How to Apply AI Technology to Your Project – with Paul Boudreau

Introduction

Imagine the insightful change achieved by projects consistently delivered on time, within budget, and with minimal scope changes made possible through integrating artificial intelligence (AI) in project management.

However, substantial challenges arise from misconceptions and inadequate preparation when implementing AI.

This webinar started with a brief AI overview, providing practical guidelines, and showcasing its capabilities in project management. Recognising AI as a disruptive technology to revolutionise project management understanding its potential is crucial for delivering the performance excellence this profession deserves.

Watch the webinar now to gain valuable insights into applying AI to your projects.

Video

Watch the video, download the presentation or audio, or read the full transcript of the webinar below.

About Paul Boudreau

Paul, a project management expert with over 35 years in the tech industry, is a college professor in Ottawa, Canada. He imparts knowledge in an AI in project management course at two European business colleges.

A global leader in AI applied to project management, Paul specialises in machine learning, natural language processing, and genetic algorithms.

Committed to enhancing project performance, he authored three impactful books: “Applying Artificial Intelligence to Project Management” (2019), “How the Project Management Office Can Use Artificial Intelligence to Improve the Bottom Line” (2020), and “The Self-Driving Project: Using Artificial Intelligence to Deliver Project Success” (2021).

Renowned as a speaker, Paul eagerly advocates for the essential role of AI in revolutionising project delivery.

Agile courses

Agile represents an iterative and adaptable approach to project management and team collaboration, prioritising the delivery of high-quality results in dynamic and rapidly changing environments.

Agile underscores regular communication, collaboration, and incremental progress, advocating adaptive planning where requirements and solutions evolve through the collaborative efforts of self-organising teams.

Enrolling in Agile courses can greatly enhance your team’s capabilities, fostering improved collaboration, communication, flexibility, and adaptability to change, ultimately boosting overall productivity. These courses offer valuable insights and techniques for cultivating a collaborative culture, streamlining workflows, and effectively prioritising tasks. Embracing Agile principles enables teams to better address customer needs, achieve faster results, and continually enhance performance.

Agile training

Agile certifications

In-house agile training

Agile consulting

Agile coaching

Agile recruitment

Transcript

Here’s the full transcript of the video.

00:00:01 Sevcan Yasa: I think it safe to say we should start, so just to give everyone a bit of background. I’m Sevcan, I’m the marketing executive for Knowledge Train, Knowledge Train and AgileKRC are in partnership. We have Paul with us today, so thank you so much for joining Paul. Just before we start the webinar is recorded. So, you will receive a link most probably next week with the recording and the PDF version of Paul’s slides. If you do have any questions, please pop them down in the chat and then we’re gonna go over them right at the end.

00:00:38 Sevcan Yasa: Also, if some of you have been in our previous webinars, we usually do have a survey right at the end, but this time we are doing slightly different. So, if you do answer the survey, you will be getting a certification of attendance. So, make sure you stay right to the end. I think that’s all from me. So, Paul, you ready?

00:01:01 Paul Boudreau: I’m ready to go, thank you, all right let’s get going. Everybody, this is going to be exciting. Listen, you’re going to want to stay right to the end. So, I’ve got a lot of really good content, you know, AI applying AI to project management. People are curious about what’s going on and what’s happening. There’s a lot of kinds of misinformation out there. There’s a lot of good information out there. What I do in my presentations, I try to stick to the facts. I do let my opinion slip in every now and then. And I do like to complain and point out people who are misleading people on this. So, let’s get started talk about the agenda. Here we go, what is AI? So, we’re all on the same page. That’s I guess that’s an old expression. How about? So, we all have kind of the same basic foundation of understanding of AI, how we can apply these AI concepts to the project methodology.

00:02:00 Paul Boudreau: We’ll talk about the value of AI have to talk about the importance of data because data is really critical when you’re using these types of tools. We want to talk about vendors, I’m going to give you some vendors because there’s another myth out there that says, oh yeah, nobody is doing AI for project management yet. However, there are a lot of vendors that are actually making sales to clients, and I’ll talk about them. I have a couple of case studies, but I do want to get through this. So, let’s get started.

00:02:31 Paul Boudreau: What is AI and before I start, I have to tell you I’m in Canada. I know this is you were in the UK; you will have to put up with my Canadian accent. Hopefully, hopefully not too bad, but it’s a Canadian accent. AI, what is it? It consists of software, software code. I use a programming language called Python. It’s very common you can use Java, C++, any number of different languages. The logic is based on calculus, so there are calculus equations that go through and build this model based on the data. So, AI is a learning process it learns by creating these data models. So, it uses regression analysis based on the data that you’ve entered. It creates a model of that data and then it uses that as a reference to make a decision.

00:03:27 Paul Boudreau: OK.

00:03:29 Paul Boudreau: By the way, I have an example of this coming up. I’m going to do 2 live demos and I’ve been told never to demo technology live because it doesn’t work so we will cross our fingers and hope for the best. Learning generally the three main learning types are supervised, unsupervised or reinforcement learning. Supervised learning with labelled data sets, I have an example coming up. Unsupervised learning you have data that has no labels, but you can do groupings or clusters or classification and reinforcement learning like learning to ride a bicycle, and I have an example of that coming up.

00:04:07 Paul Boudreau: So, what is AI? This is not AI. OK, when I do this in United States, everybody’s yelling Arnold and laughing and all that. This is what I call the Hollywood myth and the Hollywood myth says that one day robots will become alive and decide to, I don’t know, go into the kitchen, and make a sandwich. Wait a minute, even better, one day, robots will become alive and decide to kill people and take over the world. That’s not AI, that’s called entertainment, and there’s a lot of great. There’s a great show in the UK, I think it’s called humans with Z on the end where this, I mean look like a 14-year-old girl has a laptop and puts in three lines of code and all of a sudden, the robots become alive. OK, it’s great entertainment but this is AI, its calculus equations, so my expression is AI is about math not myth, and you are quite welcome to use that. I do not have copyright on it, feel free to tell all of your friends.

