Utover ledetekster: Slik bruker du kunstig intelligens til smartere prosjektledelse

Les transkripsjonen:

Vi prøver ut transkribering av podkastene våre ved hjelp av et program. Tilgi eventuelle skrivefeil, siden boten ikke er korrekt 100 % av tiden.

Galen Low: You're about 11 months into your journey with generative AI, and you've arrived at three conclusions.

Number one, GenAI's capabilities truly are unprecedented. Your uncle Rastin genuinely thinks it's sorcery.

Number two, Generative AI is a global phenomenon. People aren't going to stop talking about it anytime soon — not at your dinner parties, and definitely not in your LinkedIn feed.

And number three, even amidst all the awe and fervor around generative AI, so far, all you're using it for is note taking and a little bit of meal planning.

If you're a project person who feels like you're stuck in a rut with your AI chatbots, or know someone who is, this episode is for you. We're going to be diving into how to shape your mindset and reflexes around generative AI to avoid "doing GenAI for GenAI's sake". We're going to be talking about what the future of AI holds beyond prompt engineering. And we're going to be exploring how expanding your understanding of AI and how AI projects work might actually open up a whole new branch of your career that you never knew existed. Ready to dive in?

Hey folks, thanks for tuning in. My name is Galen Low with The Digital Project Manager. We are a community of digital professionals on a mission to help each other get skilled, get confident, and get connected so that we can amplify the value of project management in a digital world. If you want to hear more about that, head on over to thedpm.com/membership.

Okay, today we are talking about, surprise, generative AI, but also prompt engineering, and whether professionals like us project managers may be looking at Bruce Lee's proverbial finger while it's pointing at the moon. And I've brought in the big guns to tackle questions about what GenAI has on offer for the craft of project management beyond just chat-based interfaces and prompting.

So with me today is Kathleen Walch, Director of AI Engagement and Learning at the Project Management Institute and proven AI thought leader and educator.

Kathleen, thanks for joining me here today.

Kathleen Walch: Yeah, thanks so much for having me. I'm really looking forward to this discussion.

Galen Low: I am so happy you came back on the show. So Kathleen's been on the show before with Ron Schmelzer and talking about AI. Gosh, probably two years ago, maybe two years ago?

Kathleen Walch: Maybe. I know, time flies, right?

Galen Low: It's moving so fast. And I was thinking about that conversation. I was like, so much has changed. So much has changed. It was almost fringe back then.

It was like AI and project management. Yeah. Okay. Now it's just it's mainstream. It's baked in. Everyone is. Just immersing themselves in it. You can't get away from it. And I'm excited to have your expertise on the show today.

For folks who don't know, you've been steeping in the AI world for a while now, and you've seen our current understanding of AI in the professional context evolve over like the past seven or eight years. And so as fast as some of this change has seemed to us newcomers to AI, I'm just wondering, do you feel that most professionals are maybe getting stuck on everyday prompting without seeing the bigger picture of what the technology is capable of?

Kathleen Walch: Yeah, that's a great question because you're right, I have been in this space.

I say I've been in AI since before GenAI made it cool, right? I know. And I always call it the oldest, newest technology because the term was officially coined in 1956. So it's 70 plus years old, yet it feels so new. And why is that? We've been into two previous AI winners, which is a period of decline in investment, decline in popularity.

Big reasons for that is we over promise and under deliver on what the technology can do. So we really need to understand AI as a tool. And that it's not good at everything, but it is good at certain things, and so make sure that you're using it in that way. And now with generative AI, what's made it so exciting is it's really put it in the hands of everybody.

Seven, eight years ago, we still were using AI. On a pretty regular basis, it just didn't feel like it. Where we would have predictive text with our emails or we would have spam filters, right? And that would be using AI or with GPS and driving around, Waze or Google Maps that would help us with route optimization.

But it didn't really feel like AI because it just was in an application that we were already using. And then generative AI, ChatGPT in particular, right? Because it was the first one out. Put it in the hands of everybody. And so now I was able, it's what we call augmented intelligence, where it's not replacing the human, but helping you do your job better.

