to have actual data scientists or, um, people who have gone for these, um, full degrees that can utilize this technology and come in with a little bit less skills and, and catch up. I I, I don't think that part is right in the way that when Excel made it, so you didn't need to, you know, write a computer program in order to manipulate lots of data that didn't eliminate programming jobs. Right.

In fact, there were more, and in fact we will be get, you know, we are also at the same time this is happening, getting more and more data sources, having to figure out how to combine them, how to put them into you, how are you gonna design a place to store it? Your AI needs to find that data somehow. And so I, like I said, I don't think this eliminates jobs at all.

It transforms some of them and it's gonna create great oceans of jobs that we didn't see before in the way that like when, when web development happened, you know, web development, all of a sudden there's more need for, for, uh, for people who can do different types of programming. And, and the definition of programming got wider and wider and wider from when I started, when it was all basically c or c plus plus to, you know, there's, there's 50 different languages that attack different problems and, uh, there become more and more niche things. So, so I think it's, it's a net job creator, but it's gonna transform 'em.

Well, that's good. I'm sure that most people would be glad to hear that and relieve that they don't have to worry about their jobs because that is a concern as, as these, um, technology progresses, I hear that a lot. Are we gonna lose our jobs?

And, uh, but really it's just gonna create new ones and change the way they do the jobs. Our great-great grandfathers thought the same thing with machines, right? They, they, they really did.

And, and then electricity and every, every new technology we see the things that it's gonna make easier and we think, but that's someone's job today. But in reality, you know, we always find, we always find more work to do, but it, but it ends up being higher level work every time, right. It ends up being higher level work.

Yes. And one thing we do have to think about is there are humans behind all this technology. So, you know, at the end of the day, humans have to be there and the technology is just augmenting and enhancing and speeding up the work.

Absolutely. Yes. So, um, what are, let's talk about, um, we've talked a lot about the pros and how amazing the technology is, but what about some of the cons and the drawbacks?

Um, we, there's quite a few that I've, um, been hearing about, like, um, bias and ethical concerns. Of course. Um, AI hallucinations is another one.

So can we go into some of these issues? I think, I think there are tons and tons of cons. And just like with the other technological revolutions we've had, um, we, uh, have to be via very aware of them.

And the technology is already getting ahead. And you touched on a couple, um, I think that the desire to use, um, to use AI to make decisions, uh, is, is pretty scary. Um, you know, we see things with facial recognition and, and people being, uh, you know, accused of crimes because an algorithm said, well, I think that's the same face as as, as this face, right?

And, and that will be happening all over the place. You see, you know, which credit applications get, get, uh, accepted and which don't, ML is, is playing a part in there. And so I think that where AI is used, there are gonna be a lot of ethical issues where the AI is trained, what data it is trained on is already a huge issue.

And are you training on data that you had the ability to do? You know, you see this in the generative AI world for art. You know, I can say, give me a painting of me and the style of whatever artist I want to, and maybe that artist would never have wanted to do that.

Was it fair to train the AI on on that? Um, and uh, and there, there are countless others. There's a good quote from IBM in the 1970s that says, uh, a machine should never be, uh, should, should never make an important decision because a machine can never be held accountable.

And I think that's a really, really interesting proposition. So where these things get trained, you know, what data they are trained on, does the data that they're trained on have bias? Cause if so, then, then we know the system will have bias.

There have been some really embarrassing, uh, accounts already, um, and, and how it gets used. So without a doubt, as, as much as I am optimistic about this technology, I also think that, um, as with all technology, we have to be very careful and think about these, these ethical issues. There are places to use it safely.

Absolutely. Um, and there are, there are places to use it where I, it doesn't bother me ethically, but we need to, we need an active, uh, part of our community who is thinking about these issues, writing about these issues, establishing standards on these issues so that we can proceed safely. Absolutely.

So last question. Where do you see the world in five years from now? Ooh, that's a, that's a, that's a great question.

Um, I think I, I'm just, I'm gonna go optimistic on this one. I think education is gonna be transformed and I think education, uh, is gonna be transformed in a lot of different ways, technical and non-technical. You know, as I said, how we, how we learn new things.

Al already we're thinking about, um, how we can, uh, help. As you know, I just gave those examples of, you know, starting to learn to code. I I, you know, it's gonna be fantastic when kids can, can move to a, a truly interactive process.

We've already seen, I have three kids and I've already seen technology helping them a lot more, but when they can ask a question, they're using apps to learn things like math pretty frequently now. Um, when those become interactive, it's gonna be amazing. I think how we, how we, uh, think about writing is gonna change.

Just like, how will you think about spelling changed in the beginning? Like, is it okay to use a spell checker on your paper? Right.

And so I'm, I'm really excited to see, I think education five years from now is gonna look extremely different than it does right now. And, and the end result is we'll be learning faster, right? The end result is education is gonna be improved and we're gonna learn, there's be certain things, and I can't predict exactly which ones, but there are some things that we're gonna learn faster than ever before.

And, you know, when we learn faster, that means we can learn more. You know, in our 12 years of education, just primary education, uh, we'll get farther than we ever thought. And so, uh, I'm very excited about that.

I think some industries like customer service are gonna get improved so much, right? When you can deal with a, a bot that actually kind of knows what it's doing and can actually help you, you know, change your cable subscription or cancel your cell phone subscription or do that and oh, it's gonna be amazing. Um, and so there's a lot of industries are, are really gonna fundamentally change very, very quickly and, um, you know, make our lives a little bit smoother.

Worried about the ethics. I'm worried about a lot of the things, but, but, uh, i, I do know positives will come. Yeah, I agree.

I think the future is looking really bright and I'm excited to watch it unfold. So thank you so much for coming on today and speaking with us and giving your insights. Uh, glad to have the opportunity.

Thanks for having me.