00:05:17 Paul Boudreau: There’s two main components of AI that I think are the most important, the most relevant for project management. One is machine learning and the other is natural language processing. Machine learning the programme learns by building an example, and I have examples coming up. Natural language processing I have a live demo on that too. I hope so if it works, I’m going to talk about document analysis, sentiment analysis, and the virtual assistant.

00:05:47 Paul Boudreau: So first of all, Supervised learning uses labelled data sets. So, what we have here is an image. We have an image of a cat image of a dog, and we take these images labelled images, feed them into an algorithm. This is what a cat looks like, this is what a dog looks like. Feed in 100 images, 1000 images and the algorithm builds a model of what a cat looks like and what a dog looks like. And then you give it an image that’s not labelled and say, well, what is this?

00:06:21 Paul Boudreau: And it was the year 2005 that a computer algorithm was more accurate than a human at determining what that image was, the cat or a dog. Now this is used in the medical field right now. I guess the greatest non project management area it’s used is the medical field. Two areas, one for pneumonia for children under two years old. The New York City Hospital is using an AI programme that was developed specially for them. The radiologist or clinician can’t detect from the X-ray they can’t really detect whether a child under two has pneumonia, but the AI tool they’re using can detect it quite accurately.

00:07:04 Paul Boudreau: And the other one is MRI brain scans, this was out of California University of Southern California did a study where they fed MRI brain scans to detect, you know, whether people had aneurysms or tumours in their brain. You know, that colourful image, the bird’s eye view of the top of the human skull, where it’s all colours. And it’s very important because the decision is whether they’re going to give you medication or get a saw and cut open your head, the radiologist, the clinician in that case, was accurate about 90 to 91% of the time. And so, they fed in thousands of images that they had labelled.

00:07:44 Paul Boudreau: Yes, this is a tumour, no this doesn’t a tumour and the AI programme brings the accurate accuracy rate up to 97%. So already we know that AI is very effective in doing some things that’s helping out humanity, right?

00:08:01 Paul Boudreau: So, how are we going to use this in project management. Well, I developed something I call a predictor tool. It’s a three-layer neural network and neural network is something that’s popular for doing machine learning based on 87 factors. I had my research students go out and do a survey of project managers that don’t know if any of you saw it. You know what factors are most important for project success, then the model predicts success, or failure of a new project. So, let’s if I can share my screen.

00:08:42 Paul Boudreau: And there it is, let’s see. All right so, I hope you can see this Sevcan, can you see my AI based predictor tool?

00:08:50 Sevcan Yasa: Yep, we can.

00:08:51 Paul Boudreau: OK, great. So, this is how it works, there’s three it’s a column with three nodes on the left. This is for input now I know there’s only three here, but I’m going to input 87 pieces of data. Three columns in the middle they represent the neurons in the human brain. This is not a human brain, it’s software that is meant to represent what happens in the human brain and the right-hand side is an output which is a prediction. Train this model I’m going to upload some data. Where’s my labelled data set? I use Excel spreadsheets cause its easy.

00:09:30 Paul Boudreau: So now what I have is I have historical projects in the first column you can see I have some historical projects. These are projects that are complete. OK, and then I have characteristics of those projects, you know, did they have a scope statement this can be defined, right, more clearly WBS, logging requirements, change process, so these all have extensive definitions behind them. And then what would happen is the students that I had do this research went through and if it had whatever this definition was, they put a 1 for a yes, and if it didn’t have it, they put a 0 because it did not have it in the documentation.

00:10:11 Paul Boudreau: This is supervised learning, so we have a label. The label is, was this project successful or not. One if it was successful, a zero if it was not successful, how do you define success? You can define it however you want. Its user defined as long as you’re consistent. We defined it as delivered the project scope, no more than 5% over budget or 5% late to schedule and that’s how we classified all these projects and we’re going to train the model. Comes back with a validation score that means that you put everything in properly.

00:10:46 Paul Boudreau: It gives me and that doesn’t always come up with 100%, but it says this model, the training data and the test data match the model. So, I’m going to call it Paul’s model 1, it is a demo. I’m going to save that model. OK, so I’ve saved that correlation of data into a reference model, and now I’m going to take a couple of projects.

00:11:08 Paul Boudreau: These are projects, so I have project A and project B. They haven’t started yet but what I want to do is compare them so based on the documentation, based on the characteristics, what characteristics do they have and what characteristics do they not? OK and I’m going to compare them to that model of success that I just developed. And I come up with a prediction, so project A comes up at 99, oh it is pretty good. So, project A is predicted to be quite successful and Project B 26% usually, it’s a bit higher.

00:11:47 Paul Boudreau: Not too bad. So, what do you do with this?

00:11:53 Paul Boudreau: Well, I would suggest if your organisation has a project selection and screening process, maybe you want to make this one of the criteria. You have OK, I want a financial value to the organisation ROI, payback period, strategic alignment to what the organisation goals are and then what is the probability this project will be successful?

00:12:18 Paul Boudreau: “Because I don’t know about you, but I would rather implement a project that looks like it’s going to be successful than one that looks like it’s going to be an absolute disaster”. OK, that’s my project management stance anyway. Now what I’ve done is I’ve shown you as the project or before the project begins. We have a template for as the project is being executed. How do you keep this project on track? And the example I use is you’re in a video game called Project Manager. Long hallway in front of you right. You take a step forward, you turn to a door on the right-hand side, you put your hand on that doorknob before you open the door run the predictor tool.

00:13:03 Paul Boudreau: It says if you open this door, 44% probability of success don’t go there. Step back, walk down the hall three steps. Turn to your left put your hand on the doorknob there before you open the door. Run the predictor tool it says you now have a 95% probability of success alright. What’s happening here? What are we doing here? We now have an AI based tool that’s helping and supporting a project manager. Make good decisions as the project is being executed.

00:13:43 Paul Boudreau: Isn’t that amazing? And for you agile users, because I know that Knowledge Train does some agile work. Let’s do an agile prediction. I’ll just throw this in and hope I don’t let it at time. Where’s my agile? Where’s my agile stuff go?

00:14:01 Paul Boudreau: There it is right there.

00:14:03 Paul Boudreau: So, I have sprints and I have characteristics of each sprint, right? And I have these labelled there’s a success value. Was this Sprint successful or not?