And you could feel it every day. So it would help you write a better email or help you brainstorm, or it could help you with translation. It could help you create images for your PowerPoint presentation. And it really felt that collaboration and you saw the direct benefit from it. That's where, we've seen the whole world go, and it's been really wonderful, but then at the same time, people still need to understand that it's a tool, and they still need to understand when you should and when you shouldn't use AI.

Galen Low: I love that sort of AI in the background, and now AI is like a person, that you like interact with every day that everyone has access to. That is the new popular kid. I like what you said about GenAI made AI popular, but I love that whole notion of some of the winters, right?

The AI winter is because we over promised and under delivered and granted technology was not at a certain point that it may have needed to be to deliver on those promises. Now, maybe it is. We're talking about yes, ChatGPT, and yes, all of the sort of big LLMs that are running the show right now, but we're also talking about, here in February 2025, we're talking about DeepSeek.

We're talking about, technology that is mind blowing that maybe doesn't need all the technology necessarily that it's been touted as needing. I'm getting a little off topic here, but I like the sort of prospect of what it brings. I like that people are talking about it. They're embracing it. But, that can be a complicated thing where everybody is in there. Everybody has an opinion. Everybody thinks they know what it does and everybody thinks it's for everything. And maybe it's not.

If I was to put a project management lens on it, like what are some of your favorite use cases for GenAI chat based interfaces for project management?

Kathleen Walch: Yeah, that's a great question.

And, at Project Management Institute we have a lot of learning courses that are free for members or, pretty cheap for non members. And we go over a bunch of different use cases. And what I like to say is I always break it down and I say, put down a list of all of the different pain points that you have or areas that you need help or areas that you can see improvement.

And then figure out which one of those can be easily done with a generative AI system. So when we think about project managers, we always think about meeting minutes, right? It's like the stereotypical example that we've been given. So if that's a pain point for you, then how do you fix that? How do you get help with AI?

And there's a lot of tools out there that can do that already. But then what I always say too is, yes, it's great to have something that can help you once. So we talk about getting help with a project charter. Okay, that's wonderful. But how many times do you do that during a project? Probably once, right? I mean, you're not revisiting a project charter every week.

What's something that's going to help you on a regular basis? A daily, a weekly basis. Maybe that's stakeholder engagement or better communication, right? How do I craft emails or documentation for different levels? Sometimes I need to have a high level executive summary or I need to tailor it towards stakeholders, or I need to tailor it towards internal versus external customers.

Figure out what Those pain points are and work to address them because that's where you're going to see that real incremental improvement. And then also I say, just practice, right? Because you only get better with time. There really is very low failure when it comes to prompt engineering, because just redo it.

But then this also brings in what PMI calls power skills, right? Which are soft skills. And critical thinking, collaboration, communication, I always say, how do you use GenAI to help you with your power skills and how do your power skills help you be better with prompting? For example, with communication it can help you be a better communicator, right?

Because I said it can help maybe write emails or put it in different tones or shorten things, summarize stuff. But then how does your communication skills help you be a better prompter? Maybe you need to tweak your prompt, or maybe you need to change the length of the prompt, edit it over time.

So I really like to see the kind of two sides of that coin, how it can help with your power skills and how power skills can help you be better at prompt engineering.

Galen Low: I wanna come back to that later on. But first I wondered if maybe we could zoom out a bit because I have the context, not all of our listeners do.

But during your time as a Managing Partner at Cognilytica, you co-developed the CPMAI certification, which is a project management framework for specialists. And I've written this and tell me if I'm right or wrong, but my interpretation was it was built for specialists like data scientists and analysts working on AI projects.

And then more recently, Cognolitica joined forces with the Project Management Institute, making CPMAI like an official part of the PMI's rather prestigious portfolio of project management certifications. And just cause we were talking about courses and PMI and learning, I was wondering could you just talk to me a bit about who the certification is for nowadays and how is it different than say, taking just like the prompt engineering courses, the, the free one on PMI for members.

Or something on Udemy, what makes the CPMAI certification important and different?

Kathleen Walch: Yeah, that's a great question. And the CPMAI certification, I always like to talk about this as two sides of the coin, right? 95 percent of the conversation is focused on how can I use AI tools to help me do my job better, right?

So we talk about, there's a ton of different tools out there. People always ask me, what's the best tool? And I go, it depends on what you're trying to do, right? I mean, new tools come out literally every single day. So whatever it is that you're trying to do, I'm sure that there's a tool out there and you need to understand how to use it and, how to get better at it.

