Tips for Using Generative AI Tools in Project Management

STEVE HENDERSHOT

ChatGPT and similar tools are changing the way people work. And this is just the beginning.

Generative artificial intelligence consists of machine learning models that can generate text, images and other content based on how the model was trained. And right now, there are several generative AI tools that are openly available to the public.

Today we’re looking at how project leaders can use these emerging technologies to increase productivity—while keeping risk and privacy top of mind.

NARRATOR

The world is changing fast. And every day, project professionals are turning ideas into reality—delivering value to their organizations and society as a whole. On Projectified^®, we’ll help you stay on top of the trends and see what’s ahead for The Project Economy—and your career.

STEVE HENDERSHOT

This is Projectified^®. I’m Steve Hendershot.

How can generative AI tools like ChatGPT be useful to project managers? I asked ChatGPT that very question.

It told me it could be quote incredibly useful unquote in areas ranging from project planning to risk management and resource allocation by generating insights and recommendations after project leaders feed it data such as project requirements, objectives and constraints.

ChatGPT is the most prominent example of generative AI, a technology that has advanced from a fantasy to a full-blown phenomenon in what seems like a flash. OpenAI first released ChatGPT in November 2022, and already nearly a quarter of adults in the U.S. have used ChatGPT or similar tools such as Google’s Bard, according to the Harris Poll. And McKinsey predicts generative AI could add up to 4.4 trillion U.S. dollars to the global economy.

All this while the technology still has plenty of kinks. A recent Salesforce survey found that a majority of workers believe generative AI introduces new security risks and produces biased and incorrect outputs.

So as organizations, teams and project managers look for ways to safely incorporate generative AI into their workflows, they’ve got to be smart about it. To help with those efforts, today we’re speaking to two leaders who have been using and experimenting with generative AI tools. We begin with Kristian Bainey, CEO of the IT consulting company K-Pic Systems Inc. in Edmonton, Alberta, Canada.

MUSICAL TRANSITION

STEVE HENDERSHOT

Thanks for talking with me, Kristian. So ChatGPT has marked a huge inflection point where people stop thinking generative AI is some interesting future technology and it’s perceived as being useful now. What’s your take?

KRISTIAN BAINEY

ChatGPT is definitely a tool that can help anybody in any field with the correct prompts. So the better you learn about this tool, the better responses you’ll get. Even myself, there’s not a guru in ChatGPT, as it came out to the public, right? I mean, the people that invented it probably, yes. But people that are just like me and everybody else that paid for the subscription or used the free product, I’ve just practiced and practiced and practiced. And the more you use the tool and the more you understand it, the more educated you are about it, [and that] can really help you pinpoint how you can use this tool.

ChatGPT is artificial intelligence. Now, artificial intelligence is basically a power tool used to improve data-driven decision-making that cannot be separated from its human or societal content. Machine learning is a tool that helps derive useful insights from large data sets. Decision-making should always involve risk analysis to assess how likely it would be that the prediction your system makes is correct and from what the cost it would be if it wasn’t incorporated properly, and the matter of use and trust in this system. There [are] different tones. There [are] different formats that you could use, and there are different actors involved. For example, for ChatGPT, a lot of people will say, “Okay, I’ve heard all over the internet to use a lot of syntax—a lot of words or whatnot—to get a proper response.” Now, in my opinion, my professional opinion, I’ve been using it since day one, that is absolutely incorrect when it comes to acting as a project manager. What I would suggest is using clear and concise prompts where you get a result back, and then you keep on revising. There’s a lot of tips and tricks on how to use it.

STEVE HENDERSHOT

Let’s walk through an example of how a project manager can use ChatGPT. Tell me about a possible prompt and, if the initial return isn’t quite what they’re looking for, how they can revise the next prompt to get an answer closer to what they need.

KRISTIAN BAINEY

Let’s go into an individual question. When you’re asking an individual question, what you want to use is a framework. The framework that I suggest is “act as a role, question, action, using a tone.” It’s that simple. I’ll say it again. “Act as a role, question, action, using a tone.”

So, for example, “Act as a role” is “Act as a project manager.” What’s my question? I want a project charter describing the construction industry within Edmonton, Alberta, compared to Trinidad and Tobago in the Caribbean, and I would like it in quantifiable results. Please use sample data. “Using a formal tone.” That will give you a good response. And remember, this is a digital helper. It’s not going to give you the answer. It’s up to the human. The human has empathy. The human has emotional intelligence. The human is going to make the decision at the end of the day. What ChatGPT is going to give you is information. It’s going to give you the options. It might even give you the root cause—actually, it can give you the root causes, plural—but the human is going to make that decision.