00:14:14 Paul Boudreau: OK, you don’t need to have every characteristic in a Sprint to make it successful. What we’re doing is building a model for your organisation, for your project, for your sprints of what it takes to be successful. So, I’m going to train this model 90% match on this. I can save it call it Sprint Model 1. I’ll save that as a demo. Save the model, and now we’re going to prediction and let’s see if I can find the sprints that I want to predict. Where’s my sprints?

00:14:48 Paul Boudreau: So, there I’ve got 3 sprints that I’m about to execute or deliver and I want to know compared to the reference of other sprints that have been successful, will these sprints be successful, and it comes up and gives them some values? So, what do you do if it’s 49%? What do you do? Well, what I would do is I’d say, well hey, wait a minute, maybe this strategy we have for implementing this this Sprint, uh, whatever the strategy is, doesn’t look like it’s going to work.

00:15:18 Paul Boudreau: Let’s go rethink our strategy about what we should do and then come back and try it again. Run the predictor tool again. This predictor tool, by the way, is run it on the AWS Amazon Web Services I offer it. It’s an education and training tool that I use, but I do offer it for free. Yeah, I know you just came for a for a small training session. All of a sudden, I’m giving you free stuff. If you send me an e-mail, my information will be at the end. I can have one of my students set up an account for you and if you want to go through and do some practice with it.

00:15:53 Paul Boudreau: You can help yourself to do that. OK, all right that is. Let’s stop sharing this and go back to the slides. Stop sharing, I’ll go back here.

00:16:09 Paul Boudreau: And if anybody has any questions about that, you can throw them in the chat, share screen.

00:16:24 Paul Boudreau: So hopefully we’re back to the supervised learning screen. Project success we labelled person for failure, so that was the demo. Now there are other predictor tool templates. I’ve had students work on this. We obviously have project prediction. We have Sprint prediction, we have risk. What risks are most likely to cause an impact to your project? Stakeholder management one is fascinating. It has all the characteristics of different stakeholders and whether during the process of your project they caused an issue that caught for that had an impact on your project caused it to fail or caused it to get into trouble.

00:17:03 Paul Boudreau: Resource schedule budget, now I have these characteristics in these templates. They’re all customizable, you don’t have to use what I put in your definition can be whatever the definition that you want, OK. So that is supervised learning using a prediction tool to predict various elements. Second one is unsupervised learning, so unsupervised learning uses characteristics. OK, they’re called features in machine learning language. And it looks for patterns. OK, so there’s no label, but based on, typically when they people first start learning this, they use fruit. So, based on size and colour and weight, you can group things together, right?

00:17:46 Paul Boudreau: How do we use that in project management? We do something called clustering and I’ve done this. I have a programme on my laptop in Python that does this. I just take all the characteristics of tasks and I throw them in and put them into groups and the purpose for this is what if you discovered that keep percent of your tasks on the project with high complexity?

00:18:07 Paul Boudreau: For me, what I do is I say, well wait a minute if 80% of my task are high complexity. I need to go check my resources. Do they have the skills? Do they have the technical ability to complete these tasks successfully? Or maybe I need to give them more training? Or maybe I need to bring in an expert to do some coaching and mentoring? This can also be used for risks. I’ve used it on risks where for risks. For example, you cluster your risks, you have 200 risks on a project, and you put them into groups and then maybe one of those groups has the same cause, or maybe one of those groups. You can use the same mitigation strategy.

00:18:53 Paul Boudreau: So, we’re starting to gain insights into our project that we wouldn’t normally be able to do for a number of reasons. First of all, because machine learning tools can manage more data than humans. And second of all, because we don’t have the time right. What happens is to use machine learning and we all want to do well with planning, right? That’s where we want to spend our time in the planning stage of our project and what are our sponsors, what do our customers say? Why don’t I see a shovel in the ground? Why don’t I see you? When are you going to start? So, we’re being pushed to start before we can spend all that time planning and what AI and machine learning can do for you is do that planning instead of taking two days to review a scope document, for example, you can do it in two hours.

00:19:42 Paul Boudreau: The third one, machine learning reinforcement learning. You know, self-corrections, right algorithms is based on feedback like learning to ride a bicycle and I say here imagine if humans never made the same mistake twice. I have to whisper this; this is my second marriage, OK.

00:20:02 Paul Boudreau: Anyway, it’s all right, it’s going well.

00:20:05 Paul Boudreau: So, reinforcement learning, I tell my students, listen guys you’re going into the workforce as project managers. Here’s what I suggest you do build your own knowledge repository. Become your own AI kind of person and what happens is you start building all this information, so you have an issue, or you have a problem in our project. What is the issue? What are the characteristics of that issue? What were the project conditions? What was the decision? And then was that decision successful? Yes, or no?

00:20:39 Paul Boudreau: And then you build this up maybe you have 20 issues. In your first project, you go on to the second project another 20 issues. In project management, we call this experience. We have a project manager 20- or 30-years’ experience. The problem is the human brain, the human memory is terrible. I can ask each and every one of you what did you have for lunch two weeks ago on Tuesday. You have no idea, right? It’s not something you wanted to learn, but even so, they’ve learned in court as well when they ask witnesses what they remember, human memory is just fallible. So, what you can do is build this up and imagine. “Imagine if project managers never made the same mistake twice”. So, you have this repository you’re building excel and you look at a similar issue and ohh that didn’t work. I better think of something else alright and what we do is we have a machine learning programme that goes through this and can pick out the results of it. Second one natural language processing analyses scope description for completeness and errors. What I call honesty and documentation.

00:21:50 Paul Boudreau: And I know some organisations prefer just to get the scope document done and then move forward. Now we’ll fix it when we get into the implementation don’t worry about it. I’m not sure that’s the best strategy for project managers’ careers, but sometimes it happens. Natural language processing has sentiment analysis. There’s a whole process behind this on you know, words and understanding words and phrases. You can send an e-mail, and somebody sends the word happy. You know, enjoy excited that’s a positive sentiment if you come out of a project kickoff meeting and in your project team group chat or whatever it is, the words that are coming through are frustrated, annoyed, uncertain. Then that’s a negative sentiment, and maybe you know, we need to improve the communication that we’re delivering to our team. So, there’s a good possibility there. OK, natural language processing.