But that's about where 95 percent of the conversations are. And that's where a lot of the e learning that PMI offers comes in. So we have an overview of generative AI, data landscape for AI. We have our prompt engineering course. We have a applications of AI course as well. So it talks about how to put all of these different tools and applications into use as a project manager and a project professional.

And that's wonderful, but that helps you do your job better. Then we have to say as a project manager or a project professional, or these project adjacent that you talked about a data scientist, a data engineer, an AIML engineer, you think about whatever that title is, you're being tasked with running and managing an AI project.

And we have to understand that. AI projects are data projects, and so you have to use data centric methodologies, and you can't, run it in a traditional software application development style, or you're going to quickly realize that's not the right approach and your project has a higher rate of failure.

A number of years ago now, that's when we developed the CPMAI methodology, because organizations were coming to us and saying, and this was, again, seven or eight years ago, well before generative AI was around, we needed to build these systems from scratch, figure out what algorithm we needed, have all of our data requirements.

A lot of it is still the same, even if we are using one of these, but they said, where do we begin? And so we looked out there, and there really was no step by step approach. So we created CPMAI methodology with a large bank and a large government institution to have that step by step approach. And now back in September of 2024, we officially joined PMI, they acquired Cognilytica.

So now CPMAI is an official PMI certification and it's so incredibly wonderful. It always has been for project managers, project professionals, product managers as well, but then going beyond that, those project adjacent folks. I always give this example, my husband's a software engineer, has been for about two decades now.

He sometimes needs to run and manage projects, and if he's being tasked with running and managing an AI project, He doesn't identify as a project manager, nor would he ever identify as a project manager, right? Because being a software engineer is how he identifies, but he's put in that position. And more and more, we're seeing that, especially with AI projects, people who don't traditionally identify as a project manager are being put in that project management role.

And so this certification really is for everybody who fits in those different categories.

Galen Low: I think it's really important what you said. I didn't know that you started this out with a government institution and a bank. And when you're talking about the difference between a software project and a data project, and you think about data that a bank might have or the government, serious business, and it actually it brings it into sharp relief for me, the idea that, yeah, you can't necessarily run it like a regular project. There's things to take into account and consideration. And I agree with you now that like more and more, you want the whole team to have that sensibility because we've seen it gone wrong ethically.

We've seen it gone wrong in terms of just printing too fast towards a goal that we don't really know with a technology that's moving very fast and culturally is shifting in terms of our, people's adoptions and their feelings and their anxiety around it. I really like that notion that it's an understanding of how to make a data project that involves AI or machine learning go well, because sometimes the regular approach, the software development approach might not yield the right result.

You might hit that dead end. You might hit these roadblocks that you could probably avoid if you were thinking about it, ahead of time.

Kathleen Walch: Yeah, and I always like to frame it to where with software development, code is the most important part, right? Like you would never give away your code because that is your most important part.

But with an AI project, data is your most important part. So you would never give that away. And code is a really small part of it and not a super critical part of it, right? It's your data. It's the data that's unique. It's the data that's going to make or break this project, right? We say garbage in is garbage out.

And I think that helps frame it too. Especially if you have somebody who's project adjacent, not a traditional project manager where they need to understand, okay, that's how I need to shift in my mindset, or it's not about the code, it's about the data. And that's why I need a data centric methodology.

Galen Low: There's so many places I want to go with that, but I'll try and stay on top of it. But I think that's a wonderful slash mind blowing point that. Like data is the important bit coming out of a data project. Code is almost secondary. It's off to the side, it's a weird idea, at least for me. Coming from like a digital space, but I really like it.

I thought maybe I'd shift it back around to career and certifications because I know a lot of folks in my community, yes, the PMP is a great credential and it has a lot of weight in the industry. And I think a lot of people look to it after they've been doing a job a while as a project manager or even being project adjacent and just accruing a bunch of project management hours.

PMP is like the pinnacle thing to get to tell people on your CV and in your, profile and in job interviews that, you're serious. Like you play at this level, you're in demand and you should probably get paid more. So now that CPMAI is in the mix as well. Like I'm wondering if employers in the AI-enabled software space, or actually, I guess, maybe anywhere are they looking for the CPMAI certification already?