When you have a role, question, action and tone, like I said, and the result comes out, you could say revise. You say, “Hey, can you also make this part in a table format using the categories of cost-benefit analysis, risk assessment, stakeholder org chart or whatnot?” You could put stuff like that in. So that was a good example of just a simple scenario for project management.

However, there [are] a lot of organizations that might be thinking, “Well, my organization has a template for everything. We have a PMO, [so] why would a question like that help me?” Well, the first thing you would do is exactly what I said in the sense of putting role, question and action, and tone. Then copy and paste a template that you have in your Word document or PDF. Copy it, put it in your clipboard, then go into ChatGPT and put: “Revise [with a] period. Reformat to the following [and add a] semicolon <paste>.” What’s that going to do? It’s going to take the same information it gave you and put it in exactly the same format that your template at work does now. You can customize it and configure it the way that your clients like to see or [with] the different type of templates. But again, using limited scope and revising that response in the way that you want it to is very beneficial.

Now, let’s say that you have a lot of questions on the same subject, but you don’t want to keep on going role, question, action, tone. What is a good way? So I like to call it a bulk question, and I like to use the term bulk because I like to tailor ChatGPT for all of my future questions. So say, “For every question I have, I want you to act as a senior project manager, being clear and concise using a formal tone.” Then you’re going to put the word “Ready [with a] question mark.” If you don’t put that ready [with the] question mark there, you’re going to get a huge response back about project management and clear and concise, formal tone. If you put in “Ready?” all it’s going to come back [with] and say is, “Yes, I understand the requirement. As an AI language model, I am prepared to answer your question with clear and concise and formal tone while providing insights as if you are a senior project manager. Please feel free to go ahead with your questions.”

STEVE HENDERSHOT

That’s a nice look at possible workflow efficiency. How else can project managers use generative AI tools like ChatGPT?

KRISTIAN BAINEY

You could use it for problem-solving techniques. You can use it for innovation and idea generation. You could use it for a competitive analysis. You could use it to create project charters, statements of works, RFPs [request for proposals], briefing notes, a stakeholder management plan, a project management plan, forecasting, communication and quality assurance, [a] training plan, agile user stories, agile personas, risk analysis, [a] power interest grid, earned value management requirements, [a] traceability matrix, status and progress reports.

Let’s say you’re doing a business case. What are you going to do? People want to get money from their sponsor, right? So let’s put in “Act as a project manager, create this business case,” and here’s all the information that I came up with: “Can you please say the return on investment is this? Can you please say, this is how much money that I’m looking for? Here [are] the stakeholders. Here [are] our concerns.” And put all the information that comes from you. Don’t worry about grammar and all that kind of stuff. Just put in everything that comes from your head. Then, at the end, put, “Please use a persuasive tone and highlight in bold the important information.”

When you do something like that, it’s going to come back with an extremely cool response and you can say, “Great. Now can you put the second half of it in a table format so it’s easier to understand?” It will come back and give you a table format. Highlight the most important things that it thinks that you should bring it up to your sponsor. And you just tweak it. You have saved yourself probably five hours when you can just spend 45 minutes now customizing what you really want to emphasize and tweak in your business case.

STEVE HENDERSHOT

Let’s turn from the good examples to things project managers might want to avoid when it comes to using generative AI tools. What are some potential inaccuracies and risks to watch out for?

KRISTIAN BAINEY

What [the] majority of people are doing as project managers or even leads of organizations trying to implement ChatGPT within their organization is missing the concept of what artificial intelligence really is. A misconception is “Let’s introduce new technologies with AI to improve our business efficiencies. Great. Yeah, let’s do that.” That is not the proper way to go because you’re going to get a lot of sometimes chaos or a lot of resistance to change and people that just don’t understand what they’re trying to do for the organization for efficiency. The better approach is to take existing processes within your organization and feeding that towards artificial intelligence.

You know why that will work better? Because the people already are introduced to this concept. They already know how the workflow works. When you introduce AI in increments or whatnot within existing processes, people will understand the benefits that are coming out of it. How would people really understand the benefits when you’re just introducing a new technology? What is it really going to help? How has it helped in the past? Well, we don’t know because we didn’t do it before. Right? So if you take existing processes and you do that first, then when people are really up to date and understand what we’re trying to do and see what the benefits are for your organization, then you can introduce new technologies moving forward.