00:22:52 Paul Boudreau: Virtual assistant so I use Alexa good for interactive communication, human test dialogue, and ubiquitous project management. I’ll talk about that in a minute. So, what about a virtual assistant for project manager, let’s see if this works, can I try this?

00:23:11 Paul Boudreau: Alexa, is there any training in the scope document?

00:23:18 Alexa AI: Yes, there is training, five people are identified for one week each of training. Would you like more details?

00:23:25 Paul Boudreau: Woohoo. So obviously I have my project in there. Listen guys, when I started this, I did this about 24 years ago and it was blowing people away like ohh man, that’s great. Now there’s about four organisations that do this. I’ll talk about it in a minute, but the second level I defined these three levels by the way. This is my definition we have a virtual assistant understands project management concepts. So, let’s see what else I can do here.

00:23:54 Paul Boudreau: Alexa, there are three tasks this week. Can I move them to next week?

00:24:01 Alexa AI: The tasks scheduled for this week are not on the critical path. However, one of the tasks has an identified risk based on the risk register. Should I contact the risk owner and ask her to contact you?

00:24:13 Paul Boudreau: So, there we go, thank you Alexa. So now what we have, I call this ubiquitous project management that is you can manage your project from anywhere at any time. There’s a number of organisations out there that offer this, so horizon PPM is a software application that has an interface to Alexa. You can talk about risks; you talk about your schedule at PM auto out of Germany is building this capability. They have some capability now and I can’t remember the other two.

00:24:43 Paul Boudreau: MS Project has it, but MS project is just faking it. An MS project you can say add another task well to me add another task is like an admin. It’s not really a project management kind of thing. Let’s see linked to machine learning and expert systems.

00:25:03 Paul Boudreau: So, what we have, and we have to be careful what I what I talk about with AI is something called an integrated solution and there are tools out there and I don’t want to mention any names like JIRA and what JIRA will tell you is. OK you’ve got George on this task. If you keep George on this task, your schedule is going to slip by a week. You should take George off this task and put Mary on the task.

00:25:27 Paul Boudreau: And so, you do that? Ohh isn’t AI wonderful? Look, it helped me bring my project back on track. Not so fast. I say not so fast, what about risk? Yeah, who cares? What about quality? Yeah, who cares?

00:25:45 Paul Boudreau: The tool didn’t tell you that, but as a project manager, aren’t we the people who have this overall perspective of the project? We look at all areas and maybe you assigned George to that task because they have the best quality and they’re likely to not encounter their risk or they know how to avoid the risk that’s involved. And what does the tool do? And so, we have to be careful with using AI that we understand an integrated solution. We don’t solve 1 thing and ignore other aspects that could be a problem later on. OK, so what’s the value of AI lesson guys? Here’s the problem we have you take any survey and I know PMI had said that at one point they were up around 65% project success and things were increasing, it’s not true.

00:26:36 Paul Boudreau: So according to PMI, 68% of IT projects are unsuccessful. KPMG says 70% of organisations have one project failure. The biggest issue that everyone agrees on 11% of project spending is wasted due to poor performance, inefficiency in resource allocation inefficiencies and how you’re doing bad project management decisions. I mean, we’re spending trillions of dollars on projects, if we could find a way to save 11% of that, it’s a huge amount of money. My favourite survey came out of Oxford University, and it was on large projects. They call them mega projects and I think it’s over $100 million.

00:27:23 Paul Boudreau: 2 % of mega projects are successful, 2% deliver the scope on time and within the budget Oxford University study. We can’t keep going the way we’re doing. It’s not, it’s not really working. OK, we need to find a way to change our methodology and I believe, I believe that AI offers us an opportunity. And what I see from where I am is that this disruption has already started because I’ve been working with organisations that are implementing AI for the projects, we’re moving away from this, what I call process-based adherence. And for those of you who may have caught that, that’s a scaled agile framework.

00:28:06 Paul Boudreau: Listen, I know Knowledge Train works with agile. I know there’s agile aspects of this, you have to get that knowledge in agile. Go ahead and take that training, you can’t change your project methodology without understanding what you’re changing. So go ahead and take the agile training. But what agile says what these advocates are, and advocates is a light word. They’re really, I don’t know what you call them, evangelists, or they’re so fervent in their belief that you have to implement all 50 or 80 aspects to agile to be successful. The answer is no, not true, right? It only delivers you 30 to 40% project success rate. So, we’re moving away from this process oriented project methodology to a data-driven methodology. We use AI to make decisions based on data.

00:28:59 Paul Boudreau: All right, productivity using generative AI like ChatGPT, Bard, if you’ve used any of those tools, those are being used a lot now in project management, creating a scope document, creating a scheduled template, doing air review, asking what my risks are. And it comes back, it doesn’t have a perfect and I can tell you that because I teach project management and what I do is I put my exam questions into ChatGPT to make sure that it can’t come back with all the right answers and it doesn’t, OK. So, it’s a good tool to help you on the way to creating good risk documents. To see if you’ve missed something.

00:29:43 Paul Boudreau: Decision making AI can help you take action sooner with a higher probability of success and project outcome analysis. I have a client in Europe now and there’s an organisation with 100 projects and what he’s doing is using supervised learning to decide well. I want to know which 10 projects I should prioritise because they’re going to give me the greatest value and then I want to know which 10 projects really, I should think about eliminating or terminating because they’re so far off schedule and off budget. They’re really not going to deliver what I expected them to deliver.

00:30:20 Paul Boudreau: The importance of data, hope I’m not going too fast here. The importance of data I have this picture on the right-hand side little image. Those are gold coins because I know you, you have the organisations have project management data. Listen, I work with the organisations that have years and years 10 – 20, thirty years’ worth of project management data. The question is, what are you doing with it?

00:30:49 Paul Boudreau: It’s a gold mine at least runs some data mining tools, predictive analytics through all of your historical projects. You will gain insights that you were not aware of. It’s going to find some correlations in that data that you may not have been aware of, whether it’s risks or schedule or what resource allocation, whatever it is. It is a gold mine and you’re doing nothing with it, OK?