Or in your mind, is this more about sort of like practical on the job skills for somebody who's already working on AI and data projects? Is it something that can make a project manager stand out? Is it for that?

Kathleen Walch: Yes. Oh, absolutely. It is. And we see that, you're right. The PMP is the pinnacle, right?

It is our gold standard certification at PMI. And so we're working on making other certifications gold standards as well. And so what does that mean, right? And it's, incredibly robust. It goes through a lot of different steps. There's a lot of different requirements to make it a gold standard.

And so we are working to get CPMAI to become a gold standard as well, and it does let employers know that you know how to run and manage AI projects. You've been level set on terminology. You've been level set on what AI can and cannot do. You've been level set on the, six phases of CPMAI, and you know how to run your projects like a data project and following this step by step approach.

Because far too often, we've seen organizations Not have a plan, and that's for a number of different reasons, this industry really is moving fast, people feel FOMO fear of missing out, they are like, our competition is doing it, we need to do it, we're just gonna move forward, and it's okay, we can move forward, but let's have a plan.

And so that's what CPMAI does, right? And it's this step by step approach. So we start with phase one, business understanding. What problem are we trying to solve? And I know this sounds so simple, but so many projects skip that step. And so it's, what problem are we trying to solve? Is AI the right solution?

And then we have an AI go no go that we need to go through, that walks through data feasibility, business feasibility, and implementation feasibility. And we think about these like traffic lights. So if they're all green, your chances of project success are pretty high. If they're yellow or red, Your chances of failure are increasing, and doesn't mean you can't move forward, but we say proceed with caution, and know that there might be some stumbling blocks, or know that your project might not succeed as you think.

One of them is what is the ROI, the return on investment of your project, and a lot of organizations don't necessarily think about that up front, nor maybe measure it at all, and so it doesn't need to be a financial return, but you have to have some return, and if you're not measuring that, then the project itself could be a success, right?

Technically it did what it was supposed to, but at a super high cost, that's actually negative, right? You had a negative return on investment. And if you're not measuring that, you have no idea. And so you go, Oh, this was successful, but why are we suddenly, in the red? And it's because your return on investment was not a positive return on investment.

And that's what we've seen far too often. And so CPMAI helps with that. And it does signal to employers that you have the skill set to run and manage an AI project, that you have gone through this very robust training and certification, and you're certified with a PMI certification.

Galen Low: It's funny because I'm like, for folks listening, I keep Moving my mouse, looking for the like clap reaction, like in Zoom or Google Meet, and I'm like yes.

And by the way, if you're somebody who is interested in the CPMAI and wanted to leverage like the CPMAI credential in a job interview to someone who didn't know what the CPMAI certification was, just take that clip of Kathleen just now, and then just transcribe it. Use it in your interview, because honestly, that was such a crisp and clear explanation of like why this matters and why it is a bit different.

And to your point, some of the things are like, yeah, this seems like maybe, quote unquote common sense or, other frameworks, do consider this. But when you're talking about like data, the speed of technology and that urge for businesses to just right now, everyone's just running at it yet. Like you said, like trying to stay competitive, they might not even know why they need AI to be a part of their product or their organization or their solutions. And I just liked that there is a bit of a framework around that.

I'm curious because project managers, we love our frameworks. We love our methodologies. We love our certifications. And then we get in big fights about them. You know what I mean? Like we're guilty sometimes of creating camps and factions and whether or not they were intended to be rivals, we still put them against one another. And then you go into an organization, they're like, we run projects this way.

And you're like, I'd like to run it this way. And then there's a big fist fight. Do you find that to be the case with folks who are going through this program? They go out, they're like, Hey, now I know how to run a sort of data ML, AI-oriented project. But my team doesn't want to do these things. I'm trying to like preach to them and they're like, no, we do it this way.

And this is how we're going to do it. Do you find it's an uphill battle getting people and teams and organizations like on side with the phases and the steps within the framework?

Kathleen Walch: So folks that take it obviously see the light, right? And they go, wow, this is wonderful. And then they do need to bring that back to their organization where this is different.