STEVE HENDERSHOT

One of the risks we’ve heard a lot about regarding ChatGPT is the tool giving users bad or inaccurate information. How can project managers mitigate that risk?

KRISTIAN BAINEY

Well, the only way to do that honestly is if you customize your model and fine-tune it for your organization. If you’re using ChatGPT and just getting results, and you’re not even too sure if it’s misleading information, the only way that you would know if it’s actually misleading information is to have the knowledge and the background on it. You know what you can do? Honestly, and I’ve done it before, if you get a wrong answer, and you think it might be wrong from ChatGPT, you can ask it, “Are you sure this is correct?” Sometimes it will come back and say, “My apologies. It’s not.” It actually will act in the sense of [being] like a human that makes mistakes as well.

If you’re using Bing or something like that or ChatGPT-4 that accesses the internet, you can ask it things like, “Can you please let me know where this information that you’ve provided supports in a published article or some type of reference like that?” And it will come back, and it will say, “Well, according to ...” It might say a news article two years ago or something like that, but it can do that.

So there are ways if you really want to know it. There [are] three ways. Number one is to fine-tune it. Number two, if you can’t, if you’re not doing that, [is] to ask it, “Are you sure it’s correct?” See what it comes back with. Number three, ask it to reference where this information [comes from] and if there’s any supporting documentation that’s published on the internet that you can show me. And if it cannot, [if it] insists on “I cannot find anything," because a lot of times it will, you know what? It’s probably something you don’t want to use then.

STEVE HENDERSHOT

Beyond the applications that people are tinkering with or implementing right now, how is this going to evolve to impact project management and project managers specifically over the next few years or months? Pick the right time horizon.

KRISTIAN BAINEY

I would actually say months. AI is not there to take your job, but [to be] the person next to you, or the person in the same position may have an advantage. Now what does that mean? Two people are applying for the same job; one has some experience within artificial intelligence. That person will have an advantage, and how the future for project management will go, in my opinion, is that a lot of analysis work will be automated. For example, there are plug-ins within ChatGPT-4 when you can give it a great description of a workflow in words. Say, “I want a swim lane or a flow chart based on this information.” It’s going to come back and do it for you. It can create current states for you and even future states. Any type of analysis work right now that can be automated, the future of project management will have a total advantage over that because they have the tools that can do it to automate it. Now, they will have people that need to customize it, to look over it. The same business analysts, the same project manager, whatever, will look over their cause-effect diagram or workflow or current state or whatnot. Sure, they will have that because you always need a human to do that. But the nitty-gritty work of the analysis will be automated in the future. And again, project managers need to stop thinking of how AI or ChatGPT can do what humans can and need to start thinking [about] how AI and people can work together.

MUSICAL TRANSITION

STEVE HENDERSHOT

Project leaders across industries are experimenting with generative AI tools, seeing how they benefit teams and their workflows. Projectified^®’s Hannah LaBelle spoke with Higor Prado, a project manager and software engineer at technology and software company Softfocus. Based in Mateus Leme, Brazil, Higor shared the project management applications where he has found value from generative AI.

MUSICAL TRANSITION

HANNAH LABELLE

Higor, today’s episode is all about how project professionals are using generative AI tools—like ChatGPT or Bard, to name a few—in their project management practices. So let’s start there. How are you using generative AI in your project management, and how is it changing how you work?

HIGOR PRADO

Okay. It has been a gamechanger [in] my work. I like to use tools like ChatGPT to handle things that used to take a lot of my time. For example, I need to write an email negotiating with a stakeholder, and I like to ask ChatGPT to help me with it. I like to use ChatGPT with repetitive tasks and other kinds of things.

HANNAH LABELLE

And what sort of repetitive tasks would you say? Is that kind of the report-building and things?

HIGOR PRADO

Yes, report-buildings and writing emails. We have a lot of documentation when we think about project management. So the most part of the process that makes us spend time is writing things, and ChatGPT is good because it’s a text generator. Now these tasks are done more efficiently, which frees up my schedule so I can focus on strategic planning and problem-solving.

HANNAH LABELLE

What other project management tasks do you find are most suited to using generative AI?

HIGOR PRADO

These tasks that we can review. We can create, format and edit documents quicker than ever. Then there are meeting scheduling, risk identification; these are all areas that can benefit from AI. But we need to remember, the AI is a tool, not a replacement for our judgment or experience.