00:31:13 Paul Boudreau: At the very least, work with my friend Martin Paver out of the UK, who has a data trust, and he will take your data to anonymize it. And then do the data mining and insight work for you. Structured data is essential, you may be capturing data. Anybody who’s done a data migration project as I have done, we’ll understand there’s tonnes of errors, you know, warning differences. My favourite, I have to tell you my favourite one is the owner’s name.

00:31:44 Paul Boudreau: So, this was a Kennel Club that I worked with, and they had a customer relations database that they were transitioning from a legacy application to Microsoft Dynamics. And we got to the field for who owns the dog, right? And there’s supposed to be a name in there and we found that some of them there were three names. So, the we got the company that some of the people functional primes went back and they we said to them why are the three names and they.

00:32:18 Paul Boudreau: Said Ohh, that’s because the people said, well we think all three of us should be the owners of the dog. So, you never know what’s going to happen once you implement AI 80% of the time. Initially, implementing AI is going to be data cleansing, fixing all those typos and missing values. You have to do some data maintenance and feature engineering is making sure you have the right characteristics.

00:32:48 Paul Boudreau: Now when I showed you my 87 characters in supervised learning, they could be 120, they could be 150. I think my risk one has 300. But there’s also an organisation in Australia that’s working with a client, and they’ve managed to cut those 87 down to less than 20. And they say, well yes based on these 20, we can predict early deterioration of project schedule. So, lots of opportunity, Vendor Landscape, Canada has something called mely.ai. If you don’t have a database, they have interfaces to things like MS Project and Primavera and they will put your project data into repository, and you can do some work on it.

00:33:32 Paul Boudreau: In Europe, I guess mainland Europe, I’m talking about here. Mainland, not the UK, so epicflow manages multiple resources across the project. So, one of the clients they work with has 4000 users and they’re allocating those 4000 users using an AI algorithm to optimise the efficiency of completing all the projects they’re working on. Fortean is that of Switzerland, my friend Marcus PM Otto is Ricardo. He has Alexa, not Alexa, but it’s like a virtual assistant. Lilli dot AI to France, that’s Milly Tang.

00:34:11 Paul Boudreau: The UK I’ve talked to most of the founders for these things. They asked me to give the feedback. Horizon PPM, yes Peter has the voice interface to the to their project management tool, Nodes & Links. I attended a demo from him a number of months ago. Very interesting tool for construction scope master. My friend Colin out of the UK somewhere near Manchester if you have an agile project with user stories. You absolutely must use Scope Master. He detects errors and emissions in user stories, and I’ve got a case study coming up to talk to you about it.

00:34:50 Paul Boudreau: Shark Tower out of Scotland, Craig was out of Scotland last time I talked to him. nPlan so nPlan does risk management for your schedule and Google Venture Fund gave them $18 million two years ago to improve their risk management processes for projects. Greyfly is a consultant coming to the UK often. That’s my friend Quang out of Australia, one of the most brilliant machine learning programmers I’ve ever met. So, a couple of case studies before I wrap this up, oh I’m doing well. First case study at large multinational business, so I was contacted by a company out of Germany.

00:35:33 Paul Boudreau: They had 26 project management offices around the world, OK and the CIO said yes, we have to get into this AI stuff. So, the director asked me to do a workshop for them. So, I did a workshop for these 26 project management offices and first thing I noticed is of course, there were people who were sceptical. This is a change management process, not new to AI, right? AI is just another technology. Whatever technology you’re facing, whether it’s blockchain or internet of things or virtual reality. You’re going to have to face the change management process.

00:36:09 Paul Boudreau: So, people are saying, well, AI doesn’t work here because I’ve used it, and we use AI here. These guys-built things like backhoes and shavers and industrial microwaves. So, what happened is there was one PMO that was interested. The head of that PMO said yeah, I’m on board. I’d like to be a pilot project right. So, they went through and what I always say is what’s your project issue? What is it in your project management world that you’re trying to fix?

00:36:40 Paul Boudreau: I do have to tell you I did this presentation to United States, and I asked that question and the woman said to me, well we have a target rich environment. Have you ever heard that before? A target rich environment, I said ohh let me translate. You mean you’ve got lots of problems? So, I’ve never heard of that before. So, I said to these guys out of Germany, I said well, find out what your project issue is.

00:37:08 Paul Boudreau: Go to vendors, get some demos, and figure out how to solve it. So, they implemented their big issue was managing risks across the organisation and so they used an AI tool to deploy for risk management and they’re happy, I guess. Second one, this is a one I’m very familiar with.

00:37:27 Paul Boudreau: So, this was a project to consolidate as payroll software. This was an agile based software project. So, they had a reputable event. I won’t mention the vendors’ name, the consultant they used because it would be embarrassing. It’s a brand name consulting if I said you would immediately recognise whose I’m talking about.

00:37:45 Paul Boudreau: So, they helped them put this together and put all their scope statement together. The project budget was 310 million. They ended up spending a billion dollars on it. OK, so what happened? They were consolidating payroll systems to run the organisation because they had six or seven or eight different payroll systems, good value put it into one better value by having one system. They went live, people were overpaid by $40,000. Some people didn’t get paid for two or three months, it was an absolute scandal, embarrassment, nightmare.

00:38:21 Paul Boudreau: So, they did an audit on it, what are the auditors say, incomprehensible failure of project management. What is the recommendation? Blame it on the project manager. Are you kidding me? Blame it on the project manager. Listen guys, I want you all to look at me for a minute. I know I’ve got the slide up, but look at me. Project managers, it’s not your fault we don’t have the tools. It’s like saying here, go dig a hole with your bare hands. Oh, I’ve got a shovel, but I’m not going to give it to you. I’ve got it back home, but I’m not going to give it to you.

00:38:57 Paul Boudreau: We don’t have the tools, so what happened? So, this organisation got a hold of me at 2022 late last year and they said Hey, Paul listen, we’re going through another consolidation software project. What can we do? I got a hold of Colin at scope master. They ran the programme through, and they corrected all the errors. There’s an Accenture survey that says 37% of all agile projects fail because they haven’t defined their user stories properly. And then I got another call just two months ago from the same organisation they want it. They’ve got another software project they want to use the same tool. Can we do it again? OK, so this is a large government organisation using AI tools. Guys, if the government is doing it, why aren’t you doing it?