I get it, right? It's predictive, waterfall, hybrid, agile all these different terms. How do you run a project? Usually it ends up being hybrid, agile is a real term for a reason. I know. When you run your project, your AI project, like a software application project, you'll realize that it doesn't work and you have to bring in a different set of skills and a different step by step approach.

And you can run it in an agile way. So it should be small iterative sprints. We shouldn't be running it in that, predictive waterfall way. Folks realize they look to this because they've done it and it's been failing and they need to do something else. So a lot of people who find this methodology Have been running into those problems and understand that they need to get on board with something because it's not working the other ways that they're doing it.

So that's what we've seen. Also, some folks are, they get ahead of it and they go, I want to move into this space. I need to understand this and we give a really comprehensive training. So it's for everybody, right? You do not need to have had a project under your belt to get started. You currently don't have to have any requirements.

I know PMP has some pretty robust requirements and we're working towards that with our gold standard certification. But for now, you don't need requirements, which is nice because anybody can take it. And it really walks you through everything. So it first level sets on terminology. And folks think that they know terms and they're like, Oh, I don't need this.

And then they get into it and they're like, Wow, I didn't realize what I didn't know. And this really was a great level set. And that's what we say for teams too, right? You want the team to get on board. Which is the same reason that you take a PMP, and that you hire PMP, because it's that terminology that you want, right?

It's everybody's on the same page. Everybody uses the same terminology. Everybody has the same basic level of understanding. When it comes to CPMAI, it's no different.

Galen Low: I really like that. And actually, I love your framing on it because I realized that I probably made it seem like it's a methodology, but if I'm understanding it correctly, it just layers on in terms of, you may have a certain way of working, these are considerations and steps that you can weave into that.

Kathleen Walch: Yeah, we call it a methodology, but then people get tripped up on that term. And so we always say, look, don't get tripped up on terminology. You just need to follow something for success. And so you can call it a framework. You can call it a step by step approach. You can call it a methodology, but understand that it is six phases.

It's iterative. You can go back a phase or two or three if you need. So we start with business understanding. Then we go to data understanding, so what data do we need, the sources of that data, is it internal, is it external, do we have access to that data? Then we know we need to clean our data, so data cleaning is the next step.

Unfortunately, data is never clean and nice, and it's always messy, especially unstructured data. Then we get to our model development. So now we're actually, doing the fun stuff. I air quote that, which is where people usually start and skip those super critical beginning steps. Then we have to test it, make sure that it's performing as expected, which also is a really critical step and sometimes people skip and then things go wonky and they wonder why it's hallucinating and giving terrible results or not performing as expected.

And then we operationalize it, right? We put it into the real world. And so we should be doing these in small iterative phases and steps, each iteration of CPMAI should take about two weeks, right? Like a sprint. It should not be taking seven months or twelve months. That's more of that waterfall, predictive approach, and think about how much the world changes in seven months.

But for a number of reasons, this is what happens, and usually because people get tripped up at the data phase, right? Data understanding, where they don't have access to that data, and so it takes weeks and months to get access to that data. And we say, okay think big, start small, and iterate often.

So start smaller, pick a smaller data set, or, really control that scope and pare it down a lot and say, what's the smallest thing that I can do? That will provide an incremental value and provide that positive ROI. Show wins. And then we can continue to move forward, right? And just iterate more in the next phase, the next step.

Because far too often we see that it's taking way too long and then people give up on the project in phase two because they just couldn't get access to the data.

Galen Low: It's like the Project Winter.

Kathleen Walch: Yeah. A lot of good things.

Galen Low: Just get disillusioned with it and then just leave it alone. I was going to ask you actually what Wonky looks like, but I think you framed it really well in terms of like hallucinations and when you start thinking about, even small data sets are pretty big data sets, so you can get really good at yourself and then, your solution is not performing the way you want it to. And you're wondering why, and you have to unwind this entire ball of yarn to find what was somewhere in the middle there at the beginning that.

Set this astray, probably data cleanliness, right?

Kathleen Walch: Probably, access to it, how clean it is, and then how much of it you have as well. Because some people think more data, the better, right? And that's not always the case because we can be training on noise. Data is not free, right? There's a cost to cleaning it. There's a cost to processing it. So sometimes more is not better.