HANNAH LABELLE

Let’s walk through an example of how you use generative AI in your project management. Tell me about a specific time where you used it, and how did it affect how you managed a specific project?

HIGOR PRADO

Yes, it was at Voll. There was this one project where we were developing a new software module, and I decided to use ChatGPT to generate our weekly progress report. It was so convenient. The reports were consistent, structured and clear, and it saved the team so much time that we could invest back into the project.

HANNAH LABELLE

The time—there was a little bit of a benefit. Were there any other benefits that you can think of to that specific project or to the team’s workflow?

HIGOR PRADO

Yes, the benefits were significant. It saved us time and boosted our productivity. On the flip side, there were potential risks, too. For instance, there could be inaccuracies in the generated content, and to manage that, I always [made] sure that I or someone from my team reviewed the output before it was finalized. So although the AI was doing the, let’s say heavy lifting, the human touch was still very much a part of the process.

HANNAH LABELLE

What do you need to look out for when you’re using ChatGPT [or] similar generative AI when it comes to accuracy, organizational privacy, compliance, different things like that?

HIGOR PRADO

You always need to be aware of privacy and compliance issues. It’s crucial to follow your organization’s guidelines and not allow the AI system access to sensitive information. I always make sure that the data I use with AI is carefully vetted and sanitized.

HANNAH LABELLE

When you say “sanitized,” what do you mean exactly?

HIGOR PRADO

For example, if you are writing an email or report, we always need to replace the real name of the stakeholders, the sensitive information about the ID of the people or emails, this kind of thing, and especially things related to secrets from our company. For example, I need to write an email for a stakeholder, I will not put the real name of the stakeholder. I will not put the real email of the stakeholder. I will just [say], “Okay, ChatGPT, help me to write this email. This stakeholder is a person that has this personality, and I need to ask him or her to give me more two days, and I need to negotiate it.” And when I do it, I do not put the real information about the person.

HANNAH LABELLE

When you’re looking at these risks, like we had talked about, obviously privacy concerns are a big one, how do you manage those risks? Do you determine the risks on a case-by-case basis, depending on if you’re doing an email or if you’re asking for maybe some assistance related to data that you’re inputting?

HIGOR PRADO

Yes, we need to check each case. When we are starting, it’s hard. In Brazil, we have a lot of laws about privacy, and it’s a very important topic. So we need to double-check and make sure that you are not putting sensitive information on ChatGPT or Bard or other tools. Project managers need to take special care. And, of course, read the terms of ChatGPT or Bard. A lot of people use these kinds of tools and do not read the terms of use. It’s important to read it, especially when we are using it as professionals.

HANNAH LABELLE

Now, what are your top tips for other project professionals when it comes to using generative AI tools in their project management practices?

HIGOR PRADO

I tell them to start small and experiment. Once you start seeing the benefits, get your team on board. Always keep a human in the loop to monitor the AI and step in when necessary. Don’t be afraid to test the waters and learn as you go along.

Some people think ChatGPT is really smart, but it’s not as smart as it appears. It’s more generative as the name says—generative text, in the case. So it’s good to start with these kinds of tasks, and after that, we can start doing more complex things like generating reports, or sometimes you can use it to help us to estimate the effort or other things about the project. But we always need to double-check the information because ChatGPT is not smart, as I said. But for starting, I would say to start not doing professional things, doing simple things on daily basis. For example, “ChatGPT, tell me a joke.” And see the output.

HANNAH LABELLE

We’ve talked a little bit about the present. Now let’s look to the future. How do you see generative AI tools really impacting project management in the short as well as the long term?

HIGOR PRADO

This is a question for a million dollars, but I can try to guess. In the short term, these tools are fantastic for streamlining tasks and boosting productivity. Looking further down the road, as AI technologies become even more sophisticated, they may play a more central role in strategic decision-making. I could even see them helping manage stakeholder relations. But no matter how advanced these tools become, they are just that—tools. They are here to assist us, not replace us. The human judgment and experience, it will always be at the heart of project management.

HANNAH LABELLE

Higor, thank you so much for talking with me today. This was a great conversation, and I really appreciate it.

HIGOR PRADO

Thank you, Hannah.

NARRATOR

Thanks for listening to Projectified^®. If you like what you heard, please subscribe to the show. And leave a rating or review—we’d love your feedback. To hear more episodes of Projectified^®, visit Apple Podcasts, Google Play Music, Stitcher, Spotify or SoundCloud. Or head to PMI.org/podcast.