00:39:52 Paul Boudreau: We know governments are very risk averse, but they’re also very scandal averse. They don’t want to be embarrassed. So how do you get started? How do you get started? What do you have in your organisation? What project data do you have? What does it look like? Where is it stored? Is it structured? What’s your current project issue? What’s the problem you’re having? Uh create an AI road map. What are we going to do right? What’s our problem? Which vendors do we want to talk to? There’s a make buy process here. If you have a strong IT department, you can do it themselves. My predictor tool it’s based on 14 lines of code. There’s all sorts of libraries and functionality that comes with the software now. Defined the value guys I am the business background, so I believe in a business case, so define what the expected value out of this is OK we can reduce our cost or stop overspending or reduce spending on risks or identify risks sooner. Whatever it is, to find the value put together business case for it, deploy the tools. Most of these tools are not expensive at all.

00:41:02 Paul Boudreau: They’re either on a subscription basis or a user basis. You can either have them deployed in the cloud or on site as you wish and here’s my main takeaway. If you remember nothing, risk management processes were developed in the 1980s. I know what you’re doing, I know what everybody’s doing with risk management, probability, and impact right? That’s what you do. Probability and impact that was developed in the 1980s. Ohh, we’ve got fancy simulations like Monte Carlo and statistical methods we’ve added to it.

00:41:37 Paul Boudreau: But basically, it’s a 1980s process we need to update it. The critical path was developed for the space programme in the United States in the 1960s. Really, that red line on Microsoft project. You’re still looking at that and there are organisations that do it a lot differently. Organisations like Nodes & links look at it and say, well, it’s not just the critical path. What about risks? What about resources? What about the area constraints?

00:42:04 Paul Boudreau: But the message is we need to modernise project management, we need to be more productive, we need to be more efficient, and we need to be far more successful. OK, you’re the people here, you this is your opportunity to lead this change. It’s your opportunity to embrace this technology and find ways. It’s not like a turnkey solution AI has different creative ways to be implemented and as project managers, I think there’s incredible value there for you to understand the technology and find a way to implement it in your organisation.

00:42:45 Paul Boudreau: This is me, my contact information. I guess the slides have been made available later. Feel free to send me an e-mail. My company that I started with still matter consulting. I have three books on Amazon, and I’ve been at yeah. So, love it if you start to be purchasing them, you know there’s 16.5 million project managers in the world and I’m not too sure how many can read because my book sales are not doing very well. But I won’t mention that I’d be happy if you pick them up on Amazon in the UK, either Kindle or printed version, and I want to talk about Knowledge Train.

00:43:23 Paul Boudreau: So, thanks for this. They have a lot of courses here I see PRINCE2, Agile. Guys, it’s important to take these courses and increase your knowledge. They didn’t pay me to say this. I’m just saying it. OK? Look at Lean Six Sigma, very appropriate when you’re using AI because AI can take that concept and advance it even further, right. Change management I’ve said this, you need change management to implement AI technology because people will resist it and I’ve seen studies I’ve seen, Accenture studies. I’ve seen what they do Scrum master, Scrum master studies. Where they try to automate.

00:44:03 Paul Boudreau: And people started refusing to share data because they didn’t want to deal with the automation. So, you have to be able to understand project management to a certain depth and I’m happy that this organisation, Knowledge Train, provides this opportunity for you, all right.

00:44:24 Paul Boudreau: Contact information, I’m going to stop sharing. I hope you’ve enjoyed it and listen, I’m happy to take questions.

00:44:37 Sevcan Yasa: Yep, thank you so much. Paul was a great presentation and as Paul did mention, we do have various courses. So, if you are interested, you can always e-mail me. I can send you some information. It’s gonna quickly pop down my e-mail. Alternatively, you can write in the chat, and I will e-mail you with more information. So just before we head over to the Q&A, as I mentioned there is going to be a very quick survey. So, we do have a few questions actually going to start off with, is all AI project management based on agile?

00:45:30 Paul Boudreau: AI project management, you can use AI and project management for agile, waterfall or hybrid and based on some of the latest surveys, Antonio Nieto-Rodriguez, who used to be with on the board of directors for PMI, he’s come up with something like 70% of organisations are using the hybrid version. They take the best of agile, and the best of waterfall and they use that process so AI could be applied to anything, yeah.

00:46:03 Sevcan Yasa: There’s another question, how beneficial is it to undertake data analytics course?

00:46:11 Paul Boudreau: Yeah. So that’s a great question and what I’m going to talk about here is the opportunity for project managers. When you think about it, organisations are hiring data scientists. I’m not sure you’re aware of this. So, a lot of organisations are implementing AI outside of project management, hire a data scientist to help them use the data and understand the data better. And data scientists are in huge demand, I know from North America and the US these guys are making over guys and girls are making over $200,000, some of them upwards of $250,000 – $300,000 a year in salary. So that’s a lot of money, can you afford to hire a data scientist? Even if you could find one? No, but if you’re a project manager and you take data analytics. Then all of a sudden, your value increases incredibly. Do you need a data scientist to say? Well, look what project columns should we use, and which project column should we not use? The data scientist can’t answer that. A project manager can answer that but if the project manager has some data analytics background.

00:47:20 Paul Boudreau: They’re going to be having their foot in both worlds. They’re going to have a data scientist kind of perspective and a project management perspective. How valuable are you going to be to the organisation? I think you’re gonna be incredibly valuable to the organisation, so yes, I’d recommend it. I don’t think it’s depends on what your organisation has. If you have IT people or not IT people but yeah, I think it’s a good opportunity for you.

00:47:44 Sevcan Yasa: Next question is, can we teach systems thinking with certainty?

00:47:49 Paul Boudreau: I’m sorry I missed the question.

00:47:51 Sevcan Yasa: Can we teach systems thinking with certainty?

00:47:54 Paul Boudreau: Can we teach?

00:47:55 Sevcan Yasa: Systems thinking with certainty.