Galen Low: That's interesting. It's funny. It's like I started this out, we will answer the mail on this, but I started this out with what's beyond prompt engineering, but what I find fascinating about this conversation is that, I think there's a whole world that many project managers haven't seen themselves in, they might be software and like in the software space or the digital space or the tech sector, IT, they might not be in any of those sectors.

And they're like, is project management still going to exist? We'll get into it in a little bit. But there's this whole world of like data projects, the ML projects that, there's a lot of projects going on at this very minute. And yeah, sometimes they're being led by someone who's project adjacent, right?

Someone who's not necessarily self identifying as a project manager, but there is this opportunity, like there's a lot going on there that at least folks in my community, I don't know that we all know a lot about it. It's actually inspiring.

Kathleen Walch: Yeah, we say, education is cheap, right? I mean, the cost of failure is high.

And to learn the CPMAI methodology, to get CPMAI certified is fairly cheap in comparison. We always encourage people, we say this is a yes and, right? Have your PMP and your CPMAI. Follow and learn different methodologies and learn CPMAI. This is not an either or, and organizations always do this, right?

They learn something and then they adapt it for themselves. And as long as you're, following that high level, step by step approach, then adapt it for your own organization, but learn it.

Galen Low: I love that. We opened on the topic of prompting. I was thinking in my head, right? Those use cases that involve chat based interfaces, like ChatGPT, your Gemini, your Copilot, your Claude, your Perplexity, DeepSeek, and others. But there's a big world beyond that.

What's next for AI in our professional lives as project managers? And how can someone like, a project manager like me, how can we keep up?

Kathleen Walch: Yeah, that's a great question because, there's been that saying out there that AI is not going to replace your job, but someone that knows AI is going to replace your job.

And so it really is here to stay. We hope that we've crossed the chasm. We're not going to be falling into another AI winter. But how do you use it? I always like to think about that idea of augmented intelligence. So how do I use it to not replace me, but do whatever it is better? So you can replace a task or a, certain role, but not replace me as the human.

And I know a lot of project professionals are concerned about that. But if a lot of projects are moving towards AI projects, then we're just going to continue to need project managers, right? And folks with these skillsets. So how do you learn? I think it goes back to what I said before.

Put a list together of everything that is annoying to you. Because that's different for everybody. And so to make sure that you don't replace things that you enjoy, and then keep all the things that you hate, because then you're pretty unsatisfied. So think about all the things that you enjoy, and put them maybe in one column, and all the things that you don't enjoy, and put them in another column.

And then look towards communities. PMI has a lot of, learning resources that you can go to. Or maybe internally, and I always say, advocate for internal groups where, especially if you have prompts, they should not be proprietary, right? We should be sharing prompts, we should be learning, we should be collaborating, so have a prompt library internally at your organization.

Document which platform you've used it on and when, because we know that your prompts do need to be changed and iterated over time to continue to get those results, right? That's part of our step by step approach, where you need to be testing it, and sometimes it's not going to perform as expected, so sometimes just tweaking one word really makes a big difference.

And if you have that community and you can go back and forth with, it really does help. I know that some organizations, they're leaning into AI and being AI first, right? And they're really AI driven. But if your organization isn't, then figure out how you can start with those iterative, step by step approaches.

Like I said, if you're a PMI member, reach out to the community. We have a lot of chapters as well. Internally, you can have resources, and so whether that's your group, or maybe even look externally outside of your group, so that you can be collaborative, because if you're not learning, you're not growing, and then you're going to just fall farther behind, because you're going to be afraid to use these tools, or you don't know how to use these tools, and so really, don't be afraid to reach out and ask for help.

That's the best way to learn.

Galen Low: I love the sort of prompt repository, and sharing. Sharing is the thing that moves us forward. The technology is going to move fast, people need to move fast too. The best way to do that is to share knowledge.

Kathleen Walch: Sharing is caring.

Galen Low: There you go, yes! I have two questions to wrap this up.

Some of the things that we're talking about, we're like, make that list of annoying things. And, you can chat with your GenAI tool to augment your sort of personal professional life, like your own individual professional life. On the one hand, some folks, me, are you know what, this is great.