00:47:58 Paul Boudreau: Systems thinking with certainty, you know what AI is. As I said, it’s not a if I understand it your question correctly. AI is not like a turnkey solution. As I said, it requires a bit of data management understanding for project managers, and it requires a bit more math and statistics. So, the other thing, aside from data, is you need to learn a little bit more about math. You don’t need to have a master’s degree in regression analysis and correlation, but you should know how the data you’ve put in and it’s churned through the algorithm that comes out. And you have an output result of, you know, 75%. What does that mean? What can you go back and look at and in supervised learning you saw those labels. There’s something called balanced labels, so you can’t have 95% of the projects failed and 5% of the projects were successful because your output from that algorithm won’t be correct right. It doesn’t have enough of a difference to be able to distinguish between the two success values. So, you do have to understand a little bit about math and statistics and about things like balance data, how you interpret the results, things like that. I hope that answered your question.

00:49:20 Sevcan Yasa: Thank you, another question by Charles is, how to predict I know, and the applications are clearly manifold, but what do you think the unintended consequences of AI may be in project work, both positive and negative?

00:49:34 Paul Boudreau: Positive, there’s no question it’s going to increase the project success rates. It’s going to make decisions better and faster. Negative listen, it’s software like any other software. If you implement it incorrectly, it’s going to be a disaster. Two issues, number one, you don’t implement it properly. You don’t have people trained properly how to use it and you think it’s going to work. It’s not, can I use this word? A panacea it’s not a solution for everything, right? You can’t just flip a switch and say, OK, here’s AI let’s make it work. Let’s do it right, Ok. It’s fantastic, no you have to be able to understand what you’ve done, and you have to be able to interpret it properly and apply it properly.

00:50:13 Paul Boudreau: Second thing is that data, there’s a book out there called ‘Weapons of math destruction’ by Cathy O’Neill, and it talks about upgrading data and the analogy I use is you’re going into an office building, and you walk into the office building. There’s an elevator, you have to go up to the, I don’t know, 15th floor. There’s a sign that the elevator says, oh, by the way this elevator hasn’t received any maintenance for the past 20 years. How do you feel? Same thing with the AI algorithm. You have to continuously update the data to update the model to make sure it’s accurate. So, I think that’s the positive and negative. There’s some things you have to be aware of, but I think it’s an incredible opportunity.

00:50:57 Sevcan Yasa: Thank you, Paul. So, the next question is actually referring to a specific slide I believe, which was the virtual assistant for project management slide. So, the next question is, in order to benefit from this, how much of depth do we need to get into NLP?

00:51:16 Paul Boudreau: National language processing. Like I said, there’s vendors who do this. So, if you connect with the PM Otto, just do a Google search for them or a Horizon PPM, do a Google search for them. They have they have the application that does it. NLP is a little bit easier then machine learning, but if you look at things like generative AI and ChatGPT they have machine learning algorithms running in the background. I understand that ChatGPT is going to have a version on your they’re going to release the version on your smartphone within the next six months, I believe I wasn’t sure the details. So, you don’t need to know all of but again you can’t just blindly follow all these results.

00:52:04 Paul Boudreau: OK, as a project manager you need to be a, what do they call it? A sober second thought? Can I use that expression? So, you need to use your judgement on what’s happening in the project and whether this decision or whether this output is appropriate for the result that you’re looking for.

00:52:29 Sevcan Yasa: We have one more question. When I saw this question on the machine learning slide, but I think it was before it might have been the clusters one, I’m not entirely sure, but what would you see on the graph axis here?

00:52:45 Paul Boudreau: Sorry what, I missed the question.

00:52:47 Sevcan Yasa: What would you see on the graph axis here?

00:52:51 Paul Boudreau: Well, what I see in the graph axis. Oh yeah, it’s basically a number of for the tasks it would be the number of tasks and the bottom would be level of complexity of the task.

00:53:04 Sevcan Yasa: Perfect. Just seeing if there’s any more questions.

00:53:08 Paul Boudreau: So, I can talk about one that I get asked often while we’re waiting, and you go through those. How much data do you need? So, there’s another myth out there, this, this misunderstanding and misinformation that put out people put out big data. Oh, we can’t use AI unless we have big data. We need massive amounts of data. So first of all, well if you’re doing medical field X-rays and MRI brain scans absolutely get as much data as you can. And project management with supervised learning in some cases, yes, in some cases no. So, there’s an organisation that’s developed the PMI tool, by the way, those the AI tools that I showed you, the examples of the vendors, they have AI at the core, they don’t have like a project management tool. And then they added AI they were built with AI in it and then they adapted it to project management. So that’s why I like them, but you can have as few as thirty data sets, and I’ve seen in the academic world I went to a professor’s presentation where he had fifty data sets. So that’s fifty projects between 30 and 50 projects. And then figure out how many characteristics you want. Unsupervised learning you don’t need big data I just showed you we can classify tasks, or we can classify risks based on whatever you have in your project.

00:54:32 Paul Boudreau: Reinforcement learning? Yes, probably a bit more data. There’s something called genetic algorithms which I didn’t have time to present here. I didn’t have time present about generative AI either, but genetic algorithms. They have the ability to do certain activities with the minimal of data they make decisions. They’re a great decision-making tool, they fit into reinforcement learning very well.

00:54:59 Paul Boudreau: So, a natural language processing that we’ve seen ChatGPT now with huge amounts of data, which is why it’s successful. But for a lot of the machine learning stuff, big data, you don’t need big data to use it for project management specifically. And the problem with the ChatGPT is it’s not focused on project data specifically. What I see organisations doing is they take the ChatGPT engine, and they put it inside their organisation. And then they direct, it they aim it at the data the project management data that they have and then they say, you know, what should I use for a scope document, create a schedule for me and it bases it on that organisations data right now. So, I was talked to a professor out of Europe, and he took the ChatGPT engine, and he was getting some results out of it, he added PMBOK. That’s PMI’s, project management guide or whatever and he said the answers got worse. So, we have to be careful how we use these tools, because at least initially there’s going to be a lot of misinterpretation, misdirection.

00:56:12 Sevcan Yasa: Thank you. If I did miss out your question, please let me know.

00:56:20 Paul Boudreau: I can talk about one more. There’s this other survey that came out from PricewaterhouseCoopers that said, what is it by the year 2030, 80% of the project managers tasks will be automated. I talked to those guys, Mark and Adrian, about it. They just wanted to make it controversial so they could, you know, get some attention. What they talked about in that article, if you’ve read it, is things like creating a status report for a project, organising a meeting for a project, organising and prioritising emails. Listen, guys, do you think organising a meeting is a project management task like really, it’s an admin, If you’re paying project managers to spend 80% of the time creating status reports and organising meetings like you got the wrong perspective. Get a project coordinator right. When I talk about AI, I’m talking about the real project management issues. We have to deal with making tough decisions, trying to resolve issues with customers, trying to manage stakeholders and manage communication properly, having a lot of foresight, being able to predict and prevent issues before they arrive.

00:57:33 Paul Boudreau: And I tell this one last story you have Sevcan. One last story, I talk to my students all the time. You’re going to get out there, into the world, and you’re going to want to do a lot of planning and predict issues and what’s going to happen is if you can successfully predict issues before they happen, then you’re going to be sitting there, right. And your boss, your manager, is going to say, wow. You know, it doesn’t look like you’re doing much work. And then there’s going to be another project manager and they don’t do anything; they fight fires the entire day. They ignore, they ignore the prediction, ignore preventing everything. And so, what happens is they’re working overtime to try and resolve project issues. And what does their manager say? Wow, look at that person, they’re so dedicated. Look at all those issues they’re trying to resolve, we’ve got this perception wrong.

00:58:24 Paul Boudreau: The person who’s doing the firefighting should have taken more action. They should have been done more work in the planning stage, and if we can use AI tools to predict things and run a check on your scope document to make sure it’s accurate instead of taking two days, you can do it in two hours. So, the person who does all that planning and prevention, we need to change the perception. They are the project manager that’s doing better than the project manager is running around like crazy trying to fix everything because they didn’t do a good job of planning and not doing a good job of communicating whatever else it is.

00:59:01 Paul Boudreau: OK, any other questions?

00:59:12 Paul Boudreau: Sevcan I can’t hear you, any other questions.

00:59:20 Sevcan Yasa: It’s just we do have a little thank you.

00:59:24 Paul Boudreau: Ohh the best 60 minutes of my day today. Thank you, thanks so much I appreciate that. Listen guys, I as I said, I try and stick to the facts as much as I can. I do have opinions on things but be careful. There are people out there who think they’ve read, you know they read a book on AI and all of a sudden, they’re an expert and they start talking about it and I get upset, right I try and be a gatekeeper in those things. But I encourage you to improve all of your learning, whether it’s through Knowledge Train or whether it’s through reading anything about AI. I have articles that I published on a blog, so you know, go ahead, improve your skills, but make sure you’re getting the facts. Don’t get the misinformation all right.

01:00:15 Sevcan Yasa: We do have one more question. Do you have any advice reading about AI for research management?

01:00:28 Paul Boudreau: Reading in terms of reading, well I have three books out there, so I would prefer if you go to buy, there’s a lot more now than when I published my first book. There are a lot more that are out there. In terms of researching AI specifically not on the project management side. I had some books, you know, there’s like a data analytics for dummies books. Some of them are pretty good. There’s machine learning for dummies if you want to get the basic background, the fundamentals of it. But my first book was applying AI to project management. A lot of fundamentals is for project managers. The second book is PMO, it’s more about value and a higher-level perspective. And then the third book is the self-driving project where I wanted to be controversial. So, I said let’s tell people how a project can be implemented without using a project manager at all. Now at the end of the book, I say, you know, we need to collaborate. Project managers need to collaborate with AI to be successful. And I try to keep the books the prices are pretty reasonable so. “Weapons of math destruction” I mentioned that Cathy O’Neill the other one I can mention is a “Superintelligence” by Nick Bostrom. He is a negative guy on AI, so it’s always good to get a different perspective and out of Oxford University anything you can read by Professor. Bent Flyvbjerg out of Oxford University. He doesn’t have so much on the AI side, but he talks about bias. He talks about all sorts of things and he’s one of my favourite people to read something.

01:02:20 Sevcan Yasa: Thank you for that, Paul. Does anyone have any more questions?

01:02:47 Sevcan Yasa: Doesn’t seem so few people are typing.

01:02:51 Paul Boudreau: I’m happy if you connect with me on LinkedIn, just type in my name. I’m in Ottawa, you probably seen my picture there Ottawa Canada. So, type in Paul Boudreaux, space, Ottawa and I should pop up. I’m happy to answer any questions send me a message in LinkedIn or send me an e-mail. Happy to deal with people.

01:03:13 Sevcan Yasa: In the emails I did send out to everyone I did actually link Paul’s e-mail so you can always go back. What I can also do is put Paul’s LinkedIn link the e-mail that I will be sending out next week. Those one more question is your training tool open source?

01:03:37 Paul Boudreau: In the sense that I offer it for free? Yes, do I give it to other people? You’d have to negotiate that with me. Right now, the interface is written in Heroku it’s hosted by Amazon Web Services. It’s available on the web. You just I give you an account and you get a template to start off with. You get instructions how to log into your account and you get a short video on how to use it and you will be off and running so you can play with it. So, it’s not open source and it and it’s data in the meaning of that I can give you all the code, but it is open source in terms of it’s available on the web for free. For now, I’ve had a couple of people ask me if I if I would commercialise it and I haven’t done it yet.

01:04:34 Sevcan Yasa: Perfect, thank you, if anyone doesn’t have any more questions then I’m going to slowly end the webinar. Thank you so much everyone for joining. Thank you, Paul, so much for attending.

01:04:48 Paul Boudreau: Thanks, it’s been great. Always happy to share my knowledge. You know what we need to really, really increase our knowledge of AI and not be scared of it.

01:05:02 Paul Boudreau: So, thanks Sevcan, Thanks everyone.

01:05:05 Sevcan Yasa: Thank you. Thank you everyone and hope you have a nice evening.

01:05:15 Sevcan Yasa: Thank you, Paul.

01:05:17 Paul Boudreau: OK. Thanks, I’m going to drop off.

01:05:18 Sevcan Yasa: Thank you everyone.