It's a fantastic technology. It is fundamentally the way I use it. The tab based interfaces for me, it's, it's natural language processing, to the nth degree from where it was, a few decades ago. And then I'm like, isn't it just a fancy parlor trick where it's languagifying stuff and giving it back to us and we're like, Oh my goodness, this is like a living, breathing, sentient creature.

Whereas actually it's like remixing language. Are we maybe even, I started this with, are we like selling ourselves short by not using enough AI? On the opposite side, is there a risk of us thinking that AI is doing like really magical things and can do anything and not understanding that actually it is good at doing a certain thing? And is that dangerous or is it inspiring?

Kathleen Walch: It's dangerous because it's over promising and under delivering. So a number of years ago now, about in 2019, because AI is such an umbrella term, and because now with generative AI and, these large language models that are out there, people think it can do everything, and are asking it to do everything, and then maybe not getting the results that they want, they go why can't it do this math problem?

Why can't it do this? A human can do this, and it's because we need to understand what it can and what it cannot do. So we said, why don't we break it down one level deeper. And say, what are we trying to accomplish? And then let's see if AI is the right solution to that.

So we looked at hundreds, if not thousands of use cases and broke it down into the seven patterns of AI. So that's hyper personalization, right? Treating each individual as an individual. So we think about this as a marketer's dream, but also it can be with hyper personalized education or healthcare or finance, right?

And this is a really hot topic with education, especially lifelong learning. How do I, beyond that K to 12, tailor education to fit different people's, the way that they learn. Then we have recognition. So this is making sense of unstructured data. And we think about image recognition here, right? But also audio recognition or hand gesture recognition.

And then we have our predictive analytics and decision support, so this is taking past our current data and helping humans make better decisions. We have our predictive analytics, our patterns and anomalies. So that's looking at large amounts of data and being able to spot trends in that data, but then also some of the outliers, so we think about fraud detection.

We have goal driven systems, which is really around reinforcement learning and optimization. We have the autonomous pattern, so that's, the goal of that is to remove the human from the loop. So whenever you're trying to remove the human, that's obviously pretty hard. So we say that is the most difficult pattern of AI.

But what do we have now? We can have hardware or software. So we can have autonomous vehicles, for example, or autonomous delivery bots. But then we can also have autonomous business processes. And so how do we have systems that can Autonomously navigate within our workflows, and we don't need humans in there.

So it's different than automation, right? Because automation is not intelligence. We're just repeating something over and over, which is absolutely incredibly useful, but just not intelligence. So if there's exceptions, you need a human to go in. Can't handle that, do that exception handling. If fields change, the human needs to go in, and say, okay, these fields have changed.

We think about robotic process automation there, but you know, what's agentic AI going to look like, especially, in this year, but then in the coming years, and how can we have more of the, autonomous agents out there working with other agents. So when we think about the seven patterns, it really helps us break down what AI is good at and what AI is not good at and when we should use it and what we shouldn't use it.

And we go through that in phase one. Of business understanding of CPMAI, and I think that really helps too, so if people are trying to say where should I apply AI, and how can I apply AI, if we have our conversational pattern of AI that's where we have humans and machines talking to each other in the language of humans, that's where large language models fall, and AI enabled chatbots, but that's just one pattern of AI, and maybe that we shouldn't have our LLMs try and do, hyper personalized offerings or we can't use an LLM to drive a vehicle.

Galen Low: I love that. And honestly, thank you for that crash course. It actually brings things a lot into perspective. And guess what?

Perfect segue into my last question, which is just, we're talking about things that AI coming in, helping you do your day to day.

Kathleen Walch: Oh, augmented intelligence.

Galen Low: Augmented intelligence. Thank you. But then, you had mentioned autonomous, you had mentioned agentic, everyone's talking about it right now, a world where, yeah, it's just often doing its own thing. Maybe you're bidding, but not in the same way where you're like having a chat and we're like, Hey, could you do this thing for me?

Do you have an opinion on where that goes in terms of the project management world or just the world in general? Are people's anxieties founded around that? Or is it also one of those things where maybe people think it's like, Oh, that's basically AGI. I guess we can skip to the apocalypse. Whereas actually it's just, within these patterns?

Kathleen Walch: If we understand what AI can and cannot do, and we understand we're still in narrow AI. So we're applying one or more of these seven patterns, but we aren't in AGI where we have a system that's human right? And can do everything. Cause we still don't have machine reasoning. So I like to talk about this DIKUW pyramid.

So data is at the bottom, the foundation, but data on its own, you can't do much with, right? We need to apply something to it. So we get our information level, and this is where we have dashboards and, things like that. And then we go to D I K, right? So we're at the knowledge, and that's where machine learning comes into play.

Then we have the U, the understanding, and this is machine reasoning, we still aren't there. And then at the top we have wisdom. So we're far ways off from AGI, it's important to understand that. I know that there's different things in the media about how far off we actually are. Some people think we're maybe a decade away.

Some people think we're a year away. Some people think we're 100 years away. Some people think we're never going to get there, right? I mean, experts in the industry cannot agree on this. But it is important to understand the limitations that we have right now, and again, apply it where it makes sense. Agentic AI is the hot topic of 2025.

It's going to continue to be the hot topic of 2025. At PMI, we have Infinity, which is our, tool, our AI tool. And we are going to have agentic AI capabilities there. But it is important to understand, too, I think, because we don't have a commonly accepted definition of AI, then that means What is agentic AI?

We don't really have a commonly accepted definition of agentic AI in the industry either. So is it actually autonomous? Is it augmented? Is it automated? We're still moving towards that, but it is exciting. I mean, things are changing so incredibly fast. So if people are listening to this in February of 2025, what's it going to look like in February of 2026?

What's it going to look like in June of 2025? Things change. And you're just that one real breakthrough away or that one different platform, right? Earlier you brought up DeepSeek that wasn't around a month ago. How do things continue to evolve and how are project professionals?

Adopting that agentic AI and how do they bring it into their workflow? These are larger conversations, your organization, how are they bringing this in? How are they adopting these tools and technologies? We need to make sure that we're doing it in a trustworthy, ethical and responsible way. I know there's a lot of concerns around that.

Data privacy, right? Governance issues. the trustworthy aspects. Do we trust these systems? Do we have them internally? Can we use external ones? So it's a really exciting time. And these tools are continuing to get better. New ones are coming out every single day, like I had said earlier on the podcast. And so that can get overwhelming.

So it's where do you even start? And you had said, how do you even keep up with it? And I think Learning and doing every day. You have to practice, right? It has to start being that reflex. How do you make it part of your every day? I say never start with a blank page anymore, right? If you're starting with a blank page, you're doing something wrong.

Have it help you with that brainstorming, or have it write a first draft, and then you can edit it, or whatever it is that you're trying to do. But it is that reflex that you need to get used to, right? And how do you do that? Just continue to practice.

Galen Low: I love that. The mindset and the reflex. Kathleen, this was a really inspiring conversation.

I don't think I've taken so many notes in an episode, probably ever. And I mean this in a good way, as in capturing knowledge for myself, not like what to edit in post. I think PMI is lucky to have you. I'm excited about what comes next, and maybe we should have you back February 2026, if not in between to do a comparison on, where we landed after 365 days.

Kathleen Walch: Yeah. I know it's changing so fast. Sometimes people go where are we going to be, a year from now, five years from now? And I'm like, where are we going to be a month from now?

Galen Low: Oh, I love it. I love it. I love it. Where can people find more about the CPMAI certification?

Kathleen Walch: Yeah, so go to PMI.org. You can find it there. Also, I have a podcast, AI Today, where we talk about CPMAI. We're in the middle of a use case series right now, so that is now an official PMI podcast, and I'm really excited about that. I know we're transitioning everything over to the PMI system, as anybody with acquisitions know, it does take a little bit of time, but we're getting there, and I'm really excited about that. You can also find me on LinkedIn, Kathleen Walch, or PMI Cognilytica is on LinkedIn as well.

Galen Low: Amazing. I will also include all those links in the show notes.

Kathleen, thank you so much for joining me today. It has been so much fun.

Kathleen Walch: Yeah, thanks. I always love to talk to you, Galen.

Galen Low: All right folks, there you have it. As always, if you'd like to join the conversation with over a thousand like-minded project management champions, come join our collective. Head on over to thedpm.com/membership to learn more. And if you like what you heard today, please subscribe and stay in touch on thedigitalprojectmanager.com. Until next time, thanks for